Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]
New Posts  All Forums:Forum Nav:

Haswell Overclocking Guide [With Statistics] - Page 133

post #1321 of 19539
I basically follow a similar stress testing methodology as you for my own personal validation. I run typically stress test for 18-24 hours before i consider anything stable.

I think 8 hours of stress testing is an absolute minimum . if your stable for 8 hours typically you are gonna be pretty stable overall, thats my opinion and what i recommend in my OC guide.

overclocking haswell is time consuming. after 45 days im still tweaking!

i wouldn't discredit p95 in anyway. use whatever stress testing methodology works best for you. i have used p95 for 13 years of overclocking and was happy to put it down.

what is the "best" stress testing program for haswell as of right now? from a universal standpoint. simple answer intel extreme tuning utility.

why?

it is the ONLY stress test that can cope with adaptive voltage and stresses the cpu 100%. im not saying its the only way to find stability with haswell, im just saying from a universal standpoint it is the one to use and recommend. when your making recommendations to noobie overclockers i think its best to simplify as many variables as possible.

thats how i feel
post #1322 of 19539
Well, just to be devil's advocate, if it stresses the CPU 100% but doesn't cause the adaptive voltage to go up, then it may not be stressing the CPU much at all. Quite a few people have mentioned that Crysis 3 causes adaptive voltages to go up (above the normal limit), so that would imply (in one way of looking at it) that Crysis 3 is harder the XTU. Assuming we use the adaptive increase as evidence of significant or difficult CPU load, that is.
post #1323 of 19539
Oh I agree with what you're saying. I just think there seems to be a large proportion of people not really stress testing for very long at all and then wondering why they have stability problems, and I'm not referring solely to this thread or this site.

I just think it's important for people to understand that putting the hard yards in to begin with by stability testing you will and up with a much more stable system.

Sure your temps will be higher stability testing compared to normal use for the most part, and maybe that's what puts people off using Prime etc, but atleast you'll know your system should be rock solid.

Don't get me wrong, I'm all for pushing for the highest clocks you can get and I've played around with 4.8Ghz on my 4770k too and had my old Sandy at 4.9Ghz 24/7 stable, but I'd rather be rock solid stable at 4.5Ghz than have some instability at 4.8Ghz and BSOD during a session of BF3 for example.

That being said I'm planning to push my 4770K further. Hoping to get to 4.7Ghz stable without delid if possible, but happily running with 4.5Ghz for the time being.
post #1324 of 19539
Quote:
Originally Posted by ProKoN View Post

I basically follow a similar stress testing methodology as you for my own personal validation. I run typically stress test for 18-24 hours before i consider anything stable.

I think 8 hours of stress testing is an absolute minimum . if your stable for 8 hours typically you are gonna be pretty stable overall, thats my opinion and what i recommend in my OC guide.

overclocking haswell is time consuming. after 45 days im still tweaking!

i wouldn't discredit p95 in anyway. use whatever stress testing methodology works best for you. i have used p95 for 13 years of overclocking and was happy to put it down.

what is the "best" stress testing program for haswell as of right now? from a universal standpoint. simple answer intel extreme tuning utility.

why?

it is the ONLY stress test that can cope with adaptive voltage and stresses the cpu 100%. im not saying its the only way to find stability with haswell, im just saying from a universal standpoint it is the one to use and recommend. when your making recommendations to noobie overclockers i think its best to simplify as many variables as possible.

thats how i feel
but the funny thing is, I got a random reboot (seemingly like a dip in VCCIN) while doing H.264 after passing XTU for 13hrs. frown.gif
and temps in Handbrake (2nd pass at least) are identical to XTU. Could this mean XTU is not stressful enough?
Quote:
Originally Posted by Forceman View Post

Well, just to be devil's advocate, if it stresses the CPU 100% but doesn't cause the adaptive voltage to go up, then it may not be stressing the CPU much at all. Quite a few people have mentioned that Crysis 3 causes adaptive voltages to go up (above the normal limit), so that would imply (in one way of looking at it) that Crysis 3 is harder the XTU. Assuming we use the adaptive increase as evidence of significant or difficult CPU load, that is.
i still don't believe it has anything to do with the CPU load alone. it is CPU load + use of AVX instructions. for example, Prime 95 26.6 (AVX non-supported version) doesn't cause Vcore to go up by 0.1V.
post #1325 of 19539
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
post #1326 of 19539
Quote:
Originally Posted by Forceman View Post

Well, just to be devil's advocate, if it stresses the CPU 100% but doesn't cause the adaptive voltage to go up, then it may not be stressing the CPU much at all. Quite a few people have mentioned that Crysis 3 causes adaptive voltages to go up (above the normal limit), so that would imply (in one way of looking at it) that Crysis 3 is harder the XTU. Assuming we use the adaptive increase as evidence of significant or difficult CPU load, that is.

I don't have Crysis 3 at the moment but I'll see if I can play around with it and report back if I had any voltage spikes.

I agree with what Anusha has said though, it would have something to do with the AVX instruction set but not the actual CPU load. It's for this same reason that we're recommended not to stress with adaptive voltage.
Quote:
Originally Posted by Anusha View Post

but the funny thing is, I got a random reboot (seemingly like a dip in VCCIN) while doing H.264 after passing XTU for 13hrs. frown.gif
and temps in Handbrake (2nd pass at least) are identical to XTU. Could this mean XTU is not stressful enough?
i still don't believe it has anything to do with the CPU load alone. it is CPU load + use of AVX instructions. for example, Prime 95 26.6 (AVX non-supported version) doesn't cause Vcore to go up by 0.1V.

Are you using XTU Stability test or the benchmark?

The stability test isn't all that stressfu compared to the benchmark. I had the same problem before where I would pass Intel XTU stability test or Aida 64 stability test all day, but Prime95 blend would crash.
post #1327 of 19539
Quote:
Originally Posted by combatant3219 View Post

Are you using XTU Stability test or the benchmark?

The stability test isn't all that stressfu compared to the benchmark. I had the same problem before where I would pass Intel XTU stability test or Aida 64 stability test all day, but Prime95 blend would crash.
I was running the stability test. Can you loop the benchmark over and over?

Btw, into the 7.5hr mark with H.264. Still going. No sudden non-BSOD reboots. This is with VCCIN at 1.77V in bios and 1.808V with the LLC kicked in. I have like 5 more hrs left in the queue. It better be stable after that!
Edited by Anusha - 8/28/13 at 11:53pm
post #1328 of 19539
Quote:
Originally Posted by Anusha View Post

I was running the stability test. Can you loop the benchmark over and over?

Btw, into the 7.5hr mark with H.264. Still going. No sudden non-BSOD reboots. This is with VCCIN at 1.77V in bios and 1.808V with the LLC kicked in. I have like 5 more hrs left in the queue. It better be stable after that!

Not sure to be honest, I just know it stresses the CPU more but haven't played around with making it loop etc.
post #1329 of 19539
Quote:
Originally Posted by Anusha View Post

I was running the stability test. Can you loop the benchmark over and over?

No, it only runs once. And according to the help file, it is just running Prime95:
Quote:
Benchmarking allows you to gauge the performance of your processor. Intel® Extreme Tuning Utility uses Prime95 as the benchmarking engine. Since Prime95 returns a time as a result, Intel® Extreme Tuning Utility uses a simple formula to convert the time into a score. The higher the score, the better your system is performing.
post #1330 of 19539
Quote:
Originally Posted by Forceman View Post

No, it only runs once. And according to the help file, it is just running Prime95:
Seriously? That's nuts! I mean, they say Prime95 is not certified for Haswell but that's what they are using in the benchmark? Maybe that's what they are using in the stress test too! (A custom prime test; maybe without AVX!!!)
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]