Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 21

Thread: ASRock Z87 Extreme6 - Core Speed Fluctuating




  1. #11
    parsec's Avatar
    parsec is offline Senior Member
    Join Date
    Jun 2012
    Location
    Third stone from the sun
    Posts
    4,824

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    Shiari, when you use AIDA64 for stress testing, have you ever run the Stress CPU and Stress FPU tests checked at the same time?

    Running Stress CPU only is not bad, I only hit 70C at most for the core temps. But CPU plus FPU tests together, and I'm throttling in 30 seconds, 100C core temps. I stop that test when that happens of course.

    Yes, the CPU and FPU combo test is difficult, but if you want to check your CPU cooling ability, that's the one to try. The FPU setting enables AVX instructions.

  2. #12
    Shiari is offline Member
    Join Date
    Jun 2013
    Location
    Nottingham
    Posts
    143

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    Yup, got all relevant options enabled:


  3. #13
    parsec's Avatar
    parsec is offline Senior Member
    Join Date
    Jun 2012
    Location
    Third stone from the sun
    Posts
    4,824

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    That is... wow... just... wow

    Either you have a golden CPU, or I have no clue what I'm doing, or both!

    The version of CPU-Z you're using does not show the correct core voltage, try to find version 1.64.2. Who knows what core temp is displaying.

    What OC settings in general are you using? The built in ones, or your own?

    Did you check the voltages in AIDA64? You must know that they aren't 0.936V. I've been using v3.00.2505 Beta, but there is a newer version.

    If your Noctua CPU cooler does that well, I must get one!

    You also must know that those results would shock most Haswell over clockers... even the delidders, or are you a member of that club?

    Those are great results, but I am just... stunned

  4. #14
    parsec's Avatar
    parsec is offline Senior Member
    Join Date
    Jun 2012
    Location
    Third stone from the sun
    Posts
    4,824

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    Ok, Ok, did they change AIDA64 or what? Yes, the 2514 build is very different than 2505, much easier on the CPU, much cooler core temps than 2505.

    I really wonder about that, what was the motivation for that? Inspired by someone... or was that corrected.

    Of course, if you're not running 2514, then I'm still shocked...

  5. #15
    Shiari is offline Member
    Join Date
    Jun 2013
    Location
    Nottingham
    Posts
    143

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    You're right about CPU-Z; I hadn't noticed because I don't worry about what the voltage is, I worry about what my temperatures are. I've started up 1.64.2 instead (could've used AIDA's too actually), and made some more screenshots. I can't say I've seen temperature differences between 2505 and 2515 (though all these screenshots were taken with 2515). Are you sure everything else was the same in your comparison?

    AIDA64 idle:



    AIDA64 stresstest:



    Also my settings in EUFI:

    UEFI OC settings #1:



    UEFI OC settings #2:



    UEFI OC settings #3:



    For OC settings though, rather than copying these (I guess they might serve as a start, but you may not be able to POST as they were tweaked for my CPU) I'd suggest importing my HWbot profile in XTU instead: Shiari @ HWBOT, or direct link to most recent XTU profile (which should be very similar to these UEFI screenshots, if not identical): http://hwbot.org/xtu/download/1281

    My CPU is certainly not the fastest around, some reach 4.7-4.8GHz without effort, while I could say I've reached 4.6 only just (so far at 4.7 temps just go through the roof and it started throttling, though I've not yet gone through the process of seeing what voltage I'd need minimally to get 4.7 stable ... it might only just be doable under stress testing). And then again, some have problems getting at 4.3-4.4. What's your CPU's batch number?

    As to the Noctua, yeah it is very good, best of what's doable on air cooling and about as good as a closed loop CPU cooler. There are other air coolers of similar quality, such as the Alpenföhn K2. It is expensive too though. And no, I have not delidded just yet. Might do one day when I want moar speed. :)

  6. #16
    parsec's Avatar
    parsec is offline Senior Member
    Join Date
    Jun 2012
    Location
    Third stone from the sun
    Posts
    4,824

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    Thanks, but the OC settings are not my problem, it's the core temps, although voltage settings matter of course. Using AIDA64 2505, with CPU and FPU tests (only) checked, at 4.4 I was seeing throttling after 30 seconds. I never considered checking all four (adding cache and memory) since I thought it would just be worse, temperature-wise.

    Using AIDA64 build 2514, with my usual CPU settings, with just CPU stress checked, it glided through the test at ~60C on the cores, at least 10C lower than build 2505. So I tried all four settings, and as I watched for throttling to occur, it never did. Is the CPU and FPU only test different than all four? Did not try that yet.

    My batch is L310B386, and is only a 4670K. How they keep the IHS to die interface consistent with the IHS glued in place is beyond me. Given the method that Noctua uses to apply pressure to the CPU, which allows pressure adjustment (correct?), it must work better than my Corsair H60, that simply reaches a limit due to its standoffs. Not that a H60 is a fantastic cooler, but I've never had throttling with my i7-2600K and i5-3570K CPUs with it.

    Surprised to hear no temperature differences with different versions of AIDA, I might be using different CPU voltage settings in my testing, but I don't think so. Time to start over...

  7. #17
    upstate_luke is offline Junior Member
    Join Date
    Jun 2013
    Location
    East Coast
    Posts
    6

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    I found some time to try a few things, I've disabled OC Fixed Mode again and have switched to Aida64 3.00.2500 for the stress test and I'm now seeing a constanct 4.2 with no throttling.

    I'm not sure why the core speed is staying steady when earlier it was fluctuating quite a bit.

    In terms of temps, I have never gone about high 50s/low 60s for any tests that I've run thus far (both with prime95 and Aida64.)

    Here's my batch number: L311B463

    I'm going to push the OC a little more to see if the throttling returns.

    Question for Shiari - Any idea why my core speed would have fluctuated under load earlier seeing as my temps never reached the 90-100 range? I'm not sure what to make of it.

  8. #18
    upstate_luke is offline Junior Member
    Join Date
    Jun 2013
    Location
    East Coast
    Posts
    6

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    I've determined the fluctuations in clock speed that I was seeing in my OP were due to using prime95 v27.9 build1. After running a stable 4.2 with Aida64 , I switched to prime95 with no other changes and the core speed fluctuations immediately returned (regularly floating between 4.0 - 4.2). The OC Fixed Mode setting I enabled while using prime95 resolved it but I'm not sure how.

    I'll continue to work on my OC using Aida64 and plan to leave the OC Fixed Mode setting disabled going forward.

    If anyone knows why prime95 and Aida64 would have different clock speed results (stable vs fluctuating), I'd be interested to learn more about it.


    Thanks again to all that responded with helpful insight, tips, suggestions and Emily again for the updated Bios.
    Last edited by upstate_luke; 06-23-2013 at 12:29 PM.

  9. #19
    Shiari is offline Member
    Join Date
    Jun 2013
    Location
    Nottingham
    Posts
    143

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    From what I've read about Haswell so far (though statistical data is only available in small numbers so far), L309 aren't the greatest of overclockers and may have trouble even hitting 4.4. L310's are a bit better than L309's, and L311's are again apparently a bit better than L310's. But whatever the batch number, you're still going to need to provide high end cooling to reach high clocks of course.

    My guess is that these differences might be related to small changes in the manufacturing process in these batches. The IHS sits over the silicon, with some thermal paste between them. The IHS is also glued to the circuit board, and the thickness of this glue is what creates a bit of distance between the silicon and IHS. It's likely that Intel has changed something in their process in later batches which has narrowed the distance ever so slightly (probably in the micrometer region), but that small distance can be enough to make noticeable temperature differences. Short of delidding and removing the glue there's really nothing else you can do about that, and the best of cooling can only do so much if there's poor thermal transfer inside the chip.

    @parsec:

    I've seen it mentioned that 4670K's need a bit more voltage applied than 4770K's ... I don't know why though. The temperatures should be more contained however due to the lack of Hyperthreading, just like it also makes a difference to disable HT on the 4770K.

    Not that a H60 is a fantastic cooler, but I've never had throttling with my i7-2600K and i5-3570K CPUs with it.
    Sandy Bridge CPU's have the die soldered to the IHS - trying to delid those would destroy the CPU. They have pretty good thermal transfer however, so there's no need for delidding.

    Ivy Bridge CPU's are just like Haswell ... some achieve great clocks, some struggle to overclock them. Ivy Bridge is where the delidding started, due to them stopping to solder the IHS to the die (possibly related to changes in the manufacturing process).

    The Noctua (like most coolers I've seen that I can remember) is essentially spring mounted. The screws are turned as far as they can go; springs on the screws that attach the heatsink to the mounting system apply the pressure. You really can't apply more force than that, and trying to do so might break your board too I guess. It likely would have to be a lot more force (well beyond the breaking point of any delicate parts like the circuit board of the CPU) to "flatten" the glue between IHS and circuit board of the CPU so that the IHS gets to sit closer on the die.

    @upstate_luke:

    I've just tested with the Prime95 build you mentioned which supports AVX, and I see the same behaviour: every now and then the core speed drops, while there's no throttling or excessive temperatures are involved. I don't know why that is, but I've not seen it anywhere but with Prime95 so far. I think it's entirely related to Prime95, and it doesn't concern me in the slightest. :)

    Also, if you're running Prime95 for pure stress testing, that really may not be the most suitable candidate. The AVX builds may have changed things a little, but builds before that simply do not stress enough of the Haswell archictecture to test stability all that well or generate a lot of heat. I still recommend AIDA64 and XTU for stress testing, and video encoding with Handbrake for a more realistic test. I haven't had a crash here since making sure AIDA64 and XTU pass their stress tests, though admittedly I've only done maybe two Handbrake movie encodings since. And of course this is my personal experience and advice, the method worked for me so I can only hope it works well for others too.

  10. #20
    parsec's Avatar
    parsec is offline Senior Member
    Join Date
    Jun 2012
    Location
    Third stone from the sun
    Posts
    4,824

    Default Re: ASRock Z87 Extreme6 - Core Speed Fluctuating

    Shiari, I currently have an i7-2600K and an i5-3570K, both in ASRock Z77 Extreme 4 boards. I'm quite aware of IB CPUs being the first generation (22nm, fin FETs) to not use solder between the die and IHS. Frankly, my favorite PC is the one with the i7-2600K, SB CPUs were a gift, nothing else like them overall from Intel or AMD. It runs so cool and to OC it is as easy as it gets. IB is more of a challenge due to the poor heat transfer from die to IHS from simply using TIM. The die - IHS gap is the other problem, but with solder the gap is a non-issue, and really an unknown factor.

    Why Intel stopped using solder between die and IHS is a mystery. My conjecture is the ROHS standard calling for lead-free solder, which has a significantly higher melting temp than standard solder, could not be used since the die would be damaged by those temperatures. The price of TIM is many times more that of solder of any kind, so I don't see a cost saving there. The manufacturing process using solder must be more expensive and difficult, which may be a motivating factor.

    I thought the Noctua coolers had spring loaded screws, but could not see any in some photos I checked.

    The Haswell on-die VRMs obviously are adding the additional heat, since they are hotter than IB CPUs. As I said, my Haswell CPU is the first one I've ever had throttle itself due to high temperatures. The rated TDP of 84Watts I find hard to believe, as my X58/i7-930 CPU system is 130Watts TDP, and at a ~1GHz OC, it's not as hot as Haswell. Of course, different benchmark tests with different instructions is comparing apples and oranges.

Page 2 of 3 FirstFirst 123 LastLast

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •