Prime is melting my 4770k, oh the humanity. Other options?

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
I'm using a Hyper 212+.

Prime95 v28.1 is nuking my chip, even at 4.3 @ 1.15v I'm hitting 100 degrees at the same 15-16 minute mark. The rest of the time it runs around 60.

I'm all about a rock solid OC, but this seems impassible temp-wise. Not willing to delid.

[Nov 28 00:23] Test 1, 1800000 Lucas-Lehmer iterations of M250519 using FMA3 FFT length 12K, Pass1=256, Pass2=48.
[Nov 28 00:26] Test 2, 1800000 Lucas-Lehmer iterations of M245759 using FMA3 FFT length 12K, Pass1=256, Pass2=48.
[Nov 28 00:28] Self-test 12K passed!

Others experience the same? Is this still the golden standard?
 

cortexodus

[H]ard|Gawd
Joined
Feb 20, 2009
Messages
1,614
I dunno man, this 28.1 is a real roaster.

I've capped 80C at stock clocks and I've got a push/pull + shrouds Kuhler 620 with Blademaster fans on it. This is the first time I've ever heard the fans kick into high gear before. And it's pretty damned cold here in the basement (~63F)

System barely tops 67C on 27.9.
 

Bluesun311

2[H]4U
Joined
Sep 21, 2013
Messages
2,523
There's something about 28.1 that isn't quite right. Heat if you like, but even at stock settings (turbo off) the 4770S I had previously couldn't run it for even a second. Instant in-app crash, system unaffected. I sent that chip to Intel and they sent me a new one I dumped on eBay NIB for a bit of a loss. The 4770K I'm running atm can do 28.1 fine but it gets hotter than anything else. Very impressed with how the Noctua NH-c14 handles it. I think RAM speed is an issue maybe, not sure.

prime doesn't seem like a very useful OC test anymore... sooner or later it will crash and you're not sure if only p95 would possibly cause it or not. Then you may find you can run a higher clock with anything normal and never crash.

I have yet to hear anyone definitively argue what is the best way to stress test a 4770k OC on air that is throttling in stress tests but not in real-world applications.

This may sound irresponsible but I basically just gave the core/cache/PCH/SA/I-O (iirc) a good voltage bump the last time I got a crash in an application and that was pretty much it. If I get a crash in the future I'll try not to grumble about it and do the same thing again.
 
Last edited:

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
I'm going to try Aida64. If it can pass an overnight, I'm going to run with it. 4.2 @ 1.1 right now, for a conservative OC.
 
Last edited:

wyqtor

Limp Gawd
Joined
Dec 30, 2011
Messages
474
For stress testing I would recommend the Intel Extreme Tuning utility instead of anything else. AIDA is good too, but I trust Intel more.
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
21,608
Not sure if it's just the program itself or something else, but Prime always seems more ripe for errors than similar programs. While Prime95 (any setting) might not ever go over 73 degrees, Intel Burn Test might go up to 80 or 81. At the same time, it's error free pass after pass. Prime might start dropping errors after a while using the same OC settings. For the sake of stability, I've bumped a few voltages just to keep Prime happy.
 

Riccochet

Fully [H]
Joined
Apr 11, 2007
Messages
27,965
I don't trust prime. It's doing something to load the cores in a way that 99.99% of us will never encounter. Use it if you want, but I'll stick with Aida64. Asus recommends Aida64 for stability testing and I'm sure they have a good amount of research in to it.
 

jolli

Gawd
Joined
Dec 17, 2010
Messages
798
I used both AIDA and Prime to set my overclock, and I found that AIDA would allow me to run settings that were unstable under Prime.

I ended up going with what was stable under Prime (4.3GHz core, ~1.22 override volts, 4.1 cache @ ~1.16V). Those settings also spiked to absurd temperatures on a few of the prime problems, but it never crashed. I bought the Intel overclocking warranty for added peace of mind.

TLDR => I'd still use Prime, just let it throttle the chip if need be when you do your final stability check. A stable system is a happy system!

Edit: I also verified those settings under the program that runs an AVX2 stress test (talk about heat!). The name escapes me at the moment...
 
Joined
Sep 7, 2004
Messages
3,014
I'm using a Hyper 212+.

Prime95 v28.1 is nuking my chip, even at 4.3 @ 1.15v I'm hitting 100 degrees at the same 15-16 minute mark. The rest of the time it runs around 60.

I was hitting 100 degrees around the 17-18 min. mark at 4.5 Ghz @ 1.32V ...I'm on a Kraken x60....I just said fuck it, if I can run BF4 without crashing then my shit is fine.
 

Pitbully

[H]ard|Gawd
Joined
Nov 29, 2011
Messages
1,155
So glad I didn't "sidegrade" to a 4770k. My 2600k has happily chugging along at 4.8Ghz@63C.
 

MacLeod

Supreme [H]ardness
Joined
Jul 28, 2009
Messages
8,177
I keep reading story after story like this. Mild overclocks setting these chips on fire. Maybe I should just cough up the extra cash for a 2011 motherboard and get a 4820K when I upgrade.

OP, a Hyper 212 is the best cooler on the market dollar for dollar but it's not the best out there. It might just be a little over matched with your i7. Thought about upgrading to something like a D14 or H100i?
 
Joined
Sep 7, 2004
Messages
3,014
So glad I didn't "sidegrade" to a 4770k. My 2600k has happily chugging along at 4.8Ghz@63C.

I honestly thought these chips run cooler...THIS SHIT OVERHEATS QUICKER THAN MY i7 930 AT 4.4 Ghz DID on Prime! That never got up past 70 C on Prime. I wouldve stuck with that for another year if I knew about all this beforehand.
 

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
OP, a Hyper 212 is the best cooler on the market dollar for dollar but it's not the best out there. It might just be a little over matched with your i7. Thought about upgrading to something like a D14 or H100i?
I think it is over matched if I wanted 4.3 or higher. But I'm cheap; the cooler, mobo, and CPU cost under $300. I'm at 4.2@1.1v, leaving it there. And I'm pretty happy with that. It's better than my 2500k @ 4.3.

AIDA64 CPU/FPU/Cache test has been running for 8 hours, temps never exceeded 69C.

Testing FPU alone, my temps can get to the low 80s, which is fine w/ me.

Bye Prime, it was good while it lasted! :)
 

mothman

Supreme [H]ardness
Joined
May 28, 2009
Messages
5,479
I've just founded a new organization called S.K.I.P. Stop Killing Innocent Processors. It was founded on the premise that overclocking and stress testing are useless endeavors, needlessly killing, shortening the life of so many innocent CPU's. ;)
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
21,608
For me, I like a combination of Prime95 and Intel Burn Test. Prime95 does a better job of sniffing out errors and memory issues. Intel Burn Test is better for pure heat.

For gaming, nothing is going to push your system (CPU-wise) like those, but they at least let me know that my stuff works. I tend to test everything pretty rigorously whenever I get any new hardware or if I think it might be time for a new application of thermal paste every 9 months or so.

Once I got my H110, stress testing helped me figure out what kind of settings I could get away with and keep everything quiet. I stay under 80 degrees at 50% fans on my 3770K...so that's dandy with me. Using Prime95 (which isn't as heat-heavy) I never even go over the mid 70's.
 

Riccochet

Fully [H]
Joined
Apr 11, 2007
Messages
27,965
For me, who games mostly on my rig, Aida64 stresses my system enough to be game stable. My temps are fine, system is stable and doesn't crash during heavy gaming. What else do I need to stress for?
 

MacLeod

Supreme [H]ardness
Joined
Jul 28, 2009
Messages
8,177
OCCT is my method of choice. AMD FX chips seem to be able to pass Intel Burn Test at much higher clocks than they can OCCT or even P95. My 8150 was passing IBT at 4.8 but crashing under games. So I pretty much what using linpack altogether and just relying on OCCT. Plus it's got a lot of handy measuring tools built right in.

I've just founded a new organization called S.K.I.P. Stop Killing Innocent Processors. It was founded on the premise that overclocking and stress testing are useless endeavors, needlessly killing, shortening the life of so many innocent CPU's. ;)

Luckily somebody else founded HardOCP which is dedicated to overclocking the balls off our processors, GPU's and possibly even RAM in an order to get more free performance out of them but mainly because it's a shit load of fun. :D
 

ZodaEX

Supreme [H]ardness
Joined
Sep 17, 2004
Messages
4,569
For me, who games mostly on my rig, Aida64 stresses my system enough to be game stable. My temps are fine, system is stable and doesn't crash during heavy gaming. What else do I need to stress for?

Unraring.
 
Joined
Dec 22, 2007
Messages
987
Agree 100% with Prime95 not being safe for 4770k. Check the Prime95 site, it was never updated for Haswell.

I would get about 10-20 minutes in and the temps would suddenly spike from 70C to 100C in a split second. It made no sense. Intel Burn Test is no better. Running Stock settings I was hitting 90C almost instantly.

I'm using the stress tester in the Intel Extreme Tuning Utility to test my OC currently. I'm going to try Aida64 as well.
 
Joined
Nov 3, 2006
Messages
973
Agree 100% with Prime95 not being safe for 4770k. Check the Prime95 site, it was never updated for Haswell.

I would get about 10-20 minutes in and the temps would suddenly spike from 70C to 100C in a split second. It made no sense. Intel Burn Test is no better. Running Stock settings I was hitting 90C almost instantly.

I'm using the stress tester in the Intel Extreme Tuning Utility to test my OC currently. I'm going to try Aida64 as well.


Its not safe for a 4770k on air you mean. I have a very modest Corsair H110 running a 4770K@4.5 and it can do it all day sitting a smig under 80C with around 25C room temp. I don't consider this cool, but it is very solid and worst case.

Aida is rubbish, I have seen a unstable OC pass with Aida64 and fail on Prime.

If I want to run a fake stress test I will open up notepad :)
 
Joined
Dec 22, 2007
Messages
987
Its not safe for a 4770k on air you mean. I have a very modest Corsair H110 running a 4770K@4.5 and it can do it all day sitting a smig under 80C with around 25C room temp. I don't consider this cool, but it is very solid and worst case.

Aida is rubbish, I have seen a unstable OC pass with Aida64 and fail on Prime.

If I want to run a fake stress test I will open up notepad :)

Around 3:55, JJ from Asus explains why Prime95 and OCCT are not safe as they are "unvalidated applications" and not representative of realistic stress testing.

Not only do they lack the complexity to test all components of a Haswell CPU, but they will cause extremely high temps on your CPU. Dude knows his shit too, he's the Asus overclocking expert.

I'll stick to AIDA64 & Intel Stress Test.
 
Joined
Nov 3, 2006
Messages
973
Around 3:55, JJ from Asus explains why Prime95 and OCCT are not safe as they are "unvalidated applications" and not representative of realistic stress testing.

Not only do they lack the complexity to test all components of a Haswell CPU, but they will cause extremely high temps on your CPU. Dude knows his shit too, he's the Asus overclocking expert.

I'll stick to AIDA64 & Intel Stress Test.

yea thanks, looked like spin the first time, looks like spin now.
 

MacLeod

Supreme [H]ardness
Joined
Jul 28, 2009
Messages
8,177
Around 3:55, JJ from Asus explains why Prime95 and OCCT are not safe as they are "unvalidated applications" and not representative of realistic stress testing.

Not only do they lack the complexity to test all components of a Haswell CPU, but they will cause extremely high temps on your CPU. Dude knows his shit too, he's the Asus overclocking expert.

I'll stick to AIDA64 & Intel Stress Test.

That is pretty interesting. Id never heard that before. Wonder if thats the same for AMD FX chips. I can get much higher clocks on say Intel Burn Test like around 4.8 but significantly lower on OCCT and P95 like my current 4.5.
 

ccityinstaller

Supreme [H]ardness
Joined
Feb 23, 2007
Messages
4,241
That is pretty interesting. Id never heard that before. Wonder if thats the same for AMD FX chips. I can get much higher clocks on say Intel Burn Test like around 4.8 but significantly lower on OCCT and P95 like my current 4.5.

That makes sense if you are using IBT without manually updating to the latest binaries that Intel releases on a regular basis...

My personal testing method is as follows:

1) Set desired CPU speed and attempt to boot into Windows. If successful go to Step 2. If not, add Vcore and try again..

2) I open IBT and using the "Extreme" preset, and all available ram, run 5 passes. If successful, go to Step 3. IF not, reboot and add Vcore and try again.

3) After 5 IBT runs, I go to 20, again using the extreme preset. (I also load my GPU to 100% to maximize temperatures in my loop). If successful, I go to Step 4. If not, well you know the drill by now.;)

4) After 20 IBT runs, I go into Event Viewer and make sure there are ZERO WHEA ERRORS* showing. If there is, I reboot and add Vcore. If not, I go to Step 5.

5) If everything above is good to go, I then go to a 24hr Prime 95 Torture Test, using Custom Preset..I set Prime to use all available threads, and every MB of memory. I then allow it to run for 24 hrs. If I can't get a stable 24hr run, I retweak and try again..

These steps are important just for dialing in the CPU Core Speed. Once I have that rock solid, I will run through some games to make sure 3D apps are stable as well. I then preceed to tweak my memory settings. This was time consuming on older platforms, especially 1366 setups, but with my 3770K I just run the default XMP setting of DDR 2133 right outta the gate.

My method is time consuming, but since I run F@H on my systems, I need to be 99.999% sure my O/C's are stable (Nobody knows 100%)..It does no good for me to crunch WU's and return them only for the data to be wrong and it have to sent out to another member to recalculate. It wastes my electric, and slows the entire project.

*For those of you that are ONLY using IBT/ADIA64 for stability testing, PLEASE, PLEASE check your Event Viewer to make sure you do not have any WHEA errors showing. If you do, 99% of the time you NEED more Vcore for the CPU. Here is a great guide that explains how to to identify them and how to correct them.
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
21,608
I'm going to have to look into the WHEA thing when I get home. Thanks for posting!
 

sharknice

2[H]4U
Joined
Nov 12, 2012
Messages
3,116
Its not safe for a 4770k on air you mean. I have a very modest Corsair H110 running a 4770K@4.5 and it can do it all day sitting a smig under 80C with around 25C room temp. I don't consider this cool, but it is very solid and worst case.

Aida is rubbish, I have seen a unstable OC pass with Aida64 and fail on Prime.

If I want to run a fake stress test I will open up notepad :)

It depends on how lucky you were with your chip. I have my 4770k@4.5 air cooled with a hyper 212. It idles in the 30s, games in the 60s, and prime will bring it up to 80. I think Microcenter had a bunch of good chips.
 
Joined
Dec 22, 2007
Messages
987
It depends on how lucky you were with your chip. I have my 4770k@4.5 air cooled with a hyper 212. It idles in the 30s, games in the 60s, and prime will bring it up to 80. I think Microcenter had a bunch of good chips.

What voltage?

I'm 4.5 @ 1.25 and I hit 100C in Prime95 Blend after about 15 minutes (when it switches to either small/large FFT, not sure which)

I can run Intel Stress test (not burn test) for hours on end without crossing 77C though so I'm calling it stable.
 

StoleMyOwnCar

2[H]4U
Joined
Sep 30, 2013
Messages
2,380
I've been doing all of my stress testing with Prime95 25.11 (not the version everyone here is complaining about)... because I'm too lazy to get anything else?

Generally I find that if it stays stable for about 45 mins or so doing large in-place FFT's (that's my stability test of choice), it's rock solid. Right now I have my 4770k at 4.5, 1.305 Vcore (anything lower than that and I get that lovely blue screen). Been solid for weeks now, no blue screening when I'm doing anything. Temperatures in that Prime95 test got to about 74-80C on my Kraken X40. During gaming they don't really go over 6x-ish. And that's in silent mode.

I'm wondering if part of the reason I needed that vcore was because one of my RAM sticks is a Corsair XMP, though. Requires some "overclocking" to hit its 1600. I probably have a crappy chip, too. Oh well, It's stable so I'm rolling with it.

I'm amazed that some people are actually able to get 4770k's to 4.5ghz on a Hyper Evo 212. I've been telling everyone on PC Part Picker to choose an AIO or a high-end air cooler (that huge Noctua for instance). From everything I've seen air cooling is not sufficient to do any kind of OC with these Haswells.
 

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
It depends on how lucky you were with your chip. I have my 4770k@4.5 air cooled with a hyper 212. It idles in the 30s, games in the 60s, and prime will bring it up to 80. I think Microcenter had a bunch of good chips.
What volts, version of Prime, and how long are you testing?
I've been doing all of my stress testing with Prime95 25.11 (not the version everyone here is complaining about)... because I'm too lazy to get anything else?

Generally I find that if it stays stable for about 45 mins or so doing large in-place FFT's (that's my stability test of choice), it's rock solid. Right now I have my 4770k at 4.5, 1.305 Vcore (anything lower than that and I get that lovely blue screen). Been solid for weeks now, no blue screening when I'm doing anything. Temperatures in that Prime95 test got to about 74-80C on my Kraken X40. During gaming they don't really go over 6x-ish. And that's in silent mode.

I'm wondering if part of the reason I needed that vcore was because one of my RAM sticks is a Corsair XMP, though. Requires some "overclocking" to hit its 1600. I probably have a crappy chip, too. Oh well, It's stable so I'm rolling with it.

I'm amazed that some people are actually able to get 4770k's to 4.5ghz on a Hyper Evo 212. I've been telling everyone on PC Part Picker to choose an AIO or a high-end air cooler (that huge Noctua for instance). From everything I've seen air cooling is not sufficient to do any kind of OC with these Haswells.
With that version of Prime, I can do 4.5ghz under a 212+ for hours with low temps. It all depends on what stress test you choose. With Aida64, I had to turn it down to 4.2.
 
Last edited:

StoleMyOwnCar

2[H]4U
Joined
Sep 30, 2013
Messages
2,380
With that version of Prime, I can do 4.5ghz under a 212+ for hours with low temps. It all depends on what stress test you choose. With Aida64, I had to turn it down to 4.2.

Which test were you doing, blend or something? If you were doing the one I was doing (large in place FFT's), that's pretty amazing, because I had to kick my Kraken into higher fan usage just to keep it at 74-80 eventually. The temps just keep slowly creeping upwards on it.

Heck, my Hyper Evo could barely keep up with stock clocks under 70C with my ambient temps here in Georgia (granted not with how low they are now).
 

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
Yes, I've always done blend. *hides*

Too many variables to compare apples to apples.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
61,367
I'm going to try Aida64. If it can pass an overnight, I'm going to run with it. 4.2 @ 1.1 right now, for a conservative OC.

Except that Aida isn't a reliable indicator of stability. It will validate a system as stable when it clearly isn't. I've had OC's tested by Aida that tested successful for several hours. When I launched an application like Handbrake on that same system with those same settings the system crashed immediately. You'd be seeing lots of 5GHz OC results in the reviews if I could believe Aida for one second.
 

david_

2[H]4U
Joined
Jan 30, 2012
Messages
2,502
So aida is out. Prime is nutso. OCCT or IBT or Intel tuning test? Or all of the above?
 

StoleMyOwnCar

2[H]4U
Joined
Sep 30, 2013
Messages
2,380
Yes, I've always done blend. *hides*

Too many variables to compare apples to apples.

Well, I'm not sure why you're hiding. I'm actually not sure in any way that blend is not the optimal way to do stress testing. I just noticed that the one I picked for my stress testing had "maximum heat" as its caption, so I was like "well, this would be the one to do, wouldn't it?" No particular factual basis there, just my intuition.

I didn't even know that the Prime version I was using was outdated as heck back then. I'm rather new to overclocking CPU's in general... I've had the tendency to leave mine at stock until now. Overall the Prime95 version I used seems to have given me perfect stability for literally all the gaming I've done though.

Except that Aida isn't a reliable indicator of stability. It will validate a system as stable when it clearly isn't. I've had OC's tested by Aida that tested successful for several hours. When I launched an application like Handbrake on that same system with those same settings the system crashed immediately. You'd be seeing lots of 5GHz OC results in the reviews if I could believe Aida for one second.

You know the one thing I haven't tried for my actual real-world stress testing after trying Prime for a bit is video conversion. I've been gaming on this setup fine for weeks (including hours and hours of Civ 5 with a friend hahaha). Never tried anything beyond that. Should probably try that some time... it's just I don't need to convert videos very often lol.
 
Joined
Dec 22, 2007
Messages
987
Except that Aida isn't a reliable indicator of stability. It will validate a system as stable when it clearly isn't. I've had OC's tested by Aida that tested successful for several hours. When I launched an application like Handbrake on that same system with those same settings the system crashed immediately. You'd be seeing lots of 5GHz OC results in the reviews if I could believe Aida for one second.

What do you recommend then?

Is the Intel Stress Test a valid indicator?

Prime95/IBT seem to produce insane temps.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
61,367
What do you recommend then?

Is the Intel Stress Test a valid indicator?

Prime95/IBT seem to produce insane temps.

Prime95 does produce ridiculous temps. And that's what we are currently using. We generally don't have the luxury of long term testing in tight review schedules. So an hour or so is what we normally get. Now we also tend to set for manual voltage to make things easier. You'd be able to use adaptive voltage most of the time in the real world. The problem is that some applications cause the CPU to call for unrealistic voltages from the IVR. This leads to insane temps and even worse power usage. Manual voltage lets us put a finite cap on how much power the CPU can use. This is what allows Prime95 to work more or less correctly. Upwards of around 1.285 or so is the max for my testing setup most of the time. More than that and the heat soak you see on the CPU ends up pushing temperatures into the 90c+ range and crashing the system in 30 minutes or less.
 
Top