Ivy Bridge Temperatures – It’s Gettin’ Hot in Here

Why is Ivy Bridge so hot? Ask that question in any forum currently, and you are likely to receive one of two different popular (but not entirely correct) answers that everyone has been parroting:

  1. “Power density is greater on Ivy Bridge than Sandy Bridge”
  2. “Intel has problems with tri-gate/22nm”

The first answer is correct, but wrong at the same time – power density is greater, but it isn’t what is causing temperatures to be as much as 20 °C higher on Ivy Bridge compared to Sandy Bridge when overclocked. The second answer is jumping to conclusions without sufficient evidence. If you aren’t in the loop, there’s evidence of a considerable temperature difference nearly everywhere you look – we confirmed it by mirroring settings in our Ivy Bridge review, and we have read similar reports in solid testing at Anandtech as well as from other sites.

So why is Ivy Bridge hot?

Intel is using TIM paste between the Integrated Heat Spreader (IHS) and the CPU die on Ivy Bridge chips, instead of fluxless solder.

Ivy Bridge Delidded, showing traditional TIM

Ivy Bridge Delidded, showing traditional TIM (Image courtesy of pt1t.eu)

 

How does TIM paste generally compare with fluxless solder for conducting heat? Heat conductivity can be measured in watts per meter Kelvin. To be technically exact, we would need to know exactly what Intel is using for TIM paste/solder. When I went to Intel and asked, their polite answer may not surprise you – “Secret sauce”!  Given that, we can use some rough approximations. A solder attach could have a heat conductivity in the range of 80 W/mK. A TIM paste could have a heat conductivity in the range of 5 W/mK. That’s your problem right there! Note that these values are not exact, as we don’t know the exact heat conductivity of Intel’s “Secret sauce”. However, these are values representative of solder or TIM paste, and there is a giant gap between how TIM paste and solder perform in regards to conducting heat. They are in different leagues.

Demonstrating the Problem

Most importantly here, if Intel is using TIM paste between the IHS and CPU die, the IHS effectively becomes a heat barrier rather than a heat spreader. Here is a rough diagram of the current heat transfer on Ivy Bridge:

  • CPU Die -> 5 W/mK  TIM -> IHS -> 5 W/mK  TIM -> Heatsink

It would be far more beneficial for temperatures to take a more direct route such as:

  • CPU Die -> 5 W/mK TIM -> Heatsink

Extra heat interfaces are a bad thing, especially when they have relatively low thermal conductivity. On a fundamental level, it doesn’t make much sense to do things this way from the perspective of optimal cooling. However, it could make sense from a die-protection standpoint.

In contrast, a fluxless solder attach like that described in Intel patents was invented for the specific purpose of quickly and effectively radiating heat away from the CPU die. In this situation with a solder that can conduct heat in the range of 80 W/mK and in light of tighter and tighter power densities as Intel continues to shrink its processor die, you can start to  see on a fundamental level how quickly getting the heat from a very small area to a slightly larger area may be helped by the design of a soldered IHS. This still leaves the problem of a 5 W/mK TIM paste interface between IHS and heatsink, but before you get there you have a high conductivity solder attach between die and IHS that radiates the die heat to a larger area.

Ivy Bridge Power Density

Power density likely became a popular answer because Intel has referenced the challenges it presents with process shrinks, and it just makes sense on basic level. Very hot die, smaller area to conduct heat away from. Blaming power density for the heat issues is easy! However, Ivy Bridge has approximately 75% the die size of Sandy Bridge, which is a big difference certainly, but not enough to explain the stark contrast in temperatures obtained by our peers across review sites and the forums. Where Sandy Bridge would often be around the 60 °C range at a 4.5 GHz overclock, Ivy Bridge has been tested to be in the 80-90 °C range.  How can we blame power density for a difference that large? That dog just doesn’t hunt!

In light of this contrast, we can gain further insight as well from what history has taught us. If you’ve been paying attention, we saw similar issues between the E6XXX and E4XXX processor lines. The E6XXX used a solder attach under the IHS and were far easier to keep cool. The E4XXX used a TIM paste under the IHS and ran hot! Those aren’t the only examples, and I’m certain enthusiasts in the community with better memories than myself can lend further supporting evidence from our past experience. Given hindsight, it is hard to explain why Intel would make a return to TIM paste for Ivy Bridge.

Bottom Line

So based on what evidence we could find from our own investigation, as well as what experience has taught us, Ivy Bridge is running hot when overclocked because of TIM paste between the IHS compared to solder attach used on Sandy Bridge. Why Intel made this choice we aren’t yet sure. We also aren’t sure if they will continue using TIM paste on the Ivy Bridge line, or if this will only be seen on the Engineering Samples like the units sent out for review.  However, we’ve put word out again to Intel and are waiting to hear back if they have any further insight or comment to offer. If nothing else, we can hope their reply will again be in good humor… “Secret Sauce” did give us a laugh!

- I.M.O.G.

Tags: , , , ,

399 Comments:

gutlessVADER's Avatar
I'm glad somebody took the time to explain all that. I'm planning on buying a 3770K around December, so I'll definitely be keeping up with Intel's statements. They've already publicly stated that Ivy Bridge runs hotter, though, so I can't imagine them switching to a better heat conductor after already publicly denouncing their CPU's ability to stay cool. Maybe TIM is cheaper, and they've decided that with AMD out of the picture they can do whatever the hell they want.
I.M.O.G.'s Avatar
I didn't want to wander too far into the weeds within the article itself, but I want to mention as well about delidding. Many people know it can be dangerous and easy to kill a chip that way. However people often forget a challenge with making contact with the die - if you delid, you may need to modify the socket/socket retention clip to ensure the base of your cooler can contact the die, as without the IHS the chip will ride lower on the motherboard. Just a note for anyone who may follow up with further testing on this.
hokiealumnus's Avatar
It's also important to note that those who air cool their CPU with direct-contact heatpipe coolers do NOT want to delid their CPU. You'll have heatpipes that don't touch the heat-producing die surface and be worse off than when you started.
EarthDog's Avatar
Also, you want to make sure you do not crush the die either.

Great and informative article IMOG!!
I.M.O.G.'s Avatar
Sort of similar deal with most recent waterblocks as well Hokie, they are designed to cool a heat spreader, not a focused central die - older blocks like the Storm were designed to cool bare die.

As for crushing the die, I probably won't be delidding myself - my chip will only see LN2 usage, and with remounting the F1EE on a weekly basis, I'd be bound to crack or crush the die before long. In that case, I'll take the IHS for protection alone - the LN2 will do its job well enough. A lot easier to hurt a chip without the IHS on.
wagex's Avatar
could you take the lid off, clean up tim, put some little chunks of solder were the tim was set lid back on and heat it with a heatgun or mini torch until it settles back down could work?
dejo's Avatar
very nice find, and also very informative.

one question I would have is if sandy bridge has been confirmed to have fluxless solder and not a similar paste. I would guess it is the solder method, as it would explain the huge discrepancy on temps between 2 fairly similar cpu's. I just would love to see a delidded sandy bridge cpu as well.
MattNo5ss's Avatar
What are the differences in waterblocks designed for direct die cooling vs IHS cooling? I would expect completely flat base for direct die, but what else?

The whole direct die cooling era was before my time
EarthDog's Avatar
Id imagine the pins and channels in the block were more centralized than spread out, but that is just a guess... and likely a poor one.
I.M.O.G.'s Avatar
There was something of a pinnacle in water block design, just before the time when the IHS first became popular. Once the IHS came into the picture, blocks designed for use without an IHS were less effective with an IHS, and new blocks adjusted accordingly.

With waterblocks designed to cool a die without an IHS, microchannels and impingement (don't hear that too often anymore!) were major design features. Impingement over the central area of the die was important in that it greatly increased the interaction between the water and surface area directly above the die. There was more focus on cooling that small area primarily, than the current focus on cooling the larger IHS area.

There are other old timers who are much more well versed than myself within our community who could speak in more detail and with more knowledge. Nikhsub knows, cathar knows, billa knows... Others too, but those are the guys off the top of my head. rge is still very active around here and he could probably explain in more detail than me as well.
wingman99's Avatar
Didnt SB use the same TIM on the IHS and it has no problem.

This analogy has no facts to prove this theory either a test of removing IHS and have direct contact needs to be done to confirm.

If this theory was true then water cooling would not help. The heat would build up and not make high enough dissipation for water cooling to be affective.

A heat barrier is like not locking down your heat sink then the temp is high and nothing will cool it.
Cowamundo's Avatar
I would be very surprised if they would switch this "secret sauce" for solder on later chips... but who knows, would be something interesting to wait for then!
I.M.O.G.'s Avatar
SB did not use the same TIM. SB used solder attach. Here's a reference I pulled from a quick google:
http://forums.vr-zone.com/the-overcl...ml#post9743547
bmwbaxter's Avatar
Nice editorial Matt! Glad to see the discussion from the Ivy review got made into an informative frontpage article.

Hopefully the TIM is only on the review samples.
Hicksimus's Avatar
What's next? Lamborghini sending Road and Track Lambos with tires from Wal-Mart? Oh, wait, they have competition.
I.M.O.G.'s Avatar
I don't know if that comment is accurate in any way, but brownie points because it made me laugh.
dejo's Avatar
very nice once again Matt. I suspected that SB was soldered, proof is in the pudding.
rather dissapointing that Intel would back off a few nickels savings for the heat terror that those jumping to ivy will be faced with. rather sad day for me. think I will wait it out and see if they change thier minds on a new stepping.
Hoping this thread exposes their choice enough to make them rethink the process
I.M.O.G.'s Avatar
Thanks Jonnie. I'm not totally convinced that TIM paste will be a permanent feature really... One might actually suspect that retail units will be soldered, and what we've found and explained in this article is more meaningful for the preliminary results we're seeing in most of the reviews and forums about Ivy getting hot when overclocked.

There are some exceptions, despite most people raising questions about the heat when overclocked, like the computershopper overclocking results:
http://computershopper.com/component.../4#review-body

They didn't perceive the temperature issues reported in most articles with their sample. Their test was rather quick at 10 minutes then a temp reading, but I suppose it could be an indication that not all these chips are hot. Anxious also to see if Intel can add better insight than what we have to work off of currently.
nikhsub1's Avatar
I'm glad I'm still not testing water blocks... I think a Storm would be ideal on a bare die processor for H2O cooling... you do NOT want a bowed block on a bare die - at all. We are back to flat as can be again. No one is making blocks for bare dies at all. Intel really needs to be consistent, either solder the lids like you started doing a few years back or just get rid of the IHS altogether (doubt that will ever happen).
diaz's Avatar
This is very dissapointing. I'm still upgrading to 3000k, but I will hate to see inneficient heat transfer. Here's to hoping that intel used a better process for the launch day retail versions. I am NOT delidding. I shouldn't have to do anything like that after spending $300++ on a CPU.
diaz's Avatar
Maybe some lucky chips have a better contact to ISH than others?

EDIT: Maybe it would help wo have a higher pressure mount on the cooler?
dejo's Avatar
it would definitely be a relief if intel did indeed use the solder method for retail release sku's.
one other thing that many didnt pay attention to on x58, was that they had DO steppings at initial release. Even though (to my knowledge) they werent released to public for many months. may be something similar here
wagex's Avatar
and im sticking with my 2500k unless they fix it cuz its more than enough power for me if anything im gonna get a better motherboard that doesnt have coldboot issues, 4 ghz is the highest i can go witout it randomly losing oc settings... with any bios. though i could run 4.9 ghz when i had my custom loop
TsunamiJuan's Avatar
I hope everyone remembers one of the reasons why they started putting heat spreaders on the chips after the great socket 370 Debacle. OEM's and Custom builders where cracking cores left and right during heatsink installation. It didn't help that the pressure of the heatsinks at the time to be within intels spec was something stupid like 10lbs+ of pressure. You literally had to use a tool to help get the stock heatsinks to clip to the socket. This also resulted in alot of cracked sockets over time since the mounting tabs for the heatsinks where part of the socket.

Obviously it was far cheaper to protect the core with a heat spreader than it was to continue to RMA chips that got damage due to a poor method of mounting cooling.
diaz's Avatar
That's fine, but why was SB just fine and now the new cutting edge product has crappy temps in comparison, knowing that a simple solder would solve it?
bmwbaxter's Avatar
probably to clear out SB stocks, then once they are getting low switch to solder on IB and get a bunch of upgrades without the issue of old CPU's taking up space or time to sell.
PcChip's Avatar
In my opinion the reason most of these overclocked Ivy Bridge chips are running hot is because reviewers and benchmarkers have been using the voltage they used to use on the old 32nm Sandy Bridge chips on these new 22nm Ivy Bridge chips, and it's much too high (or at least a much higher increase over stock). For example, since people were used to setting 1.4v on their 2500k/2600k chips, they're setting it on the 3770k not realizing that it's a much higher percentage increase over stock than it was on the sandy bridge chips - and temperatures skyrocket. Check out the undervolting article on AnandTech to see a graph of temperature versus voltage.
dejo's Avatar
if thats the case, then why did the westmere shrink from nehalem produce cooler temps? westmere for the most part was considerably cooler at similar clocks and voltages to nehalem. I have a gut feeling that the tim rather than solder is the criminal at work here
I.M.O.G.'s Avatar
@PcChip: I linked that Anand article in this article actually.

I know in our review, Hokie tested for the voltage necessary to run the clocks - I don't think he was overvolting in relation to the frequency he was running.
TsunamiJuan's Avatar
One of the early NDA testers was talking about the fact that they have had to drive new 3d 22nm transistors at a higher voltage so they could get them stable and to market. Which is why there wasn't a really a difference in core voltage, that would normally be associated with a process shrink like this.

btw i went digging for voltage for the new chips. Their is no set listing for the chip models yet that i can find but the data sheet says this.

VID VID Range min0.2500V — max1.5200V Note:1

Note 1 - Each processor is programmed with a maximum valid voltage identification value (VID), which is set at
manufacturing and cannot be altered. Individual maximum VID values are calibrated during manufacturing
such that two processors at the same frequency may have different settings within the VID range. This
differs from the VID employed by the processor during a power management event (Adaptive Thermal
Monitor, Enhanced Intel SpeedStep Technology, or Low Power States).

thats straight from the Intel Ivy bridge Datasheet volume 1, section 7, table 7.4
SF101's Avatar
i hope they change this back to solder..

its like a step back to my amd days of creditcard spreading.
diaz's Avatar
That's definitely a part of it, but even with a very cautious voltage setting temps are still quite high compared to SB.
wingman99's Avatar
That is what ive been thinking, the TRI-gate has increased inversion layer requiring more drive power when overclocking

Maybe it's combination with TIM and 3d transistors.

When people start delidding and running strait on the core, also comparing the watts overcloking SB vs IB we will see.
I.M.O.G.'s Avatar
The VID on our review sample was 1.0-something according to a comment I saw from Hokie, if I'm not mistaken. This is just one example of course, and I don't have any idea how representative it is of Ivy Bridge in general.
TsunamiJuan's Avatar
Yeah i think its like 1.2v-1.3v is probably close to what the normal runtime voltages usually get set at out of the factorys. Just kinda depends on the bin level of the chips it seems. Usually anything below 1v is considered either ultra low volt or is a chip running in a altered power state. So while speed step is doing its thing or your gating of other cores due to lack of load.

I would be interested to see more concrete numbers from intel, the ivy bridge section of their site is kinda a mess right now as the only official listed chip for sale is the 3770t(which is a 55w 2.5ghz 3.9ghz boost, chip), with a release date of the 29th which goes for the same for the other chips. However they are still Listed under formerly Ivy bridge and not yet under the heading of core 3rd generation core i7 or likewise.

I might mull through the data sheets more tomorrow and and see how the 2nd and 3rd gen voltages compare. If there are any major differences for the main voltages.
diaz's Avatar
For the 3D transistor conversation, it doesn't seem to have much of an impact on the power consumption, which would be exactly where it would hit... it wouldn't simply translate to more heat, it would cause more power consumption which would cause heat. I'm sure its the lid.

http://www.bit-tech.net/hardware/201...3770k-review/8
Here the review shows:

2500k @ 5Ghz = 311W
3570k @ 5Ghz = 267W

2600k @ 5Ghz = 313W
3770k @ 4.8Ghz= 244W
hokiealumnus's Avatar
Our VID was 1.080V. I ran our overclock at the lowest voltage it would remain stable. It would be pretty silly to just run the same Vcore as the 2600K.
Scout's Avatar
I really wanted to upgrade to Ivy now I am not so sure hopefully a new spin or stepping fixes the kinks of the first run. I think 22nm has alot more potential.
Eldonko's Avatar
Im sticking with the power density theory myself.
I.M.O.G.'s Avatar
Why? Color me curious.
rge's Avatar
power density plays a role, trigate, and solder vs tim would play a role, only issue is how much for each.

But regarding wattage earlier post between sandy and ivy, you cant go by total watts dissipated to figure difference in core temps when comparing across a die shrink, even though total heat output the same, what matters is power density, ie power per cm2, and only the area/size where power is actually consumed or heat "originates" matters.

150 watts consumed over a large area, the temperature would not be very hot.

150 watts consumed in a .001 micron area would exceed temp of a nuclear reactor.

See intels slide that everyone uses when explaining power density increase with die shrinks. Same wattage originating from smaller area = higher core temps.
Eldonko's Avatar
The heat only makes sense, increase the transistor count from 1.16 billion on SB to 1.4 billion on IB and shrink the die from 216mm2 to 160mm2. That and I have confidence that Intel knows what they are doing with TIM..
vargis14's Avatar
I believe intel used the paste as a kind of thermal resistor so ivy bridge cpus will not hit 6+ghz on a good air cooler or a decent AIO water cooler.Combine the 25% smaller die and the paste and they take away 6+ghz
So many Sandy bridges can hit 5+ ghz on affordable cooling.
If almost every 22nm IBk model can hit 6ghz + like the sandys hit 5+ghz it would hurt Intels sales of 6 core SB-E platforms and cpus when you could get the same or even better performance with a 6+ghz IVY
I think the new coolermaster 812s heatpipe/vaporchamber heatsink will let that naked IVY hit 6 ghz maybe more since the bus can almost pass 110mhz add ivys improved memory controller and i think 6000mhz on air

All you need to protect the die is maybe the old amd barton style hard rubber bumpers on all 4 corners or even a ATI/AMD GPU trim ring.Those chips are not cracking.One of those ideas,or a better one is all that is needed to keep the coolers/block from tilting and cracking the chip.
You can use a razorblade and a light to make sure the die is flat and does not have any high spots on the corners and edges.If you do find a high corner i am sure you could carefully shave it with a fine razor.I have seen cracked chips still work so i don't think shaving a minute high spot would hurt a thing.
Get that naked IVY chip in a motherboard and post the oc/temp/voltage results please.BTW i salute the person who found that poison under the IHS 6ghz on air baby i can feel it!!!!!!
Eldonko's Avatar
Lol if someone does 6Ghz on air because they removed the IHS Ill eat my friggin hat.
wingman99's Avatar
Looking at the chart the power density went up on the core 2 Duo however the power of CM2 went down is this because of critical dimension and what is that?
I.M.O.G.'s Avatar
Thanks.

I also have confidence that Intel knows what they are doing with TIM. I also have confidence that every solution they choose is a balance between what is technically ideal and what is realistically appropriate. They could give us 6GHz on air processors by the end of the year - they have the resources skills, and ability to execute. I don't doubt at all that it could be possible if they dumped enough money into achieving that goal - but maybe they won't do that because making it a reality is too expensive, difficult, or otherwise bad for business.

If using an inferior TIM is sufficient and cheaper/quicker/easier/etc... Despite knowing a different solution would keep things cooler, I could see how they could make a decision to use an inferior cooling solution that was good enough due to balancing priorities.

The 6GHz thing is only intended to draw out a point. What they could do, what is best, and what they actually do can very realistically be different things. And if they are different, it isn't because they are dumb.
vargis14's Avatar
What kind of sauce would you like,honey mustard,sweet and sour or just plain old ketchup replace popcorn with hat please.

I do have high hopes, unfortunately i am not wealthy.But if i was you better believe i would be on it like a fly on doo doo.

I think that paste slows heat dissipation a great deal.CPU die to paste of unknown origin and properties to IHS to paste to heatsink.3 layers of stuff before it touches the heatsink.
I just really have a gut feeling intel made a cpu a little faster then they planned on and that secret sauce will remove SB stockpiles quickly,also aiding in SB-E platform sales.
SB-e has a place for super memory intesive stuff workstation quad sli etc.
But with IVYs potential raw cpu power PCIe3.0 I mean who will need 4 video cards when the dual gpu 690s and AMDs 7990 come out.BTW i prefer the new NV cards
I.M.O.G. i know intel is not dumb they are a powerhouse with no competition.So they use tactics to sell older cpus.If that means crippling a super fast cpu just enough to make it slightly faster then its older brother at identical clock speeds but wont break 5ghz without extreme cooling is good for business.They clear inventory etc.
Good luck to the first naked IVY cpu!!!!
wingman99's Avatar
LINK: http://www.anandtech.com/show/4830/i...ture-exposed/1
QUOTE: Intel hasn't announced die size but transistor count has increased to approximately 1.4 billion (layout). This is up from 1.16 billion in Sandy Bridge, a 20.7% increase. With perfect scaling a 22nm Sandy Bridge die would be 47.3% the size of a 32nm die. Even with the increase in transistor count, it's a good bet that Ivy Bridge will be noticeably smaller than Sandy Bridge.

A good percent of the transistors go to the GPU and the CPU portion of the chip did not go through much change from SB it is a TIC, however they added greatly to the GPU transistor count in IB. intel wont give out the split in the transistors CPU and GPU.
Eldonko's Avatar
Not saying you are wrong, that could very well be the case m8 (or part of it at least), Im just saying that in my opinion the chip architecture would have a lot more of an effect on temperatures than the TIM used. Moore's Law, die size, transistor count, tri-gate, and so on.

I guess this will be proven or disproven as soon as someone removes their IHS and overclocks the chip. I have my $ on a few C less and a few more Mhz same as any other IHS-less chip. Ocers are a curious bunch so we will see this sooner rather than later. It is not gunna be me though lol
dejo's Avatar
I still see no reason for Intel to choose TIM over solder, other than dollars. I also think that most that are willing to pay for the top offering of a specific family- would also be willing to spend a few dollars more to be the best that product "could" be. Bring on 2600k price, but also bring back the solder, and let us see what "should" have been
Krazy931's Avatar
just confused on the whole people saying intels trying to sell the rest of their sandy bridge when that is not the case at all theres a very precise way things are done when new things are coming out and normally the manufacturer holds very little of their own product it would be people like newegg tigerdirect etc.... that would have the stockpiles of extra chips not intel so im sorry you guys can throw that thought into the trash
txus.palacios's Avatar
Can't understand this movement from Intel. OK. You have no competition. But normal OCers (not LN2 benchers) will not buy your "k" chips.

I will stick with the 2500k or grab a i7 SNB if I find it cheap, if the temperatures get as high as that just because you decided to save a few bucks using TIM instead of solder.
rge's Avatar
If it were cheaper to use solder, the cheaper E7200 etc would also have solder like e8400 instead of non-solder/paste die attach, but they dont. So definitely agree with non-solder/paste being cheaper, and intel saving money when die attach on a particular cpu doesnt require solder when used as intended by intel.

Also possible, if intel thought trigate was going to lower voltage, then only later in design realized wasnt the case, perhaps that played a role in die attach choice. Or the fact that at stock setting, non-solder attach works fine with Ivy.

As IMOG pointed out, intel clearly has the ability to design for overclockers, but the future isnt overclocking, it is mobile, less power hungry platforms, and intel isnt designing chips for us few that push way more volts through to overclock, but for the masses.

No one is saying intel doesnt know what they are doing, we are simply acknowledging intels needs and overclockers needs may not always be the same.
lance-M's Avatar
very good info..seems to match up with other sites around the planet on this issue..heat and chips. but this is the 1st forum to actually find a reason for it.
thanks again for the info..
evangelionstar's Avatar
Question is: is this an issue limited only to the engineering and early release samples OR does it also affect the retail versions of the 3770k that will be released in 3days?


Has anyone read or perhaps suggested it to those lucky ones around the net that have actually gotten their hands on full retail kits?

These chips have been available for a while now in China and a couple of days in Norway, Hollands, Germany etc
EarthDog's Avatar
I dont think anyone is sure as I believe, but not sure, that review samples were all ES. We have a couple days to find out though.

Not sure if a retail one has been delidded.
wagex's Avatar
welp only one way to tell, who has the classified chip? tell them to pop off the lid and attach a heatsink right on the die and see what happens lol
hokiealumnus's Avatar
Just for reference, here is what a soldered die looks like de-lidded.

109059

Image courtesy BlindFreddie @ XS.
Robert17's Avatar
This editorial got linked on The TechReport:

http://techreport.com/discussions.x/22855

Overclockers.com editorial: Ivy Bridge temperaturesit's gettin' hot in here
hokiealumnus's Avatar
....and this is what happens if you don't take a soldered IHS off properly (image found @ OCN).

109061

Here's a soldered IHS removal guide.
hokiealumnus's Avatar
Sorry for double-posting, but this is worth it - Non-soldered IHS has been verified by Shamino (of ASUS RoG team fame).
Blackbird_CaD_'s Avatar
I'm very disappointed that Intel does this to us.
I also feel anger towards Intel.
hokiealumnus's Avatar
Yep, Sandy Bridge was soldered - Reference.
Salmon91's Avatar
http://muropaketti.com/artikkelit/pr...k-ivy-bridge,7

They tested a 3770K (ES version) and a 3550K (retail version) and there was a clear temperature difference between the two.

3770K http://plaza.fi/s/f/editor/attachments/3770k_oc.png

3570K http://plaza.fi/s/f/editor/attachments/3570k_oc.png

They're not the same CPU, so I don't know how comparable these results are, but the retail 3570K ran much cooler than the ES 3770K. They were both air-cooled with a Noctua NH-D14 and both were running at about the same voltage and clocks.
EarthDog's Avatar
I sure hope so, one has HT and one does not. BIG difference there.
Aynjell's Avatar
I hope that intel either sticks with it, or changes to solder before release. Either we need to be able to COUNT on being able to delid chips, or not have to.

Also, I've got tons of experience with delidding and using direct die stuff. I can manage. ^_^

Common sense says they have to fix it though. Think about it, people will just buy 2600k's instead of 3770k's because in the end they're faster. O.o
dejo's Avatar
there is also a 100mhz cpu difference speed wise as clocked. 3550 was at 4600 and the 3770 was at 4700
Frakk's Avatar
If anyone is thinking about deliding there Ivy and replacing the Thermal Compound with something expensive off the shelve don't take the chance..... its been done here
Eldonko's Avatar
Well ya, expecting hundreds or even a ghz more from this is plain silly.
Frakk's Avatar
I don't think he was expecting that, he was trying to get temps that are more sandy than ivy at high clocks
evangelionstar's Avatar
http://forums.anandtech.com/showthread.php?p=33347200

shows the temps of a retail 3770k
"4.5 ghz 1.200 V at load (2nd highest LLC on gigabyte UD5H), ~20C idle, 72/75/76/71
testing now at 4.6 on same volts. 1 hour in and hottest core at 76
@4.8 83C"


Seems like it does use TIM and therefore shammy@asus's suggestion of changing the TIM to something thicker and filing the IHS feet down would help significantly.
dtrunk's Avatar
So who has their retail IvyB? Earth, did yours come from china yet? anyone remove the IHS on a retail chip?
EarthDog's Avatar
I was joking in ordering that from ebay. And yes, it was linked in here already that a retail chip has been delided and shows the same TIM. Post # 63.
txus.palacios's Avatar
THAT looks much much better...
diaz's Avatar
hokie went in that thread and pointed out that the cooling wasn't meant for that configuration... pressure change on the chip is much different.

Also the water bock would need to concentrate water differently for a smaller chip.
EarthDog's Avatar
This would have been great in the already created thread about IB and deliding... even though the premise of this thread was bunked just a couple posts below in the provided link.

Delete now?
dtrunk's Avatar
great post over at kingpin Hokie; hopefully the OC.c team can at least collectively give intel some crap about this and I'm personally hoping for revisions within the next batch of IB chips. lucky for me, rent is due and I won't have $$ till mid May; so i can see how this plays out. @earth, darn...
Robert17's Avatar
Reading some reviews and all the posts here has had my head spinning. Out of the spin, I'm wondering if there is something inherent in the new lithography, i.e., tri-gate 3D, that makes the use of solder in attaching the IHS to the CPU a difficult proposition. I've linked a Wiki entry for soldering for generic review and consideration. Embrittlement using RoHS compliant solders may be an issue, solder temps not so much due to the availability of several RoHS compliant solder choices. And there is mention of several more exotic 4-metal solders available, all leading me to think that if cost isn't an issue, solder availability isn't an issue, and the major change in the whole bonding equation is the lithography, the engineers found some hard-to-resolve bonding issues.

http://en.wikipedia.org/wiki/Solder
dtrunk's Avatar
i would argue that just because some1 (in their basement) de-lid a IB i7 and then re-lid it with a different TIM is NO WHERE as efficient/effective at this process as INTEL is... this needs to be done, FROM INTEL and BY INTEL, from the factory for us to truly compare the TIM issue.
EarthDog's Avatar
That plus what hokie said inside the thread in the first place make his testing pointless.
Frakk's Avatar
Its worth a try, i know AS5 is far better than the crap you get pre applied on heat sinks, you can bring your GPU temps down 2 - 3c just by replacing the factory thermal compound with something decent.
trekky's Avatar
thx for the review!
hmm the heat thing is bad for intel but whats bad for intel is great for AMD
maybe this is what amd needs to show Piledriver out and maybe it will overclock alot better then ivy because the way it stands it looks like Ivy will have a very hard time getting over 4.5ghz if they are alreading reaching 80c-90c!

only time will tell

oh and secret sauce!
what are they a Hot sauce company now?
Franks Red hot better becareful!!!!!!! (for those who dont know Franks Red hot is a hot sauce maker and its pretty good)
dtrunk's Avatar
2-3c is not worth razorblade/torch action and the possibilities of my st0n3r ar$ messing it up... what do i have, a few milliseconds of too much torching and my "new" $320 processor is junk? scary!
Aynjell's Avatar
I don't know that I can buy this chip now. I was really hoping to get a 3770k to go with my 680. I am having trouble with streaming and could use the extra horsepower... but if they're gimping the chip with hot temps I can't buy one. :\
hokiealumnus's Avatar
They might NOT be. There may be many reasons for the change, but we can't know them until Intel tells us what they are. Maybe the new transistor can't tolerate the solder. The only people that know are at Intel.
Frakk's Avatar
meh... go Sandy then. or Wait to see Piledriver, thats what i'm doing.
If Trinity is anything to go by it should be pretty good, certainly a whole lot better than Bulldozer.
Aynjell's Avatar
That may be, but intel is shooting themselves in the foot here. The 3770k is not an upgrade over the 2500k or 2600k if you aren't running extreme water, and even then... 90C is like WTFHOT.

if the release chips aren't switched to solder I can't afford to use one, and will have to buy a used 2600k or something so I can get HT.
dejo's Avatar
the nice thing is there is some debate going on, and hopefully that will prompt Intel to come forth with more information.
hokiealumnus's Avatar
What I said in response to the thread linked above:

Merging this with the existing discussion. Please try to keep it all in one place folks.
bmwbaxter's Avatar
Guess someone at Intel puts that sh*t on everything.
hokiealumnus's Avatar
I'm going to post this here too. Some people at OCN have taken this piece and run like hell with it.

hokiealumnus's Avatar
The Tech Report was able to get a comment from Intel. It's not exactly conclusive, but does shed a little light.

trekky's Avatar
im just hoping that he did not put franks in the Tim.... but that would make a tasty Tim......... hmmm franks TIM!



Fast Edit oh and wow we trade 20c at 4.5ghz but intel says OH WAIT guys it can go 5c higher the sandy
rge's Avatar
I posted this over at xtreme as well, but while responding to Beepbeep listing manufacture tim specs (which are often exaggerated like fan specs), I found something interesting, last quote in post...

And to add to what DilTech said, you cant go by manufactures data bulk conductivity for tim, some are as accurate as fan specs. There is bulk thermal conductivity, interface resistance, and most importantly non-standard testing. For example AS5 I think lists their tim as 8 w/mK, yet has been tested on actual cpus to be less than 1 w/mk by both NREL and other reputable testers. Shin itsu and Dow was consistently measured as highest at ~ 4 w/mK on more than one study, and they accurately list their as such. Found NREL, page 9,http://www.nrel.gov/docs/fy08osti/42972.pdf . Any manufacturer that lists TIM paste as performing higher than 6 w/mK, better come with independent proof, not saying its not out there, but certainly havent seen any proof yet. phase change/indigo materials/thermoplastics aside.

I will try to find that power point slide from intel I posted somewhere in this forum couple years ago, but their solder attach was updated in past few years, and melting point was 150C, thermal cond ~80 w/mk.

But as listed above post from Tech report, to me, pretty much sums it up... However, Intel claims the combination of the new interface material and Ivy's higher thermal density is responsible for the higher temperatures users are observing with overclocked CPUs.

And for those that dont think TIM matters with large resistance changes, quote from http://www1.eere.energy.gov/vehicles...narumanchi.pdf slide 15
I.M.O.G.'s Avatar
@rge: What thermal paste tests by community members do you actually put faith in? I have never seen a roundup which I thought was worth much faith. There have been various big roundups (tom's and benchmarkreviews roundups come to mind), each of which I thought had problems in their approach, as well as a lack of transparency. Transparency is critical in these kinds of tests to lend accuracy and verifiability to the testing - it is hard to get independent testing verified critically, and it isn't in the reviewers interest to put in all that time (testing TIM is very time intensive) to then go back at the end and demonstrate why their testing was innaccurate. Skinnee made a nice attempt, perhaps about as good as you can do on an "amateur" setup, however he also didn't escape some of the pitfalls inherent to testing thermal pastes.

I am not saying you are wrong to believe in community tests, but I've always been of the belief that the differences are too small and the variables too great to get any meaningful accuracy. Often times when comparing 2 sets of tests, it is demonstrated that its really hard thing to do well.

I don't mean to nitpick, but your link went to a study by the NREL, which seems like a different agency than NASA which you referenced. Was that a typo, or do the two agencies have a close relationship. I'm not that familiar with either, other than a normal casual familiarity with NASA that any American should have.
rge's Avatar
yep wrong agency had both nasa and nrel studies open, I had too many pages open, both of those are nrel. I think skinnee did a good job. But my only comment was cant go by w/mk of published specs, since no standard testing. Professional testing of consumer tim by government agencies and professional companies I have seen maxes out 4-6w/mk range. Dow does a lot of testing as well and publishes results, granted they have a horse in race, but thermal conductance is similar to nrel and others. I havent ever seen any community tester list w/mk of any tim as wouldnt have means to test, and wasnt referring to any community tester for those values.
I.M.O.G.'s Avatar
There are also some pretty strong arguments against going by w/mk alone... Heat conductivity is only one of many factors that affect the quality of the interface. I just talked to Arctic Silver earlier to get some better background experience, and they reminded me of some things I've read previously, as well as brought to light additional items I hadn't previously considered.
rge's Avatar
Nrel etc test the actual resistance through the die interface when applied to chip. And they list bulk conductance, contact resistance.

finally found intels 87 w/mk indium tim quote...well one spot, I had originally seen in a slide, but here it is in intel technical journal page 9.
http://download.intel.com/technology...ally_Green.pdf


Aynjell's Avatar
I don't think beheading is even an option with the retention mechanism, is it?
dtrunk's Avatar
microscopic stress cracks in the Indium soldier based TIM... mayb this is the issue. Though this article seems a little old to be relevant to a 2012 launch. I could see how striving for "uniform" heat dissipation and heat cycle stress cracks of the STIM could push intel to use simple TIM this go around. I still hope they revise this asap for future batches.

Nice post on the intel link, good reading just for fun.
odin_611's Avatar
Perhaps it's to try and subdue overclocking long enough for intel to milk Ivy Bridge for better performance values in later models. Same processor, release it 8 times, minor tweaks to the first 4 then for 5-8 solder instead of TiM and add a major off the shelf performance boost to the late models. Ahhh good old intel :P
rge's Avatar
@wingman99, yes page 10 if going by pdf page counter, though page 9 is printed on that page, since page 9 of the article. Debated whether to say page 9 or 10, wasnt sure which would make more sense.

@dtrunk, they test each cpu, and they worked out voids a long time ago. Just posted this to show the thermal conductivity of intels indium solder die attach, 87 w/mk and melting temp 157C.
mjw21a's Avatar
Looks to be an artificial barrier then. Does it surprise anybody that Intel would do this when they're ahead of AMD performance wise? If AMD catch up we'll probably see them go back to a soldered IHS and boom, their back in front again. Due to a new and improved IB
ghost_recon88's Avatar
Interesting, removing the IHS apparently did not help with temps at all.

http://www.overclock.net/t/1249419/p...ed-without-ihs
mjw21a's Avatar
^^^ Hmmm, very interesting.... Will have to see how the final retail products are on heat. If these chips remain hot then I suspect my next chip will definitely be Piledriver....
dejo's Avatar
Kyle, the test that you linked to is likely to have issues with mounting pressure when giving up the thickness of the IHS and a mount intended for used with a given thickness chip (chip with IHS-in this instance). would guess it had very low mounting force which could change temps for the worse very easily
I would trust a scenario much more with just having a mounting system that went through the mobo to mount the hsf or waterblock. and at least having a chance to get good pressure on the chip
wingman99's Avatar
That's what i believed also, we will have to see more testing to confirm it.

So it's back to TRI-gate and power density
xsuperbgx's Avatar
I don't think that the cooler used is meant for use on a cpu without heatspreader. I think it needs a little more mass at the base to distribute the heat to all of the heatpipes.
diaz's Avatar
That too, but I think its fine because it doesn't have bare heatpipes. The cooler is not the problem, at this point it is mounting pressure. The fact that the temps went UP when ISH was removed, means there is at least some pressure problem. This can be easily verified by applying pressure on the cooler while maintaining counterpressure on the backplate.. using a hand /(or screwdriver while watching the temperature difference (in small increments) as pressure if mounting allows for more tightening headroom) is gradually applied. This way a knee temperature could be found depending on pressure mount.
ghost_recon88's Avatar
Good points, who here is gonna be the guinea pig that takes their IHS since I know you guys have good mounting setups
MattNo5ss's Avatar
What about a Venomous X with its variable pressure knob? Sounds like it may be a good heatsink for testing direct die cooling, once it's lapped to be sure it's flat as possible to match the die.
mjw21a's Avatar
I think I'll wait and watch at this point
diaz's Avatar
That would be quite prefect..
Just a nickname's Avatar
I would like to see some one using no IHS with indigo xtreme. Like this we can remove all the speculation about bad contact with the die as this "TIM" performance doesn't vary with mounting pressure.
I.M.O.G.'s Avatar
Was late to getting back to this discussion, but thanks have been distributed to everyone that covered the basic points I would have.
nikhsub1's Avatar
Yeah that must be due to mount pressure... and if you look at the bare die / paste / HSF picture you can see the cooler wasn't making proper contact - I wonder if the cooler is bowed?
MattNo5ss's Avatar
I asked because I have one and it's been lapped, and wanted opinions

Plus, you can increase the pressure further by putting something inside where the knob screws down.
bmwbaxter's Avatar
sounds like someone volunteered to delid and test
Ivy's Avatar
Ivy Bridge ist from my view not truly impressive regarding gaming performance.

Gulftown (990X) vs. SB-E (3960X) was between 0 and 25% ~ 12.5% average. (Its just in order to compare, not that anyone ever need them for gaming).
SB vs. IB is between 0 and 15% ~ 7.5% average.

So the improvements this time certainly much lower.

I only mean CPU power, the integrated GPU indeed is much stronger but its of no use for high end user unless they use that CPU as a media center PC and plan not to use any dedicated GPU for. Thats indeed the strong spot of IB.

Regarding Overclocking, because Intel is retourning with a technique most recently used on the C2D 4000 series, it can be a pain to OC. I dont know why Intel decided to do so, probably simply wanted to produce cheaper. But im sure Intel will bring out some better IB end of year, so the powerusers can buy a new IB and sell the old one with a lot of money loss. The one truly gaining lot of money is Intel because many fanatics will buy the almost same CPU 2 times in a row, thats how its done it seems. Probably even better to slowly increase turbo clock capabilitys aswell but not this year, so they can buy a third CPU of same architecture.

The valid reason to get IB is to have higher heat capabilitys because of its lesser nm, and thats the stuff every user would expect. But why? Intel dont have to provide that kind of CPU to us right from the start, as long as there is only the slightest improvements, thats all they need. Yes, they say, its hotter than SB using a cool voice ? WHAT? Unbelievable... we have dynamic TDP control and shrinked nm and still hotter? Well... its magics.
diaz's Avatar
I am very anxiously awaiting a thread with lapped vs not-lapped... oh and possible delid?
TechNickel's Avatar
My old U12P is as flat as any heatsink I've ever seen, confirmed with a proper straight edge. If Noctua is anything like Thermalright, they won't have changed their base plate in the last couple of years...

The Noctua coolers are actually super easy to mod for this, just put a shim exactly the thickness of the IHS directly next to the springs on the screws and you have exactly the same mounting pressure as before with the IHS. Much better than using the Venomous X's adjustable mounting pressure since there you can't be sure that you keep the pressure constant between IHS/No IHS (not that too much pressure would make much of a difference, never did in my tests with a TRUE and shims).
mjw21a's Avatar
I'd love to see it tested with a nh-d14
Ivy's Avatar
From my experience its very hard to reach TjMax and even by putting a incredible 130W CPU inside a SFF system (the size of a shoebox) and not even the specified TDP does matter, its up to the CPU itself, as a absolutly individual piece. Of course some bad tongues from my local computer shop told me that a 130W TDP CPU will always run into TjMax (the point where the CPU would auto throttle itself) inside a SFF and stuff like that, but its not true**. To be honest, you cant expect people to know the truth who dont even try, everything based on a prejudiced mind. But the main reason a system having very limited cooling can handle such dinosaur CPUs, is because that CPU does transfer its heat to the heat sink in a very powerful manner, thats the main reason it works. That does start at the size of the IHS, going over to the TIM and how it been solved, reaching the DIE, and DIE size can matter too (even material, a DIE made with silver is top notch but doesnt exist). From the DIE it will matter how good the heat is catched up by the sink, and then it will matter how good it will be transfered to the other end... every little detail can add up, and the sum up of those details can create lot of difference. So the 6 core Nehalem was a burner but from my experience it had a good thermal dissipation, maybe SB even better... i was unable to test.

**The heat on the 990X under Intel Burn Test load for 15 min, using Real Temp 3.60 GT monitoring tool, ~25 ambient temp, never created results above 80, combined with a SFF, size of a shoebox.

Finally, even during hardest conditions (small system, huge CPU): Staying below TjMax is not issue, its to stay stable and that matter is increased by keeping the stuff cool (obviously, my system i tested is NOT made for OC). But how is it done by a CPU very hard to cool down? Stuff like that was easy on huge 6 core CPUs of Nehalem type even. I do worry to put an IB into a SFF system because even a 6 core Nehalem (TjMax of 101) would not be more of a pain when those results are true. Thing is, its harder to transfer the heat from the cores to the end of the heat sink, thus even when a IB may have lesser TDP, its still same amount of heat hurting the cores.

Besides, Indium is expensive (kinda same such as silver), so Intel will always have to pay more using it, thats certain. Its very soft and will be fluid at above 156 (so the CPU can not be destroyed when its done correctly). Although i dont think its good to fill entire gape with (still better than cheap thermal paste). Indium should be used in order to fill surface irregularities while the main material have to be elemental silver under perfect conditions. And the DIE should be united, in order to remove simply have to heat up the united silver DIE to 156 short time to remove from IHS. Certainly difficult, the indium need to be as thin as possible and totaly even and have to be attached while under vacuum (so no air bubbles). But surely the stuff i see on Intel CPUs from people who dismantled theyr CPUs, is making my eyes hurt (looks like childrens work). At least E types should have a special solution (which is absolutly affordable at a price of 1000$ and up). In general its no quality, despite its insance price of E types. Although many E types may have some better solution, guess thats why my 990X seems to perform better (on heat) than most other Nehalems. Intel could do much better but finally, it will cost them up to 100$ each unit (dependable on material) and why to pass on 100$ each unit if they could make millions charging those 100$ without delivering any real value?! No one gonna stop them, US is free country...
satandole666's Avatar
So I still have my trusty Storm Rev.2 from the AMD64 days. I'm wondering how a 3770K would do delidded and WCed for 24/7...
Ivy's Avatar
The stuff annoying me the most is the fact that companys are never fully transparent with what they do. The dont deliever all the details considering theyr stuff to the consumers. Most consumers may not wish to know, but for those who wish to know it... its legitime for them to be informed.
LZ_Xray's Avatar
4th'd! Very kind of you Matt to volunteer to delid your new cpu!
Janus67's Avatar
Currently running prime95 on blend at stock 3770k voltage/speeds (looks like 1.112v under load @ 3.9ghz) with my venomous-x. Currently the maximum temperature is hitting about 57C (57/55/52/52 for all 4)
Ivy's Avatar
How does the Venomous X compare with the best Noctua? No baby coolers that.
Well actually, have to be compared with Silver Arrow, which seems like Silver Arrow slightly beat it. But how is Venomous X in comparison? Most important vs. stock cooler.

As far as i can tell: Not truly to impressive because some other people running 95 W TDP SB CPUs and using Venomous X, had about same results, and i mean, its a CPU with higher specified TDP and older architecture. Because the IB is so close in performance (about 7.5% average, i already told), its kinda a mess of a upgrade. The lowest upgrade i have ever seen (taking out the IGP, its no use for high end user, unless they use it for media center).

Seems like Intels main concern was simply to promote theyr IGP and adding a lot of power considering that matter. Pretty much a foul for high end, but thats a market which isnt to be taken to seriously. The mainstream is the money.
diaz's Avatar
its not about cooling performance, its about comparison between lapped vs no lapped, possible delid vs lid. venomous has pressure adjust to compensate for lid being gone.
Ivy's Avatar
Yes the pressure adjust is kinda noticed because i checked the Thermaltake site. But the issue is that a cooler of a certain size is almost incapable of, would touch other parts of the MB. So whats the point of it... when the cooler have to be downsized. Well, anyway, keep it up and may provide interesting infos about those different conditions.
SHODAN's Avatar
Jesus built your hot rod?
MattNo5ss's Avatar
It's possible I'll test a delidded CPU. I have ~$210 in my "extra Ivy CPU" fund, so I still need about $110 or so before I can do that.

Overall max temps don't matter, it's the temp differences between the different setups: not lapped IHS, lapped IHS, no HIS, TIM change + lapped IHS, etc. So, whether the cooler is the best for the chip or not doesn't matter because we'll be looking at temp differences, not overall max temps.
Janus67's Avatar
Nice find!

although we weren't the ones that pried open the CPU, always happy to see some extra publicity for the site
zkAry's Avatar
Do you have any comments to this test? that place the heat sink directly on the die and find that temperatures are similar. The weakness in their argument is that they equate their die -> paste -> cooler assembly with a die -> solder -> IHS -> paste -> cooler assembly. They don't test Ivy Bridge with a soldered IHS.
Janus67's Avatar
you can't test Ivy Bridge with a soldered IHS as IB isn't (as far as we know) soldered with any of their chips and Intel uses a special kind of solder that melts at a lower temperature than most other solders (and I'm sure is a special blend)
LZ_Xray's Avatar
Source: http://www.maximumpc.com/article/new...e_temperatures

Ugh. This sounds suspiciously like "we're not trying anymore because we don't have to".

This, to me, is the bad side of AMD backing away from the high end CPU space. I understand Intel's point completely from a cost standpoint, but not on a -K SKU CPU. Charge the extra $50 for a soldered -K CPU and give us an unencumbered enthusiast part, because an unlocked multi and a -K in the name does not an enthusiast part make.
mjw21a's Avatar
They're not backing away, they just can't compete. This is why they're going down the APU path. Convince coders to develop to use GCN and Intel won't keep up.

They can't beat Intel at their own game, so try change the game. It's the only approach in which they have a chance of actually beating Intel. They don't have the $ to try anything else.
bmwbaxter's Avatar
just hopefully putting all their eggs in this one basket doesn't fail.
mjw21a's Avatar
Fingers crossed eh?
bmwbaxter's Avatar
Indeed! Anything to drive prices down and quality up... but more prices down
MattNo5ss's Avatar
Something I try to keep in mind is that CPUs are designed to run safely up to their TJmax. So, if a CPU has a TJmax of 105C, then running at 40C load versus 100C load doesn't matter to the CPU. In my opinion, we've just become so accustomed to being able to run our CPUs so far from the TJmax (even when OC'd in some cases) that we've forgotten that the CPU doesn't care about temps as long as it's under that TJmax; and we have seemingly decided upon a threshold of when temps are "high" based on what we're used to seeing, and not based on when temps are actually high.

So, as long as CPUs run under their TJmax I'm not concerned about them.
bmwbaxter's Avatar
Does that hold true even when we are putting more volts through them causing us to get closer to TJmax? Not just by a lack of cooling at stock speeds?
mjw21a's Avatar
^^^ Good point. Just make sure that you've a case with decent airflow and you're fine.... To an certain extent anyway. My issue has always been heat buildup in my room during summer. Gaming for 5-6 hours can easily raise room temps by 10C or so. The hotter a chip runs then the more heat I'm exhausting into my room, or is it? I'm curious as to whether this chip actually produces more heat or whether its simply concentrated over a smaller area and therefore wouldn't affect my room temps.....
Janus67's Avatar
The chip itself uses less wattage so it will actually produce less heat despite running hotter than sandy bridge
bmwbaxter's Avatar
That still won't matter if the chip reaches its TJmax since the heat has a harder time escaping from the die due to a less than top notch TIM job or solder.

back in my hole I go
mjw21a's Avatar
^^^ It does for the question I was asking.... I wonder though. Is the heat output higher dude to higher electrical resistance with tri gate transistors etc
LZ_Xray's Avatar
They don't need to change the game, the game is changing whether Intel and AMD like it or not. AMD just acted on that fact before Intel did amazingly enough.

Ten years or so from now us gamers and enthusiasts are going to be the only people that even own desktops in the form factor we recognize today. The mainstream will use mobile devices and HTPC style all-in-ones with something like an APU in them.
mjw21a's Avatar
It will be interesting to see if Intel can manage to come up with a decent programmable GPU architecture without buying out nVidia....
Brolloks's Avatar
These guys did a similar test, looks like the core just runs hot period
http://www.overclock.net/t/1249419/p...ed-without-ihs
bmwbaxter's Avatar
my bad

The heat output will be less like Janus said since the chip uses less power. the point I was making was about the temps being higher due to the heat that is being produced not being able to get away from the core thus raise core temps.
diaz's Avatar
No the heat is just higher inside the chip. The heat just doesnt transfer as efficiently so it raises further before it reaches a happy balance. The cooler only sees the 77w of heat to dissipate, so the amount of heat produced is sill less than SB. As lon as power dissipation is lower, heat will also be lower.
mjw21a's Avatar
Cool, not really an issue then. Sounds like these little guys output far less heat than current AMD furnaces like mine
MattNo5ss's Avatar
Also remember that temperature doesn't measure heat.

When you start increasing volts, you won't be able to determine whether heat or voltage kills the CPU. If heat remains under TJmax while increasing voltage, then since the CPU is operating under max rated temp, voltage or defective chip are the only things left to have killed the CPU. For 24/7 use, I always keep my voltage at or below Intel listed maximum in their datasheets, so I'm always within spec on temps and voltage.
mjw21a's Avatar
Not sure I'm on the same page with you there. I understand that it's harder to pass off heat one these due to greater density and less surface area to disperse the heat, leading to higher internal temperatures..... Is that what you're getting at?
MattNo5ss's Avatar
Heat is energy, temperature (F, C, K) isn't a direct measurement of energy, Joules and Watts (Joules/second) measure heat/energy. Temperature is just a number that relates to average molecular motion.

I'm just saying that higher temperature doesn't necessarily mean more heat.
mjw21a's Avatar
Hmmmm, I think you're getting into physics there..... I don't quite get it, but thats ok..... I know you're right.
Brolloks's Avatar
Here we go again with the thermodynamics better step out
bmwbaxter's Avatar
I think what Mattno5ss is trying to say (in a far more correct and accurate way) is, wattage will dictate how much heat is put into your room. not CPU temps... correct?
mjw21a's Avatar
^^^ Lol, I understand. I'll leave that stuff to those smarter than I. In those instances I like the dumbed down answers....
muddocktor's Avatar
There are a few problems with that testing, Brolloks. First, they didn't compensate as far as I can tell on the mounting height difference, which might be making for less mounting tension. And mainly, he isn't soldering the heatsink to the die but rather he is just mounting it to the bare die with regular TIM. And I don't know how much experience you have had mounting a large, heavy heatsink to a bare die processor, but when they get as large as something like the NH-D14 on a fairly narrow core such as the IB core, it's a bit of a crapshoot to get it mounted perfectly flush.I wonder what they would have seen if they replaced the paste with something like Indigo Extreme and properly cured it if the temps would look any better. Or use the Indigo Extreme between the core and IHS and re-lid the processor. That wouldn't be nearly as good as Intel's Indium solder job but I would think it would be better than that paste Intel is using.
mjw21a's Avatar
Ok, methinks its time for me to Google "Indigo Extreme"
Brolloks's Avatar
Very true mudd, thanks for the insight, makes perfect sense
muddocktor's Avatar
Yeah, I went through the whole mounting a large heatsink (Zalman 7700) on a bare die Pentium M back in the day of the CT-479 socket adapters. It was real tricky with that large heatsink to get it mounted squarely on that little square core and the Zalman 7700 isn't nearly as massive as a D14. I'm thinking that Intel either had problems with this new core process and the Indium soldering process or they just decided that "just good enough and cheap" would do just fine this time around.
mjw21a's Avatar
Hmmmm, how does this Indigo Xtreme stuff work on regular chips? Is there any noticeable drop in temps from my trusty old AS Ceramique?
MattNo5ss's Avatar
I'm not a guru at thermodynamics by any stretch.... I know a little about a lot and a lot about a nothing

Typically, as energy increases, molecular motion increases. However, the rate at which increasing energy increases molecular motion will determine how high temps will get and how fast they will get there.

Here are four theoretical situations:
  • High wattage and slow molecular motion = high heat and low temps
  • High wattage and fast molecular motion = high heat and high temps
  • Low wattage and fast molecular motion = low heat and high temps
  • Low wattage and slow molecular motion = low heat and low temps
I believe Ivy falls into the third situation.

Indigo could be very difficult to impossible to apply to a bare die since the applications are sold in sizes to match specific CPU sizes, and they need to be applied a certain way or it screws it up. I haven't personally dealt with it because of those issues, but maybe someone with experience can chime in
mjw21a's Avatar
Hey, you've now given me stuff I understand with those 4 situations. Sweet!
muddocktor's Avatar
Yeah, I haven't messed with it either Matt. And I know they come in certain sizes for the particular IHS it is made for too. And I have doubts on it's temp ceiling being a bit low too, since you make it phase change around 90-100 C. But it's the only commercially available product I can think of that comes close to Intel's soldering process.
Brolloks's Avatar
I'm just totally dumbstruck why Intel would use such a crappy thermal interface job with a 3rd generation CPU, especially given the fact that with a die shrink you temp to get thermal issues.
mjw21a's Avatar
Nah, I think its simple. The chip will run at the speeds they set without hurting them and will perform better than AMD's current best. There's absolutely no reason for them to improve on this right now. It's simply a case of what happens when there's no competition.
MattNo5ss's Avatar
Solder keeps temps under TJmax and is expensive.
TIM keeps temps under TJmax and is cheap.

So both work perfectly fine for the CPU, which means they will go the cheaper route.

It would be fun trying to come up with a way to get Indigo installed on a bare die. It's just a little expensive for trial and error though...lol.
66racer's Avatar
Hi guys,

What seems to be the average oc range of IB and temps? I was thinking of going with a 2600/2700k and hoping for 4.8ghz, currently I have a 2500k at 4.8ghz 1.408v. BUT since temps dont seem sooooo (not great but anyways) bad with ivy I would really want at least a 4.6ghz If I switch, just curious if most of you guys are getting stable there and what the temps are. Current cooling setup keeps my 2500k @ 1.408v prime stable at 69C with ambient around 74F right now.
Janus67's Avatar
The temps can be very high depending on the chip. Last night I booted up my system running at stock volts and speeds and was idling in the mid-40s. That was with only 1.1v. Now some people have managed to get 4.5ghz out of just 1.2v, it took 1.23 for me to be able to boot and it wasn't stable up to 1.25v (haven't tried higher yet) and was hitting temps in the low 80s. Now this is with a venomous-x cooler (which I attached two higher powered panaflo fans that I normally used on my 2600k) and not your modded kuhler, but just to give you an idea.
dejo's Avatar
what I havent seen mentioned at all is the fact that using just the core to pump heat away with a tim rated at around 5w/mK may very well be much less efficient than that same core surface with 80w/mK to the IHS (much more mass and surface area) to get heat out. Seems would work much like a great pot with mass vs a cheap pot without much mass.
EarthDog's Avatar
Cant we just solder the heatsink to the core already and call it a day?
muddocktor's Avatar
Yep, if they had soldered the IHS to the core as they did with SB and most earlier Core-based lines, the IHS would definitely add significantly more area to dissipate heat. The solder junction is so much more effective at helping transfer the heat than regular TIM, it helped lower power density quite a bit. And even though IB has a 160 mm core, I imagine at least half of that is dedicated to the GPU part of it and isn't part of heat transference directly unless you are using the built-in GPU. So effectively, you are trying to transfer the heat from just 80 mm of active core. I think this is another reason why the people who have delidded IB aren't having much luck with getting cooler temps too. Only if you could actually do a solder junction between the IHS and core slug and keep an IHS on top of it would we be seeing decent temps.

This time around, it looks like Intel stuck it to ambient overclocking by using a less less efficient process to connect the IHS to the core. But using TIM instead of solder won't hold back max overclocks from the benchmarking crowd since they deal with cooling systems that can beat the inefficiencies of the TIM, so Intel can still proudly crow about how bad-ass their new lineup is with the marketing flacks.

I can't wait to get my new IB procs myself just to see how much of a step backwards IB will be as compared to SB. I.M.O.G. was supposed to pick me up a 3770k at MC Sunday and I ordered a 3570K from TD. Since I have their SB counterparts, it shouldn't be too hard to do some comparisons on cooling at various overclocks.
EarthDog's Avatar
From the looks of IB (top), its about a 1/3. From looking at SB, seems around 1/4 (bottom).

109333

109334
wagex's Avatar
thats what i said lol some one send me their cpu il give it a shot i would but saving moneys for the baby on the way lol. and the mancave relocation because my baby is gonna get my computer room
Ivy's Avatar
I feel its half the truth but not full.

Half because indeed we are used to much lower temps. When i tell a enthusiast that im fine when my CPU is running at 70, many of those are telling me "what... omg what a bloody heat, i run my stuff max 60", im worried that your CPU goes boom". Then i simply can answer "what a overly paranoid sense"... up to 70-80 is still pretty safe and when it goes far to high, a CPU will protect itself.

However, its not fully true when it comes to the matter of "that the CPU does not care". Because when its running permannently very high the lifespan is decreased and the stability can get a bad impact aswell. So its possible for it to cause some negative effects. However, a CPU wont get burned down by it... all of a sudden. They have a protection mechanism for, so, being overly paranoid is simply wrong.

CPUs in my history are pretty endurable stuff. I never had any CPU ever burned down in my whole life. Its always stuff and junk such as HDD or SSD or motherboard or capacitors, but NOT the CPU. Thats why i sometimes are even willing to "invest" 1000$ for a CPU because... it does serve me well for a eternity, thats almost foolproof.
diaz's Avatar
True. I agree fully with this... since Intel isn't scared to mention that the Tjmax is higher now, to me this says that there nothing wrong with hitting it. I work with many servers and desktops that run with stock solutions, usually operating in the upper range of the temperature envelope, and often toying with 90C+ temperatures. Several times I have serviced desktops where the dust was piled up so high that there was no airflow - therefore the CPUs would constantly hit throttling temps. CPU was fine everytime, the fans were just really loud.

Most GPU's I have owned have operated in the 80-90C range as well.. Those didn't worry me either. I don't see why a CPU running at ~90C during stress test worries anyone..
Ivy's Avatar
Well, i mean, its a stress test. On such a test i did reach up to 80 C and i had in mind, all fine! I mean, what to expect? You got a SFF system, a cooling which comes close to stock cooling, and then you create a abnormal and pretty unnatural load to the system? What to expect? I feelt safe and i even had in mind... there is still some headroom for a OC... and the cooling was not special at all.

So finally, its good to have a cool running system but it doesnt hurt the CPU to run a bit hotter than usual, thats indeed true.

Although, i still enjoy to stay 20 C below TjMAX (at 101 C max it means not to exceed 81 C) during stresstest. So that means in term of IB 85 C is fine for me. However, using SFF and IB, this time i may have to come closer to the TjMax... which isnt the stuff im used to it, but it should work still fine.

Reaching TjMax, which was until now extremely hard (even on SFF), is not causing destruction, but it would hurt the performance, so its surely very important to clearly stay below it. 90 C is already a bit at the upper edge because 15 C away from TJMax and as soon as the CPU gets older (electro migration) and the fan gets weaker (some dust) the temperature can go up a fair deal. So thats the main reason to clearly stay below TjMax. I may however correct the distance to 15 C this time, which is unfortunately closer than i wish it to be.
mjw21a's Avatar
Makes my overclocked FX-8120 look cool. It tops out at 61C
Ivy's Avatar
With the difference that it will need twice the power for lesser performance.

Still weird that IB cant have lesser temp because 22 nm is not the entire issue, nope. There is some other problems such as TIM and maybe a IGP which tends to add additional heat. In that term, a IGP free CPU of IB could be the freezer...

I still wish to know how is the share of transistor count between CPU and IGP. I do believe, because the CPU power barely increased at all, the real transistor count for the CPU is probably equal to the one a SB got. That means, they pumped all the headroom of the 22 nm manufacturing into the IGP. I dunno what to say in term its true. But finally, they dont have to worry, no one gonna beat theyr CPU. All they had to execute is to upgrade theyr IGP, so, lesser customers of "super low end GPUs". Of course Nvidia and AMD should be mad at them (because of the fact that the HD 4000 can compare with some of the very low end GPUs) but thats how the cookie crumbles. Intel can pretty much wipe out the low end GPU market and the CPU market will be dominated aswell.

In future, as long as the CPU (power) demand keeps that criticaly low as its used to be, i could guess that Intel is just slowly increasing the IGP over and over and over, while the CPU power will barely change. Why they should? AMD cant compare, nor anyone else. So they try to set the value at some other spots.

The final question is, if the entire matter is good or bad for IT environment? Well, the efficiency can be increased but the CPU power will be delayed, and many users may have a IGP constantly leeching around, without the need for it. A 100% DISABLE mode is critical. Im not sure about because the heat still stays same when its disabled, according to some rumours, which is amost impossible. All those disabled transistors should cool down the CPU, just by common sense. Intel should give the power users a FAIR PRICED alternate CPU WITHOUT IGP but some increased CPU power. Obviously not gonna happen unless we buy a 6 core CPU with twice the price and half the application support (which is in many terms ineffective and therefore to pricy).

So, does a CPU have FPU? Yes, indeed, but its soon mainly located at the IGP, instead it could be located at the CPU itself. Obviously, the only users who do care are enthusiasts, and are probably blind fated because CPU is still better than anything they ever saw. But how good could it truly be when all ressources only focused on CPU? Good question. Intel and Nvidia, both share something in common: They both are delaying theyr own progress for the sake of secondary possibilitys and marketing related strategys.

And AMD? Well AMD, is kinda reaching out for special solutions in some way. They try to please Apple (which is of course a very powerful deal) and are increasing GPGPU power (GCN architecture, 1D shaders), which comes at the cost of gaming-power to a certain extend. And they try to create server CPUs which are somewhat inferior for desktop (to much cores, most mainstream applications doesnt like it, but server does) but much better for server. On top of that the market of many mini system solutions, including console market, which can be totaly dominated by AMD next year.
mjw21a's Avatar
Pft, GCN and GPU compute is the future dude. If you think otherwise you're deluding yourself. Intel's IGP has only just managed DX11 and still has no support for GPGPU compute. These solutions will be the ones to watch in the supercomputing space in the medium to long term future. Should have a trickle down effect with mainstream computing also.

Intel tried to buy out nVidia a while ago however nVidia's wonderful leader decided to try block the deal and succeeded.

AMD's current solutions are stepping stones to where they need to be, the current FX's laid the basics to build upon. PD will improve that slightly but the really interesting stuff looks to be coming in 2013-2014. Interesting times ahead.
Ivy's Avatar
I didnt say that GPGPU is bad, dont get it wrong and i own a Radeon myself. Still have to express that it certainly did create some impact on gaming. But its not that critical and Radeon is still a very good gamer GPU, especially for its price, i do recommend. Although for those who want raw gaming power, Nvidia surely got some advantage, but that comes at a price. Anyway, this topic is about IB, so i guess im kinda going OT here. All i said was a rather objective view about the current development of the competition, not to defame them.
Robert17's Avatar
I guess I'll stuff this post in here, more appropriate than other threads.


IB seems to be the way to go for my next build, thanks to OCF mainly, as there are so many things going on with IB + heat, mobos, ram, sata 3, usb 3, cooling options. Man, there is sure a lot to take in. But Intel is going to introduce some higher clocked IB's in the third quarter which is just around the corner. No way to know today if they have any intentions on soldering the next issue of chips or stick with TIM. Buy now? Hold off ?


Gotta believe that perhaps some solder creators are working their mojo to turn in a solution whether Intel has requested them to or not. Would make a great sales pitch if the current solders don't offer a solution, if that is even the problem. And howzabout the cooling solutionists ? (is that even a word ?) Mo better heat sinks, water blocks, rads?


Anybody got a crystal ball? What's the word on the street?
mjw21a's Avatar
Lol, I think we're talking about different things here. GCN is a different, vector based architecture made specifically for compute type tasks. I doubt it would provide any benefit whatsoever in gaming. Once varies programs are rewritten to use the GPU side of things you're generally looking at a minimum of a 10x performance boost in those apps. The GPU as a co processor rather than something for video.
Ivy's Avatar
The FPU power of a GPU is the most powerful processor, however, its hard to make it understand "different languages", while a CPU got no problem with. Now the GCN did boost the possibility to make them understand. On the other hand, many of those changes are not beneficial for gaming, thus, some wasted ressources. Anyway, your indeed correct.
mjw21a's Avatar
Cool, we're on the same page then.

I hope Intel succeeds in buying out nVidia sometime down the track otherwise if AMD succeeds Intels going to be left behind. I don't care which company it is that's on top, we just need to have competition to bring down pricing and drive performance up.

Besides, it's not like AMD is EVER going to beat Intel on pure x86 performance.
alinosa's Avatar
This was my first thought when I read the thread with the guy that de-lidded his IVB. "My cathar storm block was designed for a bare die, I could make this proc a bare die...." Prolly shouldn't plan on destroying (with my luck) an expensive part that I don't even own yet.

Then I saw the post talking about impingement and had flashbacks to my research in buying my storm g4 block. Good times.
bing's Avatar
This situation reminds me of an old method to protect cpu die from crushing or chipped corner when mounting and during fastening the cooler base above it.

Details here -> Protecting IHS-less Core !

MattNo5ss's Avatar
Hmm... another possible use for art eraser. It would be very similar to insulating the motherboard up to the top of the IHS for sub-zero cooling, except this is on a smaller scale. Seems like a good idea
xrror's Avatar
Maybe replace the paste with Coollaboratory Liquid Pro under the IHS instead. It's about as close as you're going to get to simulating what a soldered TIM would be...

http://www.sidewindercomputers.com/colipro1.html

This would be rather nice to see tested... The Coollaboratory Liquid products can be a bit of a PITA to use, but for bare dies like GPU's and such with copper heatsink (bases) I've had really good luck with them.
zkAry's Avatar
Thanks for suggesting a testable method to simulate a soldered IHS.

I would like to establish how hot a tri-gate chip is with a soldered IHS.

If a soldered IHS doesn't make Ivy Bridge significantly cooler I'm less inclined to wait for Haswell for the reason that it might reintroduce the soldered IHS.
66racer's Avatar
Thanks for the info. Its tempting! Everyone is out of the 3770k and was thinking the 3570k but microcenter has the 2700k for $280 so kinda leaning that way right now. Aside from ipc I think the 2700k would be a better choice from the 3570k but the 3770k if kept cool for 24/7 at 4.8ghz would be great
BeerHunter's Avatar
does it matter how hot it runs? I saw a review which was benchmarking at 5Ghz and over 100C but does that matter? Why or why not?
xander89's Avatar
running your chip at high temps will contribute to CPU degredation and ultimately shorten the lifespan of you CPU
EarthDog's Avatar
It is OK. The TJmax is 105C (where it throttles down to save itself) vs SB that was 95C.

As always, the cooler the better.
bmwbaxter's Avatar
It matters because too high of temps can cause stability issues and if it gets hot enough the CPU will throttle itself to prevent damage which will have a major impact on performance. Also if you are running at 100c you could easily get a temp spike that will cause your computer to BSOD.

Definitely not the desired result.
BeerHunter's Avatar
The 920 I have gets to 75C sometimes on water and never had issues. I really don't care as long as it's stable and WC for noise purposes mainly especially with loud video cards. Anyway good info - I will try and keep it lower than 105C.
EarthDog's Avatar
I wouldnt go higher than the mid 80s when stress testing personally. Keep WELL away from 105C, just as you kept well away from 100C (had wrong # up there, sorry).
Janus67's Avatar
I would set 85C to be an approximate temp limit on the CPU, I like to keep my CPUs about 20C under TJMax at 100% load to do my best to prevent any sort of degration
TNM's Avatar
I am curious now on how the design is of the new IHS to avoid multiple mounts of a heatsink from giving problems with air pockets in the TIM?

Consider that you mount a new heatsink. You tighten it up well.
Unless the TIM is of a different design than anything else I have seen, you will potentially squeeze it a bit (I mean, you actually want to do so I think) and some of it will have to go out to the side.

Remove the heatsink and the pressure, and I would guess the IHS "bounces back" giving you air pockets between the IHS and the TIM/cpu die.

I this could get even worse I guess if you just squeeze hard on the middle of the IHS with your finger and the heatsink would always eventually stop against the side walls of the IHS (depending of course on exactly how tall these are vs. the die, I am sure it is practically the same, but is it accurate enough?).

This would eventually leave it to the TIM you put on top of the IHS to make sure the IHS gets "pushed down enough" to remove any air pockets (potentially benefitting the use of rather thick TIM like arctic silver or ceramique?) if that is even possible on a second mount.

Could it be that the second mount of a heatsink will never ever get the same quality in heat transfer as the first? Just like reusing the TIM on a heatsink multiple times?

Yes, I realize the IHS most likely is is pretty stiff as well (I have not seen the IB one, but older IHS's I have played with was not easily bendable), but I assume there is some microscopic bending of it when you tighten a heatsink well on top of it so the center part will get squeezed down a bit.
ehume's Avatar
Now that is quite a thought. Makes sense. They'd really have to tighten up the IHS/CPU join to keep that from happening.
diaz's Avatar
You guys worried the chip will degrade over time? It shouldn't since it is protected at 105C.. Its not going to start melting, the melting point is much higher than that. Silicon melts at 1400C for example.. of course there are other materials used on the chip, but just saying that if they engineered it to work at 104C and protect itself at 105C, then it will be fine.

Unless someone here has concrete evidence that high temps causes CPU degradation to many people? The only chips I have seen destroyed are ones with the safeguards/throttling turned off and chips pushed WAY beyond their engineered realm of operation.

A barrier I would see is if you set it to 104C, and then your ambient temps increase on certain days. So you'd want your max ambient delta to be considered. So something around 90C, expecting a ~+10 fudge factor on your ambient.

Another situation is cooling degradation. If your PC collects dust, with time you might gain some temps, so keep an eye on temps and dust PC often
xander89's Avatar
its also going to cause stability issues, i doubt you would get a chip that can run at 100 c under prime that would stay stable. I would assume most chips are going to run into stability issues when hitting the high 80's/ early 90s
Janus67's Avatar
From reading reports of SB users that kept their temps/voltages too high to have degradation issues is my only source. I've always kept my chips as cool as I could and never put a ton of load into them (beyond benchmarking which I try to do SubZero and a quick stability test + gaming)
EarthDog's Avatar
Lulz, Im not worried about it melting!!

I wouldnt be too worried about degredation with the warm temps personally. But with stability... the cooler the better.

http://hwbot.org/forum/showthread.php?t=45302
diaz's Avatar
What are the symptoms they described as CPU degradation?
EarthDog's Avatar
Not sure what "they" said, but degredation on these things that I have seen are, higher voltage needed for the same overclock, unable to reach the peak clocks anymore...etc. Those are pretty typical symptoms of a degraded chip.
diaz's Avatar
I see... maybe a degraded motherboard filter section? Because silicon is 0 or 1, vs filter section capacitors can degrade over time or if pushed too hard (voltage), or beyond their working temperature..
EarthDog's Avatar
Highly doubtful its the mobo. I have seen people drop a degraded CPU in another board with the same results.
diaz's Avatar
Ok, point taken.. Thanks for the info
bmwbaxter's Avatar
Degradation can happen on chips that have only been run subzero, so it is not always temps that degrade chips. Once voltage is added you can start to degrade your CPU even below it's TJmax.
jason4207's Avatar
If you never really stress your chip much then electromigration occurs much more slowly. When running stress tests you get much higher current which is when the most "damage" occurs.

Lower temps allow electrons to flow more freely and thus electromigration is lower for a given load. But under extreme low temps we push extreme loads (more voltage and current). Push too far and some of those traces pop like fuses at any temp.
bmwbaxter's Avatar
Like running 2.2V through an FX-8150?
GeneO's Avatar
It is not the transistors - your ones and zeros - but the conductive traces on the chip that get eroded away over time by the current flowing through them. This erosion is accelerated by higher voltages and temperatures. The process is called electormigration. This erosion, over time, results in connections with high resistance, broken connections or shorts. The first effects are usually loss of stability requiring higher volts or requiring you to lower the frequency. Increasing the voltage accelerates the process even further and eventually the processor will fail.

If you operate the processor within its design parameters, it should last at least its rated lifetime. Operating outside accelerates the degredation, but if you keep your processor cool and don't apply too high a voltage, it will still last long enough. Electromigration depends more strongly on temperature, than voltage,

The 105C is the temperature above which the processor can be permanently damaged, immediately.
Suppressor1137's Avatar
Has anyone tried ripping off the pad and put it in oil? should be interesting to see if that were to hold up well for temps. *by oil, I mean full mobo submerge*

I don't know how effective oil is compared to air/water, I just know it is a viable alternative.

If they were to do that, the next step would be to compare a 2700k with it(or w/.e the equiv is) to see how the "pressureless" method works.

Circulation would be key...chilling would be a bonus.
bmwbaxter's Avatar
Your temps would be worse guaranteed.
jason4207's Avatar
At the lowest level silicon is actually analog. Threshold voltages are pre-defined which tells the circuit whether a particular voltage value will be a 1 or a 0. Above a certain threshold voltage the signal is considered a 1 and below a certain threshold voltage it is considered a 0. If it's in the middle then the output is undetermined and can go either way.


In oil cooled setups you still need a massive metal HS to spread the heat out from the chip effectively. Die to oil will still have the heat very concentrated at the die. Oil is far less effective than metal at wicking concentrated heat away from the die.

The only way I see too address this issue is to solder a HS or water block directly to the die.
hokiealumnus's Avatar
It appears Impress PC Watch has proven this to be accurate. VR-Zone is where I saw it originally and here is the original source (you'll need a translating browser).

Their results (image courtesy Impress PC Watch):

109791

VR-Zone can have the last word (though I think their last sentence is a stretch).

I.M.O.G.'s Avatar
Interesting, looks like they tested by putting the heat spreader back on, but just putting different grease between the die and IHS.
Ivy's Avatar
Oil is a difficult matter. Some sort of oil can penetrate a gape which can not be penetrated by water by any means. So, the layer can be reduced by several times. It could be that thin that the connection between heatsink and die is the smallest distance possible. That means, the potency in theory is very good, however, its hard to have that oil at a stable condition. I know some sort of oil, it could penetrate my fingers, even when i use all the force possible in order to provide zero gape, its still not zero.
Archer0915's Avatar
But the question is this: Is the CPU really hotter or is the measured temperature hotter because of internal factors such as density? When I get time I will test this unless someone has an SB and an IB they can check?

If someone can check this sooner than me (My IB is busy working on some reviews) please do.

What I want to do is find a ratio of internal measurement to MB temp on both. If the external temps are similar but the ratios are way off then it could be due to design and location.
hokiealumnus's Avatar
Here's another - replaced stock TIM with something better and gained eight degrees.
Bobnova's Avatar
Mine'll be here Wednesday. First thing is stock:stock comparison with SB and my kill-a-watt.
Archer0915's Avatar
Well SuperPi / mod 1.5 32m gives me 38 MB sensor and (44-46-47-46) on the cores at stock with a 66w system draw with my 3570K.
Archer0915's Avatar
@ IDLE I get (29-32-33-32) CPU and 22 on the MB CPU sensor stock speeds with 45w draw.
Robert17's Avatar
So, is anyone on the phone with Intel yet? I don't have their number, well, at least I don't think the receptionist has much pull there.
Archer0915's Avatar
Why call Intel? The chips run fine.
Robert17's Avatar
Just thinking that maybe, just maybe, they might want to know that there's a better TIM out there since they apparently want to use TIM to mate the CPU to the IHS. That would make it easier for the system builders to decide to use IB chips in their rigs; afterall, who wouldn't like to buy a cooler chip out of the box?
Archer0915's Avatar
The issue is cost. Intel counts every penny and the fact is at stock the chips are fine. Now I will agree the K might need some special attention because they are supposed to be made to OC.
Bobnova's Avatar
+1 to this.

Plus, they OC just fine. Intel doesn't care how far they OC, they just want the extra $XX for the ability to OC at all.

I think what everybody is forgetting is that Intel exists for the benefit of Intel. Not us. Not the end user. Intel will do what is best for Intel.
Given that AMD cannot compete, or even come close, to Intel in any way right now there is absolutely no reason for Intel to spend more money on soldering the chips or using a better TIM.
Leave aside the fact that you want them to, and ask yourself "why would they?".
Archer0915's Avatar
Worth repeating. I do hope that AMD comes back though because I fear the 500 dollar entry level CPU coming back.
ehume's Avatar
I wonder if we could persuade Intel to put out an i7 3770X version, with a soldered IHS for people who want to go to the thermal edge. I say this because the Vcores involved seem rather low, so it's not Vcore that's limiting now, but temps, which can be controlled with better conductance.

They might cost more, but a soldered cpu would sell one or two chips, I think.
xander89's Avatar
feel like this convo has gone round in a big circle. i said the exact same thing about 4 pages ago hehe, people are underestimating how much cutting a few $ off the production cost of each processor will earn them.

As bob have said intel exist for intel, they are a corporate company and they dont care about us. Simple fact. They only reason they have up until now is because they had competition ( kinda) so had to listen to the consumer. Now they are so far ahead they can cut corners like this and it wont make the slightest bit off difference. Either we are going to buy IB or SB for the price, no1 would seriously considering AMD at this point.

Its win win for them and they are counting the cash......

Robert17's Avatar
There is no disagreement from me on this. Call me naive. When our company can produce better, safer, more enduring products, we do, cost of production becomes a matter for the customer to determine whether they want or not in the improved product. It's called Best Practices. I guess I shouldn't assume that Intel would engage in such.
SF101's Avatar
so charge us back the 10 cents to put better tim on the cpu then.. or the 50 cents to put solder on it.

i have a feeling in 6 months intel will release the ivybridge -e's which will be no more than a soldered ivy for the 2011 platform.
txus.palacios's Avatar
They should have had at least soldered the K CPUs.
xander89's Avatar

i think that is the general consensus of everyone here.
bmwbaxter's Avatar
I was hoping retail chips would have been.

But maybe intel did this for a reason (besides profit), like maybe the new chips had too high a failure rate after the solder process?
Bobnova's Avatar
Thing is, it works fine at stock. It's fine in practice.
It's us misusing it that causes higher temps, and even then it's fine really.
If it tended to explode and injure people best practices would be rather important. I doubt your company spends an extra 10% to allow 0.1% to push things a little further beyond the warranty.
SF101's Avatar
I don't think I'd put it that way bob your calling overclockers the 0.1 % but overclocking in general is become pretty common now id say the 0.1% are the people going sub ambient and those people don't seem to be having any issues pushing the chips. its the regular joe who bought a good air cooler or a liquid cooling system experiencing the heat.

And the majority of the people having problems are the average wc/aircooled people.

and out of the people that buy K specific chips id say they are 90% or so of the K sales
I.M.O.G.'s Avatar
I get the point bobnova is making.

At the same time its a K CPU specificly for the unlocked multi. It would look better to its target market if temps were better when raising multi and vcore.

I would think Intel would prefer everyone talking about how good the performance is, not "the performance is pretty good against sandy but temps can be trouble". In that sense, I think they made an unintended mistake.
wingman99's Avatar
The K CPU's are so popular they sold out across the US and they run hot OC.

+1 the k is number 1 in sales for retail CPU
Janus67's Avatar
I would imagine it would be, especially for early adopters, how many people build their own PCs that aren't enthusiasts in some manner? Sure some, but most will likely be buying the most expensive or 'special-looking-one'.
Robert17's Avatar
I understand your point. It works fine. Does just what the spec sheet says. Great. But I tend to think that before release 'K' chips were pushed pretty hard, tested against various solders, TIMs, temps, and cooling solutions. It seems they released a product that has more potential that they could have released with some more thought to performance and market perception.

As to our company: we actually have lost some substantial market share to cheaper products that should be made to the same specs but definitely aren't. A recent sample of a "competitor's" product failed during testing; in the product it was employed within damage, injury and death were a distinct possibility. Our product costs 20% more, not 10% more. So, yeah, considering the tens of millions of dollars involved in market share, to say nothing of the millions in liability, we are taking some hits but sleep well at night.
Frakk's Avatar
This, they are a money making machine, not a public service.

They have no reason to improve there products, so why bother? especially if they can make more money by cutting a few corners.

[Smaller DIE + cheaper TIM = more profit]

They have shareholders which they have to pay big dividend and bonuses
Archer0915's Avatar
Well monitoring says I am close to 100 @ 4.6 but the Heat pipes are cool. This makes my head hurt. Could we have bad sensors or pathetic heat transfer?
MattNo5ss's Avatar
High temp measurement doesn't mean high heat output.

Archer0915's Avatar
Oh I understand this but my point that I did not make was perhaps we have reached a point where pumping cold in will be more effective than getting the heat out.
MattNo5ss's Avatar
Based on the sub-zero benching results, pumping cold in definitely helps more on Ivy than it did on Sandy
White_Pawn's Avatar
based on the chinese report. Reseating the ihs was enough to make the cpu run 15-20 degrees Celsius cooler at 4.6ghz+ and definitely helped with the overclocking. They also reported that it helped them use lower voltages as well. I am sure with a reseating ihs plus water or naked die + storm rev2 (or similar) will yield 5ghz+ with very acceptable temperatures.

Additionally, as per the cold comment, i've seen fairly high CBB and CB, but it does seem to scale well.
GoD_tattoo's Avatar
Up on Engadget this morning. Well sort of. The page says overclockers.net, yet the source is overclockers.com. Damn Engadget. Even the pictures are from here...

http://www.engadget.com/2012/05/14/i...thermal-paste/
Archer0915's Avatar
When I see some other people here who bought their CUP delid and get some results I may do the same but until then I will deal with it.

Honestly I feel that a few hundred MHz is worth the sacrifice for the technology.

At the moment I am messing with the voltages to see where I can cut some heat.

I can not say I am impressed with the clock-ability/heat of the CPU but I also can not say I am unsatisfied with the package as a whole.
I.M.O.G.'s Avatar
Archer, be careful what you watch for when people delid. Some results going both ways good and bad, depending on how the cooler attaches and other factors when delidding. Doing it, and doing it right are different things. It seems like lately people are getting the best results by replacing the TIM then replacing the IHS - I imagine thats mainly a contact thing with the die/cooler.

Thanks, they fixed that for us.
Archer0915's Avatar
I am looking for cores that are not ripped to shreds only. Results are secondary. Like I said before my last delid ripped open a core. Not good.
txus.palacios's Avatar
Guys, we're famous!
EarthDog's Avatar
They are awfully sure about something that doesnt have a answer based in facts yet...

EDIT: Well, its nice to see some actual testing done on it!
Archer0915's Avatar
Well it looks like 4.4 on this air is all I am going to get. Admittedly it is IBT and I can not think of anything rougher.

Still I am going to test it against my SB setup @ 4.5 just to see how well the faster memory and CPU do at a clock deficiency of 100.
Robert17's Avatar
This is pretty interesting. Has some links to other data and articles as well. Worth a read.

http://www.extremetech.com/computing...f-overclocking

Right, wrong or otherwise, it gives one something to ponder and discuss.
Bobnova's Avatar
Got a 3570k running at 4.6GHz stock volts (~1.25v) WP1024 temps are ~60c on a Frio Extreme with one fan on low. Not bad for the speed.
xander89's Avatar
"Ivy Bridge is still an amazing push forward performance wise" i think thats a bit of an overstatement for 3-7% over the last gen lol. I would say a gentle push in performance
xander89's Avatar
probably best to do it + 200-300 mhz, will account for the average 5% difference in performance better than 100. I could tell you the results of a 4.4 Ib Vs 4.5 Sb now.
Archer0915's Avatar
I can't really understand what you are saying there.

The IB platform at stock is better, as a whole, than the SB is when overclocked.
xander89's Avatar
I'm saying running the IB at a deficiency of 100mhz isnt really a worthwhile comparison Vs the SB. The Ib runs on average 5% faster than a SB. So in order to actually get a worthwhile comparison, run the SB at 200-300 mhz faster (roughly 5% faster clocks) well 220 to be precise
Archer0915's Avatar
Thanks for the clarification.

I look at much more than just the CPU benches. I look at the big picture and unfortunately MB manufacturers have not taken full advantage of what the IB CPU has to offer. Some of those 3.0 lanes could be used to boost the poor DMI that is quickly becoming a bottle neck.

We know that the 2.0 spec was plenty for video cards and 1.0 still does fine but the poor south bridge has been neglected. With SSDs and other fast bandwidth hogging devices out there some of those lanes need to be devoted to the south bridge.
BillHenshaw's Avatar
Doesn't anyone remember when Intel did this before? It was a big disappointment then and for me it is a disappointment now.
Differences between (Intel) Northwood and Prescott Cores When Intel went from 130nm process to 90nm process the Prescott slurped enough power to destroy capacitors and power supplies on the fleet of Dell desktops I managed at the time. Dell configured the systems for Northwood power needs and then dropped Prescotts in that over whelmed their design. MB manufacturers should have learned but it'd be wise to check all the power capabilities and have someone do an actual measurement of Ivy Bridge power draw compared to Sandy Bridge and see if the published specs are real THIS TIME.
EarthDog's Avatar
Thing is that this CPU uses less power, but runs warmer.
Archer0915's Avatar
Efficient. Converting power to work creates heat.
BillHenshaw's Avatar
If Ivy is hotter that means it's wasting more power making heat instead of processing. My guess (only a guess based on Intel's actions in history) is that the published power draw of the Ivy is much much less than the actual draw - especially when overvolting / overclocking. The data in print from Intel is best case marketing speak. The heat has to come from somewhere. More heat = more power use (as I understand the physics anyway - which is limited - I only did 111 university Physics - I never took 112).
Frakk's Avatar
So does conductive resistance, just to throw a spanner in the works

What we all need is supper conductor based transistors, IBM are making good progress on that front.
Archer0915's Avatar
Cruising the web and downloading a file:

hokiealumnus's Avatar
The published power draw (at least that published here) of Ivy is what I saw from a Kill-a-Watt. When both are running stock, Ivy pulls less than Sandy. SNB stock: 158W, IVB stock: 134W. I didn't record SNB overclocked so I couldn't make the comparison, but a fair assessment is that the gap closes when both are overclocked.

The chip itself isn't producing more heat. It's showing higher temperatures. This is partially because Intel cheaped out and used thermal paste instead of fluxless solder. If they stuck with solder, the temperatures would be much closer to what you see with SNB. They might be a little bit higher because of the process shrink, but in large part it's due to the CPU-to-IHS interface.

You're confusing high temperatures with more heat production. The two are not as related as you think in this case. Ivy runs really hot temperature wise, but it doesn't spew a bunch of heat out of my radiator. Certainly not what you'd think a CPU running at 70C+ would put out. Alternatively, when I run SNB-E loaded at 50C, its radiator puts out plenty of heat. Not so with a 70C+ IVB.
Frakk's Avatar
I'm starting to see TDP numbers as pretty useless as a measure for power draw.

My CPU has a 125w TDP, i can assure you its using more then that right now.

77w TDP- Ivy vs 95w TDP Sandy, yet its power draw is only 5w less on Ivy

http://www.tomshardware.com/reviews/...k,3181-23.html

The TDP does not dictate the final power draw, even at factory clocks, I feel increasingly the TDP number is used more as a marketing ploy.
Archer0915's Avatar
Hokie if I may.

You have two pieces of metal. The first is a 1/2" steel plate that weighs 400 pounds and the other is a thin 16GA 4" x 4" square that weighs one ounce. Now the square is 1000 and the plate is 200. The square is hotter and the plate contains more heat energy.
Archer0915's Avatar
Nobody ever said TDP=Power draw.
EarthDog's Avatar
What does Core temp say about that...
Frakk's Avatar
eh? if i understand the question correctly, 50c Cores and 65c CPU @ 4Ghz on a junk air cooler (@ full load) why? have a look at benchmarks link.
EarthDog's Avatar
Core temp shows power used in TDP (which I know is not TDP exactly, but...) with version 1.0 and above.
alinosa's Avatar
EDIT: Dang this is a busy thread... and Hokie (and others) beat me to it. I'll leave the water analogy for kicks

I think this was addressed earlier in this thread. I'll do my best to recount it: The temperature is higher, but it is not necessarily a function of "wasting" more energy. As I understand it, it isn't generating more "heat"; that temperature increase is due to the fact that the "heat" generated cannot escape the core fast enough to allow the core to maintain a lower temperature.

I put heat in quotes because there is a difference between heat and temperature that is explained quite well, again, earlier in this thread.

The way I picture what is happening: the "heat" in the core is like water filling a container. The heatsink is like the opening in the container letting water out. The Sandy Bridge opening to let the water out is bigger than the one in Ivy Bridge. So, the Ivy Bridge water container is more "full" than the Sandy Bridge container. Neither of them overflows but the Ivy Bridge is designed to run a little more full so Intel says it's ok.
Archer0915's Avatar
TDP is how much heat the cooling solution needs to be able to dissipate and nothing more. This is rated at stock speeds and voltage.

As far as heat creation that is another ball of fur Frakk and I would tell you to pick up a copy of Giecks. You need more than just applied physics you need to understand material properties and.... Engineering.
rge's Avatar
Cyclops did a nice test on stock settings, then modest OC, comparing original unmodded 3770, then changed the tim out and replaced IHS, then he ran bare die. He removed socket and got good contact with waterblock.

His overclock was modest, so not as high dissipated power as some others, but for OC settings he got:
5C better temps after changing out intels tim
15.7C better temps by running bare die
Archer0915's Avatar
Thanks. I wish I could do that with mine but I cant very well do things like that because it is not representative of what every Joe can do.
Frakk's Avatar

CoreTemp will never see my rig again as they use Qinstaller, which tries to force install adware junk.
Its not just that i don't want ad aware on my rig, i don't trust it,- again not just because its listed on virus report sites, but also because of the forceful nature in which it installs that stuff, even if you unchecked the first round of adware the second round gives you an option to cancel as its installing, even if you are fast enough to hit cancel it installs it all anyway..... not having it.

Anyway, CPU-Z reads 126w TDP, HWMonitor reads 134w Package, can't give you any more then that
hokiealumnus's Avatar
So....click "More downloads..." and get the standalone CoreTemp .zip files.
Frakk's Avatar
sorta lost faith in it as a result, i'm a bit fickle like that.

I'm involved with software myself, i know how much work is involved and with that i have been known to use the Donate button on freeware sites, as do a lot of others.

[Edit] i'm to intrigued to know where this is going... so its 124.4w
Archer0915's Avatar
If we were to dissipate all W of heat then core temps would be 0 would they not?

What I am getting at is if your CPU is reading through your software that it is at 136 TDP it means absolutely nothing. A HSF can dissipate X heat and the CPU uses V x A watts which generates some heat. This heat is not a direct result of V x A. You must consider material properties at an atomic scale, the resistance at temperature and thermal conductivity of all elements involved in the process.

I, for one, dot not have time, patience or the exact design details necessary to work this out nor do I care to.

Just know what TDP is and then separate it from all assumptions because it is really irrelevant unless you are running @ stock.
Archer0915's Avatar
Do you have an IB?
Frakk's Avatar
An IB? Ivy Bridge? no, ED wanted to know what my Core Temp TDP reading was. its an x6 1090T @ 4Ghz
Archer0915's Avatar
I just wondered because if you had one you could get into this a little deeper.
Frakk's Avatar
yeah, i might end up with one if PD flops like BD (unless i can get a new SB at a good price)

I have to say there is no hiding the fact that BD is just no where, my x6 monsters it and that is based on an ancient K8.

PD needs to better my x6 or Intel gets my money when that time comes again. But honestly, from what data is out there, i prefer SB.
jd101's Avatar
well with a vcore of 1.28 for 4.4 4.5 my temps are fine 65 under load . much better than my i7 870 @4ghz. my h80 sounded like a leaf blower when gaming lol . ib @ 4.5 should be fast enuff for 90% people out there . there`s sb-e for the other`s who need more power .
EarthDog's Avatar
You had bet you would be over 125W TDP, I thought you would not be at the clocks listed in your sig. BARELY!
BillHenshaw's Avatar
Thanks everyone for all the explanations. IMOG made it clear that the TIM was a huge bottle neck in getting rid of the heat from the new Ivy cores (Heat = BTU's, Joules, Calories as expressed in temperature of X amount of mass with X thermal conductive qualities). Thus the build up of higher 'temperatures'. I could ask about the mass of a SB versus the IB (I realize I assumed that they are pretty close) but the real bottom line for me is:

How fast will it go? (compared to the competition {SB and even AMD})
and
How much will it cost? (Thanks for the kill a watt info - real costs)

I was hoping for cooler cores and higher clocks and lower electric usage. This just reminded me to much of the Prescott.

It makes sense from a marketing perspective to use TIM instead of solder. These Ivy's are cheaper to make and in higher demand than Sandy, so why enable huge overclocks right from the start when you can add a bump later by simply changing to solder. It's not like they're under pressure from AMD...

Thanks again - talk to you again in a couple more years...
Frakk's Avatar

No no no... lol, even at stock its probably using about 160w of power, i don't know what it actually is, i could Google i guess, but i'm pretty sure its around there.

I will try to put it another way, the 1075t / 1090t / 1100t - FX-8120 / FX-8150 all have a 125w TDP, those Thuban and BD's are all identical CPU's, they simply have different (stock) clock settings (as set by AMD in the factory) the lower clocked ones draw less power than the higher clocked ones, again with identical TDP numbers stamped on the box.

TDP is simply a base number, the actual power draw depends on what variant of the CPU you have, in the same way my 1090t will draw the same power as a 1100t if i clock it to the same clocks (3.3Ghz)

Add to that the stock LLC causes the vCore to fluctuate all over the place...

Its a bit like "this is a 125w CPU,- BUT.....Not really!

Any CPU can have this and that writen here and there, doesent mean thats what its pulling out of your PSU.

That's what i'm driving at.
Archer0915's Avatar
I think we said most of that.

Let me clarify as to not sound stoopid!

"TDP is simply a base number, the actual power draw depends on what variant of the CPU you have"

It is not a number pulled out of the etherial plane though. I will make another post.
Bobnova's Avatar
No....

TDP is the absolute most that any CPU with that model number will draw given absolute worst case scenario for loading at stock settings.
Unless your CPU is the worst leaking X CPU ever, it may well not hit TDP stock.
A 3.3GHz 125W CPU OC'd to the same settings as a 3.4GHz 125W CPU may well use more than 125W of power, because it was binned for 125W at 3.3GHz. It also may well use less if it was a low leakage special to begin with.

What the TDP says is that at stock clocks the CPU will not draw more than that. Doesn't mean it will draw that.
Any OCing and the TDP has just become completely irrelevant as you don't know what the actual base wattage was, and without that you can't calculate the new wattage.
Frakk's Avatar
Thats a relief, i was not at stock in CoreTemp, i was @ 4Ghz, i even rammed up the vCore to 1.4v to see if it would shift from 124w. it didn't, i don't think its reading what its actually sucking out of my PSU.
EarthDog's Avatar
This is drifitng a bit off topic so I digress after this, but Frakk, go buy a kill a watt meter and see what your system pulls at idle AT THE WALL. Your CPU cannot be using more than your TDP at idle (that makes no sense actually).

The 3770k I have at 3.7Ghz stock volts pulls 138W peak while 100% load folding. I idle on that PC around 85W or so, thats with a pump and 4 fans mobo/ram.
Frakk's Avatar
Now that makes more sense.
I remember when i first arrived here, you said my CPU was leaky and you would love to get your hands on it to freeze it in your house of CPU horrors.
I'm still trying to determine if it is as you said.
One thing i do know is it runs among the coolest of the Thuban line overclocked at nice low volts (completley stable) check link in sig.
That doesn't look like a leaky CPU to me, unless i'm missing something? it seems among the best Thubans there are.
Do tell

@ ED, have been intent on getting one to satisfy my own curiosity. i just keep forgetting
Archer0915's Avatar
Per Intel:

http://www.intel.com/content/dam/doc...ower-paper.pdf

Like I said TDP is not what people think.
Bobnova's Avatar
Intel recently changed their TDP designations to what Archer posted. Previously it was an absolute maximum.
I think AMDs still is.
Archer0915's Avatar
Yeah you cant compare Intel TDP and AMD TDP. Do notice that the AMD motherboards support 125w or 95w or 65w. I have never seen an Intel board with those limitations.
Bobnova's Avatar
That's why the IB TDP went from 77w to 95w, Intel rather specifically doesn't want them to exist.
EarthDog's Avatar
Wait... I thought that Intel stated it still is a 77W part but would like board partners to leave the 95W value there for............?

http://www.nordichardware.com/news/6...um-of-77w.html
I.M.O.G.'s Avatar
That is correct Earthdog. TDP of IB is 77W. It is marked higher on the package to ensure board designers did not skimp on the power section for z77, then effectively create boards that are compatible with IB but not with SB. Intel did this to control the 3rd party products in the market, and ensure a certain level of expectations/consistency for consumers.

AMD typically contrasts this approach by continuing to offer more freedom for their partners, and I'm not sure if they do this as a competitive advantage, or if its a necessity in order to get partners to play ball with them.
Ivy's Avatar
Boards "tuned" for 77 TDP are not compatible with IB, its fairy tale. IB needs even stronger cooling than SB, at least the exactly same level. So thats reason Intel dont want them to mistaken the lower TDP with "lower cooling", its simply not true. When a solution is trying to handle 77 TDP according to the old style, the CPU will be cooled to less. Because its higher density and some other issues (bad dissipation) is not making the matter any easyer than before. It does consume less power, thats the only real advantage. The system may heat up lesser externally but its critical to mistaken it with "lesser cooling demand", thats wrong.

TDP doesnt necessarely mean "power consume", nor does it mean "heat dissipated", its a matter more complicated than that, so i do understand when the stuff is rather a nuissance to Intel. I however do not support bad thermal solutions (in term soldering is better deal, why not doing so?), thats the only stuff i "throw" negatively at Intel, but not the matter about TDP values.
wingman99's Avatar
Here is another TDP explanation.

http://en.wikipedia.org/wiki/Thermal_design_power

Thermal design power From Wikipedia,

The thermal design power (TDP), sometimes called thermal design point, refers to the maximum amount of power the cooling system in a computer is required to dissipate. For example, a laptop's CPU cooling system may be designed for a 20 watt TDP, which means that it can dissipate up to 20 watts of heat without exceeding the maximum junction temperature for the computer chip. It can do this using an active cooling method such as a fan or any of the three passive cooling methods, convection, thermal radiation or conduction. Typically, a combination of methods are used. The TDP is typically not the most power the chip could ever draw, such as by a power virus, but rather the maximum power that it would draw when running "real applications". This ensures the computer will be able to handle essentially all applications without exceeding its thermal envelope, or requiring a cooling system for the maximum theoretical power (which would cost more but in favor of extra headroom for processing power).

In some cases the TDP has been under-estimated such that in real applications (typically strenuous, such as video encoding or games) the CPU has exceeded the TDP. In this case, the CPU will either cause a system failure (a "therm-trip") or throttle its speed down.[1] Most modern CPUs will only cause a therm-trip on a catastrophic cooling failure such as a stuck fan or a loose heatsink.

Since safety margins and the definition of what constitutes a real application vary among manufacturers, TDP values between different manufacturers cannot be accurately compared. While a processor with a TDP of 100 W will almost certainly use more power at full load than a processor with a 10 W TDP, it may or may not use more power than a processor from a different manufacturer that has a 90 W TDP. Additionally, TDPs are often specified for families of processors, with the low-end models usually using significantly less power than those at the high end of the family.

The power consumed by a CPU is approximately proportional to the square of the voltage according to Ohm's law:[2]


(where C is capacitance, f is frequency and V is voltage).

[edit] See also
Bobnova's Avatar
The board's CPU power delivery is what wouldn't be up to spec. Much like AMD boards that only support up to 65w, 95w, 125w or 140w CPUs.
Lower TDP = lower power draw.
rge's Avatar
Another person that delidded an IVY, he did pretty thorough testing with multiple tims at multiple OC settings. In spanish, but scroll down, good chart.
http://foro.noticias3d.com/vbulletin...d.php?t=391334

So we have seen several delided Ivy's that got 5-10+C better temps by changing out intels stock tim and replacing IHS when testing in range of 1.2ish vcore, and more with higher vcore. Beginning to wonder if intels secret sauce is some cheap generic non-conductive paste. I never sell anything, too lazy, so when Haswell comes out, going to have to delid mine....but multiple people getting similar results at similar vcore testing...
I.M.O.G.'s Avatar
FWIW, Janus67 and I tested a delidded 3770K with paste and IHS replaced, but only very shortly.

It ran cooler and with more even cores than the other 3770K we compared it to at idle. However under multithreaded load its temperature shot up approximately 40C, while the other 3770K shot up closer to 30C. The load temp readings were about the same for both, just one started out with much lower readings and its core readings were more even.

Under full pot LN2 and 1.8V, the unmodified 3770K ran fine - the delidded 3770K would throttle under load (CPUz would show the frequency dropping, and the score would be bad for the clockspeed).

This is only one delidded/replaced IHS sample however, and we did not delid/replace the IHS ourselves. It looked like a job well done however, you could hardly tell the IHS had been taken off and replaced... So I don't think this example means anything, other than that it might be kinda hard to delid and replace the IHS and get good results.

Our goal at the time was testing 3 different 3770Ks and running them on LN2, so we didn't waste time when one chip throttled - we swapped it out for the other chips. I'll be giving the delidded one another shot to make sure, this was only one mounting and a 15-30 minute test. I expected to see bad contact with the base of the cooler when I unmounted the LN2 pot from the delidded 3770k, however it looked like a perfect interface.
xsuperbgx's Avatar
Interesting link at xs regarding intel TIM and different replacements:
http://www.xtremesystems.org/forums/...MX4-PK1-LQP-IX


edit: same as a couple posts up....
(sorry i didn't see it.)
Archer0915's Avatar
My bet is most of the people recommending a delid don't own an IB or at least did not pay for it.
rge's Avatar
I wouldnt recommend deliding even if could get 10C better temps, odds arent it isnt going to help overclock much if at all, and mine for sure not at all. I wont go over 4.7 ghz for 24/7 because of voltage concerns, and my temps are ok with 4.7 1.31v, so I have nothing to gain. 4.8 takes 1.37+ vcore, not doing that 24/7.

But I will delid mine when haswell comes out, because I want to see for myself the temp difference on mine, no other reason.

I dont know that many are recommending deliding, just some curious few experimenting.
hokiealumnus's Avatar
+1. Delid if you want to for experimentation. Don't delid expecting it to help temps or your overclock.
White_Pawn's Avatar
>implying anyone is recommending anyone else to delid their CPU.

What people are trying to say is that at least a significant portion of the temperature difference between ivybridge and sandy bridge is a result of the TIM. If anything, what people may be recommending is to wait for a fluxless solder variant from intel.
bmwbaxter's Avatar
Gonna be a long wait I believe...
Archer0915's Avatar
Did you read IMOGs post in the its getting hot in here thread?

Honestly this thing is clobbering my SB in 24/7 clocks and performance and staying under the 105 (usually 90s) in IBT. At times it will drop to 4.4 under load but it is usually 4.6-4.7 under load.
wingman99's Avatar
SB vs IB 2-8% is not clobbering, more like a nudge with high temps.
Archer0915's Avatar
Have you run yours @ 4.6 with 2400 ram and done encodes with the new IGP @ 1500 MHZ?

http://hwbot.org/submission/2285066_

Look at the scores just above and just below mine and realize I do not evar post my best scores on HWBot.

http://3dmark.com/pcm7/349857

I am telling you a build around an IB will stomp a build based around a SB. I have both. I run my IB faster and it is faster per clock with faster ram. So yeah it is a stomping.
wingman99's Avatar
Your first link does not work.
Archer0915's Avatar
http://hwbot.org/search/submissions/...=on&offset=-20

6004 is mine. you tell me how much difference there is? We are talking IGP and CPU improvements so yeah I say it kicks my SB around.
wingman99's Avatar
Looks like PCMark 7 measures your CPU and GPU not IGP.
-Redrum-'s Avatar
With those temps though, you're trading some minor performance gains for durability. Fine for benching of course, but not really desirable for most.
xander89's Avatar
epenis scores. i think its pretty established now that if you want to pay the little bit extra and get good benches. By all means go with an ivy bridge. For normal gaming and everyday use its little or no upgrade from SB. You keep claiming that its amazing IGP and cpu improvements.

Firstly IGP is near pointless for most people. As if you are forking out that much for a computer 90% of the time people wont be using the IGP. So who really gives a rats ass about IGP improvements? Not meeeeee. Not until its actually rivalling mid end GPU's, which i doubt it ever will.

And you keep banging on about great CPU improvements, but we all know that on average that equates to about a 5% increase across the board for real world performance. A 5% that can easily be nulled with a higher clock on the SB and with lower temps. Im not saying that IB wont be worth it when they sort out their TIM issue. But until then it seem illogical to invest in an IB. Intel are keeping the prices of the SB roughly equal to IB because they know that they will not sell half as many IB CPU's if they lower SB prices. Hell that's what ebay is for XD. Pick up a 2500k or 2700k for 50 cheaper and your laughing.

Ivy bridge is just an interim Chip until haswell, its as simple as that. Id prefer to save the 30-50 and fork out extra when haswell is released. The only major advantage at the moment for me is the higher clocked RAM. Then again if i slacken the timings on my corsair ram i can hit 2200-300mhz with a minor V bump. going to 2800 is somewhat pointless when you have to slacken the timings so much to do so. Yes higher freq Vs ram timings is a debated subject but its the general consensus that high freq do = slightly better performance over tightened up timings. But then again you are paying on average twice as much for 2400 ram than for 2000-2133 ram so could get 16 Gb of 2133mhz ram for the price of 8gb of 2400 ram.
EarthDog's Avatar
Durability is not sacrificed...first of all.

Next, that igor can be used WITH your gpu so ut could help those with not as good gpus
xander89's Avatar
sorry point taken, it will help fps for people with mid to low end dedicated graphics cards. When you get into the mid end+ graphics card there are literally no gains to ingame FPS. If you are running a card with gtx 460 speeds or above it kind of becomes redundant
Archer0915's Avatar
The 6150 is using the IGP only thank you much!
I.M.O.G.'s Avatar
Which post of mine is being used as a testament for delidding? My last post in this thread I mentioned testing a delidded sample that didn't work... The sample was sent to me, free, and I just tested it. It wasn't something we are actually trying to do... Even for benchmarkig.

@xander: you don't have to buy high rated ram for ivy. A lot of stuff rated 2000 and lower is clocking to 2500 or better on ivy.
Archer0915's Avatar
If you have an IB try running IBT. I have heard of some people hitting 105 at stock clocks. IBT heats them up to levels that I have never seen from any other stressing program. So I am not pushing it if I stay under 105 in IBT.

Many overclockers fear IBT. I must have 100% stable for what I do. Others can set the records I will set the sustainability bar. For a product review everyone needs to be able to take a CPU close to the levels you test at. That is why I never publish my MAX OC or benchmark scores.
Archer0915's Avatar
I mentioned it but it was the opposite. I was disagreeing with the deliding and I used you post as reason not to delid.

Here it is and you can see what it is a reply to:

I was implying that you did not seem to think it was worth the trouble.
I.M.O.G.'s Avatar
Thank you for clarifying, and for taking my post in the proper light.
Archer0915's Avatar
NP. Honestly I am getting good results and temps in everything and I have pushed it intentionally to 105 and gotten no errors in IBT. Since I started playing with settings I get good temps, 90s in IBT and a 4.4 base OC with a turbo of 47.

The issue I see is people turning off power management, setting the voltage at 1.2 or 1.3 instead of using the dynamic offset. This is not a SB and should not be treated as such. The days of just upping the volts and multi to get a great OC are going away for this CPU.
EarthDog's Avatar
+ 1 Archer...

...this is the first platform where offset is more useful (for me personally it served to confuse more so than anything). For overclocking, I would still disable power features until you are stable, THEN re-enable them.
Archer0915's Avatar
Man when I had voltage set to 1.25 trying to get to 4.5 I could not do it. Now I usually crunch for a few days for heat and error testing but because of the issues everyone was complaining about I had to use IBT. Well 4.4 could not hold the heat.

What I did do was increase the Turbo Core TDP, lowered the A just a hair and set everything to dynamic with all power saving features on to see what happened.

Now I have a higher dynamic vCore, higher bench scores and less heat. I also hate sharing at times because it means I can't say nanny nanny boo boo.
hokiealumnus's Avatar
Moar testing!
Originally saw this @ XS.
rge's Avatar
I would disagree with the upping volts and multi going away, our mobos must have very different implementations.

I am prime 12+ hrs, linx (newest) max mem stable at 4.7 with 1.31vcore, and max temps 85C linx. And zero idle bsods or whea errors. According to fluke reading vcore, I need 1.305 vcore but I run 1 notch above stable testing, ie 1.311v. I can use LLC on turbo or high and adjust vcore accordingly but as long as vcore remains 1.305 or above, Im stable. That overclock was easy and straightforward and temps are fine. And power savings would save me $2 per year with my use (for that 20W several hrs per day), and I dont use them. Sleep mode on other hand saves me over 300W, that I use.

Now look at dynamic vcore (first without power savings). If I set vcore to normal 1.16 on mine and add .140 vcore dynamic, and boot at 35 multi I get 1.16 vcore. At 40 multi gives me 1.26 vcore, and at 47 multi gives me 1.39 vcore. Basically it overvolts, like auto overclocking with higher multi. Now I add power savings, and it overvolts at load, and I get idle bsods because it undervolts at idle/low load. I have played around and got it semistable, but at cost of still higher vcore than necessary at full load, or idle bsods when lowering vcore so load appropriate. There are multiple posts with idle bsod issues from people using power saving and dynamic vid, where mhz scales up faster than vcore.

I could probably get it stable manipulating LLC, some of power features turned off, etc but would take weeks to test for idle bsods, and what a colossal waste of time, especially when it is easy, straightforward, and better temps on mine, to up multi and vcore.

As an aside, I can run prime/linx all day with dynamic vcore + power savings, since load is always high enough (even linx) to maintain high enough vcore....but idle bsods then have to be tested and adjustments made.

Some people prefer dynamic, Im not a fan yet.
Archer0915's Avatar
You are cooling with water. I am cooling with air. Yeah I should have specified that though. My Bad.

On air I could not run IBT extreme without hitting 105 strait up clocking like the SB. Moving to an offset dynamic voltage (same as the set voltage) I could keep the temps a little lower. I added the turbo back and started getting warm again so I turned on power management adjusted the TDP to 82 (I have no idea if it really did anything) and lowered the A by two.

Now my temps dropped quite a bit due to the dynamic voltage and then even more when the CPU idles down to 1600. The turbo allows me to gain when I can without really affecting heat as well.

When I had it locked at 4.4 it sucked.
hokiealumnus's Avatar
Pretty much +1 to rge. Also on water.
Buckster's Avatar
to stop getting WHEA errors during video encodes etc - my 3770k needs 1.285V in bios with Extreme LLC - so under load thats 1.319V

this is for 4.5 gig - pretty rubbish

and even though I have a decent case (Corsair 550D) and a decent Noctua cooler - I get temps up to 92-93C during handbrake encodes (which use AVX)

pretty unacceptable for a touted "overclocking" CPU I'd say - unacceptable from a temps point of view

ALSO for note - when I was trying to get 4.6 stable and WHEA error free - I had to keep upping the voltage, and my temps went up, I got to ap oint where I was almost WHEA free .. went one notch higher, temps during encodes got to 96-97C and suddenly windows logs were full of WHEA CRC errors .. every 30 seconds ... so thats a worry - seems the CPU doesn't like temps above 95C - even though it doesn't throttle till 105
Archer0915's Avatar
Well if you and RGE were not on water how would you get those temps down? Come on guys it is what was a hobby in the beginning. Enthusiasts and poor boys. When you start adding extravagant cooling into the mix it moves away from the poor boys and into the exclusive club only. I do not run water because I choose not to. I did play with some expensive cooling way back but it involved a modified 8000-BTU window unit. I have made water blocks for fun and then never used them (testing only) because where is the challenge?

A very small percentage of people use water for one reason or another and the rest of us do not for various reasons. When you discount things that could help the majority (I know you were speaking to me here and that is fine but we have a big audience you know) you may unintentionally cause someone not to even try other ways. Hey Hokie and RGE said...

Well I know that is not the case but it can be interpreted that way. Now do not make me build a cooler and take you boys to school for 24/7 clocks. I have a buddy with a machine shop itching to play.
hokiealumnus's Avatar
Very good points Archer. I just haven't had time to try your way (or with an air cooler really). Not saying it's bad by any stretch, it's just not my preference.
ktownhero's Avatar
My LLC setting for my Asrock Z77 Extreme4 has 5 levels. Level 1 being no vdroop, level 5 being max vdroop. Right now, I keep it at level 3 so I get some droop but nothing extreme.

What are the thoughts here on whether it makes more sense to run a higher offset with more vdroop versus getting rid of vdroop altogether and running a lower offset? Are there advantages/disadvantages for either method?
Archer0915's Avatar
With these chips? Who knows and there are two (or more) schools of thought there as well.

Locked voltage with droop or dynamic with no droop are two.
ktownhero's Avatar
After I posted that I did some googling around, and discovered that apparently vdroop is an intentional feature of Intel processors to keep stability when clock speeds change dynamically. It seems to be a fairly general consensus that vdroop should not be altered except in cases where extreme cooling/voltage is being used.

I guess I'll leave it at level 3 and ride the fence
Archer0915's Avatar
Well I still get my 4.4 and then some.

Current settings:
http://youtu.be/QhCGMb4CmOU

Evidently I am @ 1.35 not 1.3
rge's Avatar
Slightly more off topic...but related, news to me, but may be old news to others..

I was playing more with LLC off and dynamic vcore to get 4.7 with semi-reasonable load and idle vcore for heck of it....but then read on another forum about power savings affecting SSD performance, Toms article is here from 2009, so I tested mine.

With power savings, ie C states enabled with dynamic vcore, I get consistent 40% decrease in random 4K writes, and 10-25% decrease in 4k random read performance on my SSD versus DISABLED C states with AS SSD and crystaldiskmark. But the effect and what it affects seems to be mobo/bios dependent. I tried timing an app installation multiple times, and it took consistently longer with power savings enabled.

Toms conclusion from 2009 was:
Archer0915's Avatar
Wonder what happens when windows is set to maximum performance with no power savings? I have no power savings enabled in windows at the moment.
I.M.O.G.'s Avatar
Probably nothing. On most desktops, the power is configured the same for the CPU in savings and maximum performance I think. On a laptop, it may be more common to configure the CPU to use something less than its max power.

On a desktop, I would expect the only difference between the default profile and max performance to be in peripherals... Disk, monitor, and sleep/hibernation settings.
S4MBO's Avatar
Hey all just wondering what peoples thoughts were about if intel were going to stop using TIM and change back to solder (at least for the k-series) or not on the next revision?

Everyone's vote and input welcome

Also does anyone know how long it is generally before a 2nd revision gets released? I really want a 3570K but much rather get a soldered 3570k...
EarthDog's Avatar

+1.

I do not believe the windows power saving features reach to the CPU and its 'c states'. so adjusting that shouldnt have an effect.
thideras's Avatar
S4MBO, I've merged your thread into the main discussion about the TIM used on Ivy Bridge processors.
Bobnova's Avatar
Windows is involved with cpu c-states and performance. Fairly deeply really, it's the only one that knows what the upcoming CPU load is.
EarthDog's Avatar
But when changing the windows power options, as is the context here, does it touch those states when you go from energy savings to power user (or w/e its called)?
Bobnova's Avatar
I know for an absolute personally tested fact that it did on 775. Newer stuff I'm assuming it does.
I don't think it ever totally disables them (Power User used to, now I don't think it does), but it definitely changes where it bottoms out and/or where it prefers to sit.
diaz's Avatar
Just built my 3570 machine... checked out temps yesterday at stock settings during battlefield 3, core usage was about 70% (GTX680 pushing framerates).. well my hottest core reached 55C on a 30C ambient. That's not bad considering my room temps, but still expected a bit lower than that at room temps. I'll do a stress test tonight and see what happens.
EarthDog's Avatar
Just gave the google machine a bit of a workout and came up with this:

http://www.sevenforums.com/tutorials...ep-states.html

Is it possible that the power options in windows controls the S states but not C states which apparently effect SSD performance?
Bobnova's Avatar
Certainly possible.
Might even be "S"oftware Based CPU Power States and "C"PU Based CPU Power States.
Who knows?
Archer0915's Avatar
So it seems that we have come to a place where it is as it was. EZ overclocking is almost non existent.
rge's Avatar
Interesting question Archer....likely to be mobo/bios/cpu dependent, but on my IVY + GB ud5h mobo.....but on playing with settings on mine

If I disable C1e and C3/C6 and EIST in bios, SSD has full performance. When I go to power options, under CPU power management, I only have the option for cooling, ie presumably fan control. I do not have min/max processor state options, since my hardware setting of disabling c1e and c3/c6 and EIST makes them nonavailable, since hardware setting overrides any software control. Note, this causes an extra 20W usage while my computer is on vs turning on eist/c1e/c3/c6. I think once hardware disables eist/c1e/c3/c6, the only power savings are peripherals like IMOG suggested.

If I enable c1e, c3/c6 and EIST and overclock using dynamic, I then lose 40% of SSD 4k write and ~20% SSD 4k read performance. But, I then have access to min/max processor power states via software by going to power options, so I could "tune" min max processor states further.

This doesnt affect my ability to use sleep mode either way, as I overclock the easy way of turning off c1e, c3/c6, eist and just raise vcore/multi. My computer at idle uses 210W idling (280W including monitor). And sleep mode just 20W for everything, which goes into after 30 mins of not using computer. And of course 185-190W idle if I run same settings, but dynamic OC with C states enabled.
Archer0915's Avatar
Very interesting and thanks for the info. It may come to a choice for air cooling of what is most important to you Higher OC or maximum SSD performance. I am using a UD3 but I am on a titening schedule and I don't have time to screw with things. to look for a work around.
EarthDog's Avatar
Maximum o/c has little or nothing to do with sleep states... does it? I mean there is trouble coming OUT of them or hibernation or something, but never knew it limited max overclocks. Not to mention, sleep states are for IDLE PC's so not sure how that would effect temperatures for a maximum overclock as idle temps are always in line.

What did I miss here?

Thanks!
rge's Avatar
I just added the side note about sleep mode, to say I get 180W power savings via sleep mode which my computer probably is in 18+hrs per day, so I dont worry about 20W power savings of C1E/C3/C6/EIST while I am using it. And yep, sleep mode obviously has nothing to do with OC/temps and obviously does not interfere with SSD performance etc.

The trade off being discussed with air cooling where heat more an issue:
option 1) use dynamic vcore + c1e/c3/c6 enabled and plus is 20W power savings most of time, better temps when cpu is not max stressed, but downside is loss in SSD performance.
option 2) disable c1e/c3/c6, use 20W more power while computer in use, have higher general temps since vcore is maxed all time, but plus side is max SSD performance.

Sleep mode not relevant for above 2 options, I just threw it in there to say I get my power savings via sleep mode 180w x 18hrs, not going to worry about 20W during use for 6hrs. And since I am on water, I dont care about hotter general temps since my vcore maxed all time, though as Archer pointed out if on air, and temps are a concern, then may be something to consider.
S4MBO's Avatar
Thanks sorry for being a noob

...anyone got any ideas when intel going to release the next revision?
EarthDog's Avatar
Nope. I certainly wouldn't bet they will use solder either. There isn't a reason to for 99% of people.
manu2b's Avatar
^Yep, I'd be surprised as well.

Sticking with SB untill second half 2013 (if God gives me life till then...).
diaz's Avatar
Ok, so after using prime95 for 20 minutes, I got 58C on my hottest core, with H80 on "low". Stock everything, voltage was showing 1.120 on CPUz. Ambient was about 26C.

I think i'll check out the bios and start tweaking to see what I can get.
Bobnova's Avatar
I can't recall a time previous to SB where 4.5GHz was a gimmie.
Both my IB chips do it (or more) easily with quiet cooling and acceptable temps.
S4MBO's Avatar
What Ghz were you running your 3570k to get that on your H80?

I have a coolmaster V8 and was thinking of getting a H80 or H100 but not sure if it's worth the upgrade or not.
wagex's Avatar
with sb dont you just need multi and a dab of vcore for 4.5? some lucky buggers didnt even need v core. oh yea and pll overide, mine wont even go past 4.4 without pll OR
Archer0915's Avatar
AMD and Intel have been pretty easy of late and it seems the end of the simple one or two part OC (multi and voltage) is here for IB. Honestly with the limits put in place by Intel that effectively locked up the Bclock (1155) it is back to multi changes or very limited FSB (bclock changes). It was not long ago that a mid grade CPU cost 500-1000 dollars and people bought them. Because there was a jumper on some of their MBs they could OC. 12 or 16 Mhz (some less) then 16, 20, 25, 33 Mhz. Then they added a bit and 40 came into play. Anyway that was easy then things got a little more complicated because everything moved to BIOS. ... So today the choices are BE or K processors to get up in speed because intel has effectively taken away the FSB/Ref/BClock and add to that the core temps of the IB and this CPU becomes a tweakers toy and a novice nightmare.

And as far as 4.5? I was not speaking about any given number I was thinking about using you head to OC and not a guide.
diaz's Avatar
Bone stock. 3.4Ghz, but CPUz was showing 3.8Ghz, so apparently it was in turbo mode for the prime95.
Bobnova's Avatar
Yeah, that's all I needed for IB too though.
Archer0915's Avatar
What are you cooling with Bob? I think that is what is very relevant to this topic. We are effectively mixing nomenclatures here. There is the Air and the more than air.
Bobnova's Avatar
TRUE and Frio Extreme, both with low fan noise.
Archer0915's Avatar
Just curious what your temps are like. Mine stay in the 80s at 4.5 in everything except some of the more intense burner programs that are meant to heat things up.

What do they look like in IBT? Also are you using dynamic or locked voltage?
Bobnova's Avatar
Running FAH on a continual basis I had cores temps in the high 60s to high 70s depending on the ambient. I haven't tried IBT.

EDIT:
Dynamic/offset on one board, fixed on the other.
Archer0915's Avatar
Have you noticed a temp difference? Also I am thinking of stripping the MX4 and going back to Diamond 24. I have noticed about a 3-4 temp drop in the past few days and I can only assume it is because of the TIM is set. Well better tim will mean a few less degrees so I figure why not give the diamond a chance.
Giatrakis's Avatar
I don’t play any games if I find something good I stick to it. I have no attention insulting anyone I was shocked by the main article seriously what I believe?

The new gigabyte motherboards have double copper this heat can’t make it to the cooler effectively like it happens in a sandy bridge I believe it will spread out thought the pins to the back of the motherboard through the double copper all over the motherboard like an umbrella!! In the end you will get high temperature readings all over the motherboard.

I still can’t believe why Intel made such a strange move. Most people will not risk to open the processor apply indigo extreme or a better thermal paste that the author of the topic make it clear will fail to remove the heat effectively as the soldier did. I will not invest any more serious thoughts to this, I guess the most people will end up down clocking it I really hope the best for them, seriously.
foxbait's Avatar
I can understand the logic of using it in the non-K version but using TIM instead of solder in the K version is just very silly and penny-pinching when the consumer is already hit with a premium price but nobbled by temp restrictions. Wake up Intel!
Bobnova's Avatar
Why though, really?
What are consumers going to buy as an alternative?

It's not polite to consumers, but it's a perfectly reasonable decision on their part.
They won't lose any customers over it and they'll make more money.
foxbait's Avatar
So, greed is good?

Remind me of the 80's.
Bobnova's Avatar
Not good at all! I'm highly annoyed with them.
My point is, it's not a mistake on their part. From their point of view it is the correct decision. They lose nothing and make more money.

They aren't in the business of making people like them, nor in the business of being warm and fuzzy, they're in the business of making as much money as possible for as long as possible.
Archer0915's Avatar
That part of it is called manufacturing! Things have changed over the years and no instead of engineers running things we have MBAs and CPAs that are clueless about much of the product and design making decisions.

It (IMHO) is less about greed and more about ignorance in the beginning. The greed kicks in when the engineers get fired for protesting.
xander89's Avatar
100% correct, people seem to have this illusion that companies like intel give a **** about any of their customers. They dont, period, all they see us as is money into their accounts.

Hey thats corporate companies for you! This is why we need AMD to bring out a kick ass chip that will really push intel into having to get their act together. I would happily defect to amd if the performance was good enough. I think many people would share that sentiment, screw brand loyalty if they cant be assed to keep their customers happy. Provided AMD thrive one day intel will rue the day they used tim on their chips! or at least i hope so
Archer0915's Avatar
Hey I have been a CPU **** for years! Always had some of each but like a true **** I love the bigger stuff! Bigger numbers that is.
foxbait's Avatar
I agree. St*ff brand loyalty. Come on AMD, bring on competition and consumer choice.
xander89's Avatar
are you compensating for something archer teehee
lilchronic's Avatar
well i tried to delid my first chip and ****ed it up now this is my new chip
i5750's Avatar
I don't know I have a new ivy bridge i5 the 3470, And I am using the stock cooler since I had no thermal paste for my silentX. And I notice a major improvement in temps over lynnfield. it idles in the 30's and never gets to 60 under cinibench load. It gets to 58c highest. The i5 750 would get past 80c sometimes.
EarthDog's Avatar
Stock and a stock cooler... makes sense.

Welcome!
Leave a Comment