• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Prime 95 v28.5, Haswell CPU's and you...is it bad for the CPU's health?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
I have been seeing random mentions on the web about this and really didn't think much of it. The web is full of BS sometimes as we all know. So finally, I ran across a post from a knowledgeable and respected person (Raja @ ASUS), that seems to explain why we should not be using it to test Haswell CPU's. Here is the snippet...

Do we agree with this assesment? I know personally I have stress tested multiple haswell and one haswell E CPU with the latest P95 with no ill effects at both stock and heavily overclocked settings for hours.

Praz nailed it really. The newer versions of Prime load in a way that they are only safe to run at near stock settings. The server processors actually downclock when AVX2 is detected to retain their TDP rating. On the desktop we're free to play and the thing most people don't know is how much current these routines can generate. It can be lethal for a CPU to see that level of current for prolonged periods.

As for the universal validity of various stability testing programs, that's a more difficult question to answer without using illustrations to simplify what occurs at the electrical level on some of the associated buses.

Being brief as possible and focusing on DRAM transfer as an example: Data is moved around the system in high and low logic or signal states. The timing of these systems and those that rely on them needs to be matched closely enough for data to be moved around and interpreted correctly.A burst of data may contain a series of 1s and 0s. The 1s pull more current as they require defined voltage level that is above 0. Each data pattern has a different effect on the timing margin. Some eat into the timing margin more than others (I may illustrate the theory of this in a future guide). If a given stress test does not generate patterns in a way that eats into the timing budget sufficiently to represent how the system is used, the stress test won't be as useful to the end-user.

That's why most stress test programs alternate between different data pattern types. Depending on how effective the rotation is, and how well that pattern causes issues for the system timing margin, it will, or will not, catch potential for instability. So it's wise not to hang one's hat on a single test type. Evaluate what your needs are from the system and try to run a variety of tools to ensure the system is stable in various ways. We also need to bear in mind that some stress tests only focus on a single part of the system, while others will impact multiple areas at once.

Seasoned users usually find a systematic way that leads them from stress tests that focus on individual areas to those that hit the entire system as part of their test regimen. Ultimately, this all comes down to what your requirements are and using enough testing to confirm reasonable stability for the system in its intended usage scenario.

We coded Realbench to generate stress with real-world apps. It's a useful tool for people that encode, render or crunch numbers with their systems. However, it's not the only method out there - there are many tools to evaluate system stability that are perfectly valid.

-Raja
http://www.overclock.net/t/1510388/haswell-e-overclock-leaderboard-owners-club/2390#post_22900116
 
This stuff really drives me crazy. I'm running prime now on my new setup and the temps that prime puts onto my 5820k are insane. I mean minimal voltage(1.150) on a noctua cooler and in seconds it's at 80c-82c. I'm hoping prime is okay to run on it, but i'm also wondering in the back of my mind if it's actually bug free on my haswell. What bench can we use though? If not prime what else? Intel's? Realbench? Sounds like just clock it up and do your daily tasks and see if it crashes. Kind of sucks though I wish there was a definitive answer on this.

I've been reading prime is over volting haswell's as well. So, I'm not even sure if my 80+ temps are real or not. I do a lot of rendering work so, I wish there was more evidence to support prime for stability
 
XTU or earlier Prime95 than 28.5 have no AVX support.
I also don't think that running stability tests with AVX support will damage cpu if you keep reasonable voltages and temps.

Prime is not overvolting CPU. It may cause higher voltage if you set LLC wrong. Highest LLC level is usually giving higher voltage under full load than you set in BIOS.
 
XTU or earlier Prime95 than 28.5 have no AVX support.
I also don't think that running stability tests with AVX support will damage cpu if you keep reasonable voltages and temps.

Prime is not overvolting CPU. It may cause higher voltage if you set LLC wrong. Highest LLC level is usually giving higher voltage under full load than you set in BIOS.
I'm with this thinking... I have not seen it raise voltage over an older version personally.
 
I will always use prime 95 for short runs to pin down quick stability settings, as it will tell me in 10 minutes what may take 10 hours with realbench, aida64, etc. For example 1.24v and 4.7ghz will fail prime 28.5 in 5 minutes each time. I can run realbench for hours at that level without issue, but will crash 24/7 and have.

At 4.7ghz, to be stable for
1.29v for prime 28.5
1.27v for prime 27.9
1.22 to 1.24v for encoding, aida64, x264, realbench.

But real question is do I need 1.29v (prime 28.5 with fma3/avx) for 24/7 stability, or will 1.27v suffice prime 27.9 stable (no fma3, less avx use).

But at oc.net, no one like running prime, because it is too difficult to get haswell stable running prime at higher overclocks and no solder so temps get out of control. Prime does pull way more wattage than cpu is rated for though, up to 2x, but first two issues are why people dont like running it, not because of raja's explanation.

I know if I get prime stable and then bump vcore up 1 notch, I will never crash from vcore issue and never have, so will continue to use it myself. But I just stay out of prime threads at oc.net, and just run same programs as everyone else. But for my own testing, I will use prime, I have zero worry about prime hurting my cpu with few runs I do. I am not running a lesser program for 8 hours when prime will tell me in 5 minutes.

But do really need as much vcore as required for prime 28.5, to be 24/7 stable, I dont know, that is only question that would keep me from running prime. In the future I may stick with 27.9.
 
Last edited:
Would you guys recommend me running an old version of prime then?

prime 27.9 will have 10C less temps than 28.5, draw less current, still use partial avx (which you may run into unlike fma3), and is likely going to get you 24/7 stable. That would be a good place to start.
 
Prime does pull way more wattage than cpu is rated for though, up to 2x, but first two issues are why people dont like running it, not because of raja's explanation.

But do really need as much vcore as required for prime 28.5, to be 24/7 stable, I dont know, that is only question that would keep me from running prime. In the future I may stick with 27.9.

You're spot on. I am currently running version 27.9 prime95 and it's running exactly 10c cooler than 28.5 does. What in the hell is 28.5 doing to have such a different temp difference?? I mean I do know it's running new haswell instruction set, but that shouldn't cause the increase in temp unless i'm wrong. Only thing that it could be doing is pulling more wattage, but why does it do this? If it's pulling more wattage wouldn't it run hotter, but be more stable? I'm completely lost on why 28.5 is pulling more wattage?
 
It was mentioned already in this thread, the AVX/AVX2 instruction set is used in 28.5 whereas others do not seem to test that (post 3/6/7). :)
 
Hrmmm... So it looks that with 27.9 version of prime95 my chip is stable at 4.2ghz @ 1.150v... Not too bad. It was not stable with 28.5. Not sure what's going on there.

Saying this my chip at full load hits a high temp of 71c. I may have more headroom to up the volts and try a higher clock. 28.5 was running my chip so hot that I wouldn't dare increase the volts. I'm using hardware monitor. I'm hoping that temp difference between both versions is an error. I'm not sure if I should push my cpu anymore since i'm not sure what's going on between both versions.
 
I've noticed that new XTU is crashing/showing info about errors faster during stability test than its older versions while it's still not heating up CPU so much as Prime95. XTU base on older Prime95 version ( before 28.5 ).
 
I suspect this would apply more to people who use adaptive voltage. As i understand it, using adaptive voltage it can and will go above what you set when the cpu runs into avx2 tests. The question then becomes is prime95 running the cpu so hard just because of avx2, or is it becoming more like linpack and furmark where it is a fully synthetic way beyond anything reasonable designed to fry your cpu program.

I still use prime95 on my haswell, although i have been everything not avx2 stable at 4.6ghz 1.3v even running a batch of handbrake overnight, but those settings will instantly throttle in prime95.
 
Prime95 is using new instructions to perform faster calculations of prime numbers. This is mainly a project with option to run stability test so it's not only heating up soft like furmark.
 
Hrmmm... So it looks that with 27.9 version of prime95 my chip is stable at 4.2ghz @ 1.150v... Not too bad. It was not stable with 28.5. Not sure what's going on there.

Saying this my chip at full load hits a high temp of 71c. I may have more headroom to up the volts and try a higher clock. 28.5 was running my chip so hot that I wouldn't dare increase the volts. I'm using hardware monitor. I'm hoping that temp difference between both versions is an error. I'm not sure if I should push my cpu anymore since i'm not sure what's going on between both versions.

You should need more vcore to run 28.5 than 27.9, mine did as well. It is working cpu harder.

The temp differences are real. At 4.7ghz, 1.29v on mine, cpu only power consumption is 180W with prime 28.5, 157W with prime 27.9, 140W with XTU, x264, etc. So temp differences are real and reflect the wattage differences. Running fma3 and increased avx with 28.5 increases power consumption and hence temps.

@Supertrucker, isnt just avx and not just with adaptive, the on cpu fivr in haswell/dc has some built in LLC logic that cant be turned off. With increasing loads, the cpu vcore increase up to .02V with adaptive on or off... same increase above level set. Higher load, higher vcore. you wont see this increase using cpuz because cpuz on haswells and DC reads VID instead of vcore. You need to use HWMonitor to see actual vcore vs vid.

1.295v at idle (multimeter reading vcore off leads UD5H GB board)
1.303v with AIDA64 load CPU only
1.308v with AIDA64 CPU + FPU
1.308v with prime 25.1 blend and small ffts same
1.308v with prime 27.9 large ffts with avx
1.311v with prime 27.9 small ffts with avx
1.309v with prime 28.5 large ffts with avx/fma3
1.314v with prime 28.5 small ffts with avx/fma3

Many others measured same .02v increase on varying mobos as load increases, again with or without adaptive.
 
Higher load, higher vcore. you wont see this increase using cpuz because cpuz on haswells and DC reads VID instead of vcore. You need to use HWMonitor to see actual vcore vs vid.
The latest version seems to... at least on s2011 haswell-e it does...I got 1.1x for stock and 1.375 for 4.5Ghz. ;)
 
Thanks for the link. I am aware of that info. :)

What I took away from it is this:
On these CPUs, the monitoring of Vcore can be a bit tricky (because it's integrated inside the CPU), but some mainboards are capable of this (though I'm not sure how they really implement this and how accurate the particular values are).
I thought MSI and Asrock boards would do it on Haswell. I know both MSI and Asrock do this on X99 as well.
 
Bumping up an old thread to confirm that P95v285 blows up my CPU temps by a good 20 C more than XTU. I saw temps of 98C for my modest OC of [email protected]. Needless to say, I exited Prime after a minute or so.
 
so now that it's been a few years, what's the verdict? i got some instability despite being fine in realbench but prime95 blue screens almost immediately. in my situation prime95=real world stability but i'm afraid to run it for more than 5 minutes. now that it's been awhile are the prime95 peoples cpu's all dead or was it just mass hysteria? i think i'd rather run something dangerous for 5-10 min that works than something "safe" for half a day but maybe i'm playing with fire. i'm lowering my oc today to what i hope is truly stable.
 
Back