Ivy Bridge (3770K & 3570K) Results and Discussion thread

Sorry I forgot to mention that I'm using a Cooler Master Hyper 212 Evo.

that is a decent value cooler, but nowhere near what some of the top end coolers are capable of. may explain the differences.
note that there are different MO's for chips, some just run hotter than others of the same flavor- just as some clock better than others

3570k @ Asrock z75 pro3
Alpenföhn Brocken CPU air cooler
8 gb corsair vengeance low profile 1600 CL 9
gigabyte windforce x3 gtx 570
120 gb ocz agility 3
1 tb Samsung Spinpoint F3
in an antec 900 case with okayish airflow

4,3 Ghz @ stock no problem. had to raise Vcore by +0,040 volts to get to 4,4: here are my results:

4,4 Ghz @ 1.152 Vcore (which sometimes goes up to 1.160 in cpu-z), 1,752 Cpu Pll, primestable for 3:30 now.
maxtemps 77°C, mostly lower @ around 72°C. roomtemperature ~24°

Must have a good sample or smthing. Dont want to raise vcore any higher, given my rather moderate cooling solution, but i think 4,5 shouldnt be much of a problem.

http://s14.directupload.net/file/d/2889/iw2hgbk3_jpg.htm

Thumbs up ivy!

Looking for advice on my setup in my rig. I think 4.6GHz is the best I can do on this chip without skyrocketing temps. So, I have been working on tightening ram timings. Manufacturers spec is 10-12-12-31 2T. I tried going for 1T, was surprised to find it aced a double pass of Windows Memory Diagnostic (Memtest 86+) yet when I am in windows I am getting random crashing of apps / services. Setting back to 2T cleared the crashes.

I was always under the impression that Memtest was a solid indicator of memory stability. Apparently there is more to it?

Any advice on what to tweak to stabilize it?

The only voltage settings I have touched are vCore (1.20) and vDIMM (1.65) in BIOS.

The trick to 2T is that it basically adds +1 to each of the major timings (so it would be running 11-13-13-32, if I understand it correctly). You may have to try seting it at 11-13-13-32-1T and see if it works with that, then drop them down one at a time to check for stability.

It seems 4.4 GHz is as high as my 3770K will go before I need to add some voltage. Unfortunately temps are just too high (>88C) with the voltage required for stability. Maybe better cooling is in my future.

Surfrider77 said:

I was always under the impression that Memtest was a solid indicator of memory stability. Apparently there is more to it?

Click to expand...

Definitely not the case in my experience. I had a couple different faulty RAM chips back in the AMD socket 939 days that wouldn't throw an error in Memtest, but would fail Prime95 only after 2+ hours. These same chips would cause Windows to BSOD while idling.

Some while I am testing







EDIT: Made some fan changes and I think I can tweak it down to ~90 @ 4.5 with a little work.

If I can not pull heat out I will force a little cool in.

I just have a 212 Evo cooler and was running my 3570k @ 4.4ghz with a +0.10 offset stable, but under IBT my temps did hit 90C at times. I really don't mind the high temps all that much since I only really web browse, use Lightroom, run XBMC and game on my machine, but I found a great deal on an Antec Kuhler 920 so I'll be replacing the Evo with that as soon as it arrives, and hopefully that will handle the temperatures much better. I plan on leaving it at 4.4 or 4.5GHz for now, but I'll feel a lot better if it caps out in the 70s rather than 90 under load. In reality, my usage temps cap out at about 20-25C lower than what I get during stress testing.

is all I am using. It is the ARCTIC COOLING ACFZI30 120mm Fluid Dynamic Freezer i30.

It is not the cheapest and you could get better for less but I like the brand and it does a decent job.

IDK, but I'd rather have something from Cooler Master Hyper series than that Arctic heatsink.

txus.palacios said:

IDK, but I'd rather have something from Cooler Master Hyper series than that Arctic heatsink.

Click to expand...

AC is quiet and functional. I have a boxed up true. A CM V6 and several others around. I chose this because I need to simulate middle of the road.

Dam those stuff is hot. Its first time i see such temps at this forum.

Lot of worry regarding my IB at my SFF. I did wait so long to get my second PC, and now it seems to turn out to be a water cooker. Intel need to improve design, use solder and what else. I dont see that stuff running fine in smaller systems, its hard stuff. Huge systems should be miles away from heat issues but coming close to 100 C at 4.4 Ghz (medium OC), is a heat issue.

But seriously, not even a 6 core Nehalem clocked at 4 Ghz, inside a SFF does heat up that much, i tested that stuff already using Intel Burn Test 1h in a row! So all i can think of, some problem with heat dissipation at the die.

However, some people say: Not a problem if you don't overclock. It really only starts to heat up when you try to reach 4.5Ghz and pushing more then the stock voltage through. Otherwise it run just as cool.

All i want to attain is 4 Ghz, so i could be safe.

Besides: Any good comparison vs. SB at same clock? I wonder the difference.


Still wonder how to tackle 10 nm, it will be like a smolten nuclear reactor.

Ivy said:

Dam those stuff is hot. Its first time i see such temps at this forum.

Lot of worry regarding my IB at my SFF. I did wait so long to get my second PC, and now it seems to turn out to be a water cooker. Intel need to improve design, use solder and what else. I dont see that stuff running fine in smaller systems, its hard stuff. Huge systems should be miles away from heat issues but coming close to 100 C at 4.4 Ghz (medium OC), is a heat issue.

But seriously, not even a 6 core Nehalem clocked at 4 Ghz, inside a SFF does heat up that much, i tested that stuff already using Intel Burn Test 1h in a row! So all i can think of, some problem with heat dissipation at the die.

However, some people say: Not a problem if you don't overclock. It really only starts to heat up when you try to reach 4.5Ghz and pushing more then the stock voltage through. Otherwise it run just as cool.

All i want to attain is 4 Ghz, so i could be safe.


Still wonder how to tackle 10 nm, it will be like a smolten nuclear reactor.

Click to expand...

Actually they run cool externally. They put off very little heat because it all stays in the CPU.

Ivy said:

Dam those stuff is hot. Its first time i see such temps at this forum.

Lot of worry regarding my IB at my SFF. I did wait so long to get my second PC, and now it seems to turn out to be a water cooker. Intel need to improve design, use solder and what else. I dont see that stuff running fine in smaller systems, its hard stuff. Huge systems should be miles away from heat issues but coming close to 100 C at 4.4 Ghz (medium OC), is a heat issue.

But seriously, not even a 6 core Nehalem clocked at 4 Ghz, inside a SFF does heat up that much, i tested that stuff already using Intel Burn Test 1h in a row! So all i can think of, some problem with heat dissipation at the die.

However, some people say: Not a problem if you don't overclock. It really only starts to heat up when you try to reach 4.5Ghz and pushing more then the stock voltage through. Otherwise it run just as cool.

All i want to attain is 4 Ghz, so i could be safe.

Besides: Any good comparison vs. SB at same clock? I wonder the difference.


Still wonder how to tackle 10 nm, it will be like a smolten nuclear reactor.

Click to expand...

temperature of the die of the CPU != the heat output into the system. It technically puts out less heat than a sandy bridge processor, despite running hotter internally.

This is a point that has been mentioned and lost several times (maybe even in this thread). Admittedly, I just re-learned that concept myself though and it requires repeating.

So it seems IB is ~10 C hotter at stock and about 20 C hotter at ~4-4.5 Ghz, than SB. 30-100% hotter at 5 Ghz (thats for freaks, keep me out of this).

SB probably can be compared to Nehalem so when Nehalem was like 70-80 C, IB could reach 90-100 C, very close to the limit. With some luck at 4 Ghz it may "only" run 15 C hotter than a 6 core Nehalem, so it means a increase of +9 C when we take the higher TjMax into account.

Intel, i do appreciate your TjMax increase of 4 C between Nehalem and IB... thank you very much, but next time 15 C would be the equal value.

Janus67 said:

temperature of the die of the CPU != the heat output into the system. It technically puts out less heat than a sandy bridge processor, despite running hotter internally.

Click to expand...

I dont care temperature at the die directly but indirectly, the problem is that it doesnt reach the die at all, so the heat can not be dissipated. Thats what i call "problem at the die".

I would be glad when the die is very hot, so we got something to cool down and it will be cooled down. Instead it would take a insane cooler who can ripp away every bit of temperature, below ambient, away from the die, in order to barely cool down the stuff. So at ambient, i give it a low chance, as soon as the OC is high.

But main issue simply is, i dont care to much its internal heat as long as it doesnt overheat. Now im not certain if it can keep below TjMax, so it would throttle, and finally end up being a worse piece for every system who is not equipped with a special cooler.

So far, the ridiculous laughing from my PC vendor regarding the 6 core Nehalem inside a small system, did completly miss... he failed. However, now at the baby IB (4 core only) he could be true with what he said, and he didnt even inform me about, because this time he didnt expect it to be true. Anyway, the 6 core Gulftown was simply a awesome heat dispenser at the die, that was how you make stuff come true.

Too many people are confusing temperature with heat in all these Ivy threads...

Heat is the stuff which goes into the system, right? And temperature is internally. Still doesnt change the fact that the IB is hot. I got not so much problem with heat, but when it can not be dissipated into the system, it simply means that the pieces will blow up... common sense. Much better having a hot system but cool processors. The better capcitors and what else nowadays got a heat tolerance of 105 C**, its not that much of a issue. Actually, even a GPU is already throttling at 105 C so it not technically possible to go past capacitor temps. Best stuff which can happen is that i get a throttled CPU while the system stays that cold that i have to use gloves in order not to freeze my fingers, and what im gonna say? Well... sad face. So, dissipation is the most critical thing by far and Intel should do any effort in order to increase that matter.

**Well i once had a 8600 GT passive cooler. Those caps was dirty cheap (no solid state and china made) and they still survived 100-120 C for up to 3 years, until they did bulge. But since that day i never used a passive card ever again for gaming. The GPU however, at 120 C was running properly all those years (playing WoW with), and the entire PCB was almost burning all the time, it was kinda a fun experience, but im now done with. The GPU would have lasted forerver, but the cheap CAPs finally gave up, the cheapest possible, and it took them 3 years! A japan made cap... could survive 15 years.

MattNo5ss said:

Too many people are confusing temperature with heat in all these Ivy threads...

Click to expand...

Let's make that easier to understand. People are confusing temperature with the rate of transfer of temperature. (heat transfer)

Ivy bridge produces less heat, as evidenced by the low power usage but has a higher temperature than sandybridge mostly due to the TIM used between the die and the ihs.

A report has shown that reseating the ihs will yield temperatures very close to sandy bridge temperatures.

Airport analogy: an airport can be busy because there are many people going in and out. An airport can also be busy because a snowstorm has grounded all the airplanes.

So why the ..bad words.. is Intel using TIM?
Last time i had a processor using that kind of TIM it was a E4400 and that thing was running really hot. 75 C. Unbelievable but, the 990X is having same temperature. Of course, that CPU is 20 times stronger or something like that. Just dunno whats going on. As far as i can remember the E4400 was rated at 65 TDP, the 990X is double. But i told already, i dont care TDP but its not fun when they are not making any effort to reduce the temp by the max possible and then they keep telling "higher density" but dont actually spell out the true issue.