-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
3570K at 4 Ghz and the truth
- Thread starter Ivy
- Start date
- Joined
- Feb 16, 2008
- Location
- East Coast
If you set vcore manually to stock value, and take it off auto, vcore wont rise like that when you increase mhz. You should be able to run 4.0ghz at/near stock vcore, without much temp increase over stock settings.
OP
- Joined
- Jun 7, 2011
- Thread Starter
- #24
In my country all shops closed at sunday, i will run tomorrow to a computer shop, kind of a emercency run because its like strangling me when i have no proper working KB, thats why i fail to answer. Occasionally i succeed to answer when the KB is working for a few min after a coold start, which sounds fun but its how the cookie crumbles... making my work nearly impossible. Next time i buy several KB at once, so i never have that issue again.
Regarding heatsink, its about equal to stock cooling because the outer surface isnt the only stuff which counts. Quality of the pipes and alignment of material is just as important and i never had any problems at all. IB is the first CPU causing problems. The CPUs in the past didnt nearly reach any critical conditions, not even at 5% OC. Although the 6 core got a improved sink (made for 130W).
The other thing which is important is the dissipation, i already explained multiple times. When the CPU does not transfer the heat to the HS, then a bigger HS is indeed more effective but its scaling in efficiency is bad and when a small HS is not being heated up, the issue is somewhere between the DIE and the HS, and not necessarely the HS itself. A HS have to heat up, and in term the HS is not heating up, either its connection is bad, or the heat is not being dissipated. So, even a increase in fan speed would not improve matters because its simply not dissipated to the other side of the HS (which is critical). Usualy i could lower the temperature by increasing fan speed, but in term of IB, its just useless because i does not transfer properly. So i know exactly what people are talking about, but im still gladful for every comment, useful or useless, at least they did contribute.
The comment from "rge" is still the most useful one, because its trying to solve a issue, instead of being destructive. In term i can lower the voltage it could stay in a green temp condition, even at 4 Ghz. But with that KB i cant truly solve issues right now because i do need a working KB for (most common sense), thats why i can do nothing but wait.
In fact i had a lot of people loling at me for using "small" systems or solutions, but when they finally saw my results, they kinda was surprised when they saw the results. I was using 6 cores already into similiar systems, who stayed green in temperature (under Prime95 load). So i know what can work and what not... im not a greenhorn. However, in term of Shuttle, they arnt using the same cooler for every system. The cooler used on Z77 is a specified for 95 TDP. Although it could be possible to upgrade to a bigger one (with more pipes). So im not done with my solutions in order to fight that issue. Of course, a increased size with better thermal capability such as the Lian Li Mini Q is being considered if that problem is continuing with the hot CPUs nowadays. But it have clearly be said, the CPU is the only issue at the whole concept, so the only reason i use a increased size is a bigger HS. The whole Z77 system does not produce much heat at all, most of the heat is from the GPU when at 100% load.
I rather avoid tower systems because i plan to have 3 active systems, and i got lot of consoles in front of my TV, so having several towers in the same room is just overkill as long as there is a way to avoid, i do appreciate to go smaller and it worked so far. IB is the first CPU creating issues. When i have to use tower just in order to run a 77 TDP baby core, i kinda wouldnt understand how i did sink that far and i would clearly solve the issue because its a error into any scientific approach. That CPU is able to run into non tower systems and i will prove it, no matter how much of work i have to do. Just because manufacturers fail to build proper stuff, one of them being Intel having inferior thermal design.
What to do next:
1. Trying to lower volt
2. Using my infrared gun (very precise) and testing the temps everywhere at the HS.
3. Comparison with a high clocked (3.5 Ghz) Nehalem 6 core on a comparable system
Nehalem 6 core @3.6 Ghz (1640 Mhz DRAM): 6 core results at outter HS surface is already known: max value is 58 C. Core at Prime95 max temp is ~80 C (~25 C ambient).
IB unknown yet over 105 C at 4 Ghz/Prime95
Regarding heatsink, its about equal to stock cooling because the outer surface isnt the only stuff which counts. Quality of the pipes and alignment of material is just as important and i never had any problems at all. IB is the first CPU causing problems. The CPUs in the past didnt nearly reach any critical conditions, not even at 5% OC. Although the 6 core got a improved sink (made for 130W).
The other thing which is important is the dissipation, i already explained multiple times. When the CPU does not transfer the heat to the HS, then a bigger HS is indeed more effective but its scaling in efficiency is bad and when a small HS is not being heated up, the issue is somewhere between the DIE and the HS, and not necessarely the HS itself. A HS have to heat up, and in term the HS is not heating up, either its connection is bad, or the heat is not being dissipated. So, even a increase in fan speed would not improve matters because its simply not dissipated to the other side of the HS (which is critical). Usualy i could lower the temperature by increasing fan speed, but in term of IB, its just useless because i does not transfer properly. So i know exactly what people are talking about, but im still gladful for every comment, useful or useless, at least they did contribute.
The comment from "rge" is still the most useful one, because its trying to solve a issue, instead of being destructive. In term i can lower the voltage it could stay in a green temp condition, even at 4 Ghz. But with that KB i cant truly solve issues right now because i do need a working KB for (most common sense), thats why i can do nothing but wait.
In fact i had a lot of people loling at me for using "small" systems or solutions, but when they finally saw my results, they kinda was surprised when they saw the results. I was using 6 cores already into similiar systems, who stayed green in temperature (under Prime95 load). So i know what can work and what not... im not a greenhorn. However, in term of Shuttle, they arnt using the same cooler for every system. The cooler used on Z77 is a specified for 95 TDP. Although it could be possible to upgrade to a bigger one (with more pipes). So im not done with my solutions in order to fight that issue. Of course, a increased size with better thermal capability such as the Lian Li Mini Q is being considered if that problem is continuing with the hot CPUs nowadays. But it have clearly be said, the CPU is the only issue at the whole concept, so the only reason i use a increased size is a bigger HS. The whole Z77 system does not produce much heat at all, most of the heat is from the GPU when at 100% load.
I rather avoid tower systems because i plan to have 3 active systems, and i got lot of consoles in front of my TV, so having several towers in the same room is just overkill as long as there is a way to avoid, i do appreciate to go smaller and it worked so far. IB is the first CPU creating issues. When i have to use tower just in order to run a 77 TDP baby core, i kinda wouldnt understand how i did sink that far and i would clearly solve the issue because its a error into any scientific approach. That CPU is able to run into non tower systems and i will prove it, no matter how much of work i have to do. Just because manufacturers fail to build proper stuff, one of them being Intel having inferior thermal design.
What to do next:
1. Trying to lower volt
2. Using my infrared gun (very precise) and testing the temps everywhere at the HS.
3. Comparison with a high clocked (3.5 Ghz) Nehalem 6 core on a comparable system
Nehalem 6 core @3.6 Ghz (1640 Mhz DRAM): 6 core results at outter HS surface is already known: max value is 58 C. Core at Prime95 max temp is ~80 C (~25 C ambient).
IB unknown yet over 105 C at 4 Ghz/Prime95
Last edited:
- Joined
- Feb 17, 2010
- Location
- Manassas, VA
Might just be me that's confused, but what does KB stand for?
Most of us here are aware that the heatpipe design will make that smaller cooler about equal to a stock cooler. However, many of us also recommend to not overclock on the stock cooler, especially a hotter running CPU like Ivy Bridge. With a stock cooler or similar, the temperatures you are seeing aren't very surprising, and the difference when you OCed it isn't surprising either if the CPU is automatically increasing it's voltage as well.
Cooling running CPUs, like Sandy Bridge, could be overclocked a good amount on the stock cooler with no temperature issues.
The reason that they're so hot is that the IHS on Ivy Bridge CPUs is MUCH worse at transferring heat than most CPUs. This is because there is TIM between the IHS rather than solder. Point being that few CPUs run as hot as Ivy Bridge does, so while a very cooling running Sandy Bridge CPU might be fine in a SFF system, a Ivy Bridge is not.
Most people are aware of this, there's an easy way to tell which one is the problem. Open up the computer and feel the heatsink while it's operating. If the heatsink is warm or hot to touch, then it's doing it's job and airflow is the problem. If it's cold, something else is the issue.
Many of us here (including myself) are big fans of SFF systems. Several top cooling companies, such as Scythe, Noctua, and others, make extremely high performing heatsinks that will fit in SFF systems. Miahallen, a member here and one of the top overclockers in the world, also has a few projects where he made some extremely powerful and overclocked SFF systems with some modded AIO liquid coolers. You can read about one of them here.
I think the only reason that some people here may have not taken you seriously is that there just wasn't enough information in the original post to know exactly how you tested and the methodology of your experiment. Also, the conclusion you came to was something that many of us already found out, along with the reason why, when this article came out almost a month ago.
Last edited:
I think the idea of a small form factor pc is pretty cool, but to me picking such a design and going with an Ivy Bridge processor and then OCing is a very challenging setup.
There are many different things you can do to try and lower the temp. Like Knufire said, what you have found has been known for awhile now.
Now going past that as to ways to fix the problem.
If your not able to or wanting to replace the heatsink then there are a few other options. Keep in mind that I don't know what your system actually looks like, just going off of ideas.
First, have you tried re-seating your heatsink on the processor with a new layer of good thermal paste? Maybe you have a bad mount?
If your more adventurous you can try removing the Intel Integrated Heat Spreader and replacing their crappy paste with again some good quality stuff. This has been shown to drop the temps by a decent amount on its own.
You could replace the fan that your CPU Cooler comes with with a better quality fan.
Finally I'd defiantly manually set every voltage in your system to it's default setting. Then you could try lowering the default cpu voltage down a step and see if your system is still stable. If it is do it again. Undervolting is a proven way of lowering temps but can obviously lead to stability problems.
Pics would be helpful.
Like others have said you have picked a cpu that is known to produce a LOT of heat. Dropping it's temps while giving you a good overclock while working in a small form factor is going to be a challenge I would guess. I wish you the best of luck on it.
There are many different things you can do to try and lower the temp. Like Knufire said, what you have found has been known for awhile now.
Now going past that as to ways to fix the problem.
If your not able to or wanting to replace the heatsink then there are a few other options. Keep in mind that I don't know what your system actually looks like, just going off of ideas.
First, have you tried re-seating your heatsink on the processor with a new layer of good thermal paste? Maybe you have a bad mount?
If your more adventurous you can try removing the Intel Integrated Heat Spreader and replacing their crappy paste with again some good quality stuff. This has been shown to drop the temps by a decent amount on its own.
You could replace the fan that your CPU Cooler comes with with a better quality fan.
Finally I'd defiantly manually set every voltage in your system to it's default setting. Then you could try lowering the default cpu voltage down a step and see if your system is still stable. If it is do it again. Undervolting is a proven way of lowering temps but can obviously lead to stability problems.
Pics would be helpful.
Like others have said you have picked a cpu that is known to produce a LOT of heat. Dropping it's temps while giving you a good overclock while working in a small form factor is going to be a challenge I would guess. I wish you the best of luck on it.
OP
- Joined
- Jun 7, 2011
- Thread Starter
- #28
Ok, finally i got a working KB. Now what can be said, i got it to work at 4 GHZ without overheat @10 min of Prime95 stress testing.
Although i made some changes which could have few performance impact so i have to test the CPU performance now.. thats all i can say for now.
What i did is to limit the TDP value to
Value 1: 616 = 77 W
Value 2: 736 = 92W
That value is 1/8 of watt, so 700 is 87.5 W max value i alllowed for it to use, 10% lower than initial value.
It should not have impact on performance because 87.5W should be enough of juice at 4 Ghz.
Although i made some changes which could have few performance impact so i have to test the CPU performance now.. thats all i can say for now.
What i did is to limit the TDP value to
Value 1: 616 = 77 W
Value 2: 736 = 92W
That value is 1/8 of watt, so 700 is 87.5 W max value i alllowed for it to use, 10% lower than initial value.
It should not have impact on performance because 87.5W should be enough of juice at 4 Ghz.
Last edited:
Still using 1.2v FOR NO REASON. You can run 4Ghz at stock voltage. I can make temps hit 105C at STOCK SPEEDS if I raise the voltage for no reason like what is done here.
Set it all back to default, go to windows and load up the CPU. Note that voltage (should be 1.0-1.1 I would imagine). Now, set to 4Ghz, and MANUALLY set your voltage to w/e you saw at stock, NOW compare temperatures. You shouldnt need anywhere close to 1.2v for 4Ghz. Again we have mentioned it a couple times in the thread... try it.
Set it all back to default, go to windows and load up the CPU. Note that voltage (should be 1.0-1.1 I would imagine). Now, set to 4Ghz, and MANUALLY set your voltage to w/e you saw at stock, NOW compare temperatures. You shouldnt need anywhere close to 1.2v for 4Ghz. Again we have mentioned it a couple times in the thread... try it.
OP
- Joined
- Jun 7, 2011
- Thread Starter
- #30
Value 1: 616 = 77 W
Value 2: 736 = 92 W
At that value temps seems to keep below 85 C, however. fan may turn up at 100% load to noisy levels. But it doesnt happen at gaming condition because no game does stress that CPU at 100% (which isnt even good because system will become less responsive).
Temperature is (full load):
Pipe: 33 C
Outter surface of sink: 30 C
Ambient temp: 25 C
Compared to Nehalem its 20-25 C lower heat on the sink.
Kinda low, considering the high CPU temps. Usualy the dissipation is much higher, so that issue does remain unchanged. But my theory simply is, that the TIM only can dissipate a X amount of heat at a given time. At above 90 TDP, the dissipation will become criticaly low and only the strongest coolers could still keep it at low temps.
The benchmark seems fine, no performance impact. http://3dmark.com/3dm11/3477903;jsessionid=1smy52f2d6sv1p6jb8m8ltd64
Unfortunately i didnt find a option in order to reduce voltage, (increase only). But at least it helps a lot to reduce thermal design.
Value 2: 736 = 92 W
At that value temps seems to keep below 85 C, however. fan may turn up at 100% load to noisy levels. But it doesnt happen at gaming condition because no game does stress that CPU at 100% (which isnt even good because system will become less responsive).
Temperature is (full load):
Pipe: 33 C
Outter surface of sink: 30 C
Ambient temp: 25 C
Compared to Nehalem its 20-25 C lower heat on the sink.
Kinda low, considering the high CPU temps. Usualy the dissipation is much higher, so that issue does remain unchanged. But my theory simply is, that the TIM only can dissipate a X amount of heat at a given time. At above 90 TDP, the dissipation will become criticaly low and only the strongest coolers could still keep it at low temps.
The benchmark seems fine, no performance impact. http://3dmark.com/3dm11/3477903;jsessionid=1smy52f2d6sv1p6jb8m8ltd64
Unfortunately i didnt find a option in order to reduce voltage, (increase only). But at least it helps a lot to reduce thermal design.
Last edited:
OP
- Joined
- Jun 7, 2011
- Thread Starter
- #32
Its proprietary Z77 Shuttle design (some kind of redesigned micro ATX) with a Amibios. I could check out BIOS date but its pretty new... because the Z77 been released less than a month ago. I cant find something to decrease, but its not unusual for them to have kinda few options. Because fewer options means to save space (and parts). However, not allowing a IB CPU to lower voltage in a SFF machine is kinda a serious flaw in design, im not protecting this, its the truth. I dont even know how the *insert mindful word* they even start to think that you could actually overvolt a IB to +250 mV inside that machine? I totaly dunno, it would reach like 1.4 V and the CPU would go boom... not less than that, it goes boom inside such a machine.
I think i could to the stuff better than that in term im the designer. But nowadays "better" is not of value, its mostly "being cheaper", i hate the current market economics, because it does reduce quality to sometimes terrible levels. Produce as cheap as possible and as much as possible. While Intel is doing exactly the same, cheap IB CPUs with bad TIM (why the heck cant they even use a proper TIM?) in order to ripp off the mainstream market, and 1 year later some overpriced IB-E CPU to ripp off the "hardcore" customers. So we all happy, but those who are happy the most... guess guess.
I try not to think to much about, as long as my stuff is working as expected....
So, indeed there is several flaws, not only 1 and im not attacking a single entity, because its simply a chain of several stuff which, added together, can create a bomb. Myself doesnt feel that guilty because i tried the best i could within the given limits. I already told that i wanted to build a SFF and there is not to much options unless i go fully custom but thats massive work. And i actually would have problems finding a appropriate board for. The issue is that that stuff is not mainstream, and not even hardcore, its kinda "nerd" stuff, and thats the reason for it to create highest amount of difficultys. On the other hand, its a true challenge... and a challenge is something which can provide true experience.
I think i could to the stuff better than that in term im the designer. But nowadays "better" is not of value, its mostly "being cheaper", i hate the current market economics, because it does reduce quality to sometimes terrible levels. Produce as cheap as possible and as much as possible. While Intel is doing exactly the same, cheap IB CPUs with bad TIM (why the heck cant they even use a proper TIM?) in order to ripp off the mainstream market, and 1 year later some overpriced IB-E CPU to ripp off the "hardcore" customers. So we all happy, but those who are happy the most... guess guess.
I try not to think to much about, as long as my stuff is working as expected....So, indeed there is several flaws, not only 1 and im not attacking a single entity, because its simply a chain of several stuff which, added together, can create a bomb. Myself doesnt feel that guilty because i tried the best i could within the given limits. I already told that i wanted to build a SFF and there is not to much options unless i go fully custom but thats massive work. And i actually would have problems finding a appropriate board for. The issue is that that stuff is not mainstream, and not even hardcore, its kinda "nerd" stuff, and thats the reason for it to create highest amount of difficultys. On the other hand, its a true challenge... and a challenge is something which can provide true experience.
Last edited:
Why are paragraphs being written about this? You bought a machine with some crap-*** custom cooler, and its performance is hurting your system. Nothing about your system or your temps are valuable input for people looking for information on the 3570k. It's in a tiny shuttle case, with poor ventilation, and is running on the worst heatsink design I've ever seen in my life.
Buy a $20 cooler, save yourself probably 20-30C, and call it a day. Or don't. But stop with this thread!
Buy a $20 cooler, save yourself probably 20-30C, and call it a day. Or don't. But stop with this thread!
- Joined
- Apr 11, 2010
- Location
- London, UK
wow..... dont really know what to say to this thread so i shall explain my feelings with a gif
- Joined
- Nov 3, 2008
Why are paragraphs being written about this? You bought a machine with some crap-*** custom cooler, and its performance is hurting your system. Nothing about your system or your temps are valuable input for people looking for information on the 3570k. It's in a tiny shuttle case, with poor ventilation, and is running on the worst heatsink design I've ever seen in my life.
Buy a $20 cooler, save yourself probably 20-30C, and call it a day. Or don't. But stop with this thread!
I have seen plenty of coolers similar to that! Why do you think that lappys almost never OC? Because they use a similar cooling solution.
- Joined
- Feb 18, 2010
In the voltage section in your bios, move the selector to the cpu voltage tab and then hit the - key near the numbers on your new keyboard you just bought. That should make the voltage go lower. It doesn't get much easier than that. You could also type in 1.1 manually to see if that works as well. Don't use offset voltage either (probably wont go negative and is probably what you are trying to do), only adjust CORE voltage.
- Joined
- Jan 9, 2012
- Location
- Massachusetts
There should be a thread "Hall of Fame" section to the forum This thread is an instant nominee LOL
This thread is an instant nominee LOL
OP
- Joined
- Jun 7, 2011
- Thread Starter
- #40
Im sure we have meet some of the thousand Intel workers who are regulary posting in there in order to tell lies such as "the CPU will decrease 20-30 C on the stock cooler they delivered with". Stop sprouting nonsense else i will deliver a proof.
No, i dont do it, because it would truly drag me down to some kindergarden behaviour which is recently pretty common. All i get for proving stuff is exactly nothing but another spit.
There should be a award for someone spitting so much junk at once... i already had 130 W CPUs working fine on equal setups. Not continuing explanation because all those narrow minded people wanna hear is "big big big big".
Generally we have many people who can do nothing better than to attack me, but are incapable to write any good text. Hilarious.
About sink:
Thing is im not sure if there is a better cooler able to fit. However, as long as the cooler doesnt heat up past 35 C, there is something wrong with the dissipation, thats certain. Its usualy 20C higher, on CPUs with lesser heat. That cooler is not that worse as it seems to be, it got a copper bottom surface and the pipes seems to be made with copper with a nickel coating, many other "cheap" coolers are not using such material. Now how much material is needed depens on how many heat actually have to be transfered. In term its less then 35 C (at the pipes) a bigger cooler may not decrease issue a lot (its hard to go below ambient on air, and its only 10 C above). Its probably another spot. The HS is attached the the CPU using copper.
Stock HS of Intel is not better than that.
CPU:
Yes they are hot, thats kinda one of the bad factors.
Mount:
I dunno but doesnt seem like it. The thing with thermal paste simply is that the difference between them is pretty low. There was once a test about those and its more of a "believing" matter. Generally thermal paste is bad for heat transfer itself, but it helps to attach the 2 surfaces to each others, kind of a catalyst. The best catalyst is solder.. no doubt.
Heat spreader:
Well i kinda worry that i could damage it.
Issue is solved anyway because i run 85 C at 4 Ghz max under Prime95 at current time, i did suceed doing so by lowering TDP value. Not that i need above 4 Ghz, the CPU is to powerful for any game i ever played so far..
No, i dont do it, because it would truly drag me down to some kindergarden behaviour which is recently pretty common. All i get for proving stuff is exactly nothing but another spit.
its performance is hurting your system.
There should be a award for someone spitting so much junk at once... i already had 130 W CPUs working fine on equal setups. Not continuing explanation because all those narrow minded people wanna hear is "big big big big".
Generally we have many people who can do nothing better than to attack me, but are incapable to write any good text. Hilarious.
About sink:
Thing is im not sure if there is a better cooler able to fit. However, as long as the cooler doesnt heat up past 35 C, there is something wrong with the dissipation, thats certain. Its usualy 20C higher, on CPUs with lesser heat. That cooler is not that worse as it seems to be, it got a copper bottom surface and the pipes seems to be made with copper with a nickel coating, many other "cheap" coolers are not using such material. Now how much material is needed depens on how many heat actually have to be transfered. In term its less then 35 C (at the pipes) a bigger cooler may not decrease issue a lot (its hard to go below ambient on air, and its only 10 C above). Its probably another spot. The HS is attached the the CPU using copper.
Stock HS of Intel is not better than that.
CPU:
Yes they are hot, thats kinda one of the bad factors.
Mount:
I dunno but doesnt seem like it. The thing with thermal paste simply is that the difference between them is pretty low. There was once a test about those and its more of a "believing" matter. Generally thermal paste is bad for heat transfer itself, but it helps to attach the 2 surfaces to each others, kind of a catalyst. The best catalyst is solder.. no doubt.
Heat spreader:
Well i kinda worry that i could damage it.
Issue is solved anyway because i run 85 C at 4 Ghz max under Prime95 at current time, i did suceed doing so by lowering TDP value. Not that i need above 4 Ghz, the CPU is to powerful for any game i ever played so far..
Last edited:
