Jump to content

mikeyten4

Members
  • Posts

    4
  • Joined

  • Last visited

Posts posted by mikeyten4

  1. Ok, thanks Fiery :)

     

    So, what I've really taken from that is that I can monitor my real-world usage via AIDA64 to see how it compares... but even if everything is fine right now, more applications may start to use things like AVX instructions in the future, so my FPU result might be a sign foretelling what I could see in the future via real-world use.  The CPU copes with it at my current clock, but in honesty I can't see it being very healthy.  I guess I'll just have to keep an eye on things over time and see what the future holds :)

     

    Thanks again!

  2. Hi everyone :)

     

    I'm just wondring if someone can explain to me a little bit about the FPU test in AIDA64.  Until now I've been running stress tests with CPU, FPU, and cache being tortured.  As I understand it this is fine, but running an FPU only SST will applying the peak temperature to a CPU in order to show that it's capable of running under such circumstances without throttling back.

     

    I've so far got a stable 4.2Ghz clock out of my 4770K and I'm going to move on (with a target of 4.4GhZ or 4.3Ghz).  At my 4.2Ghz clock the machine was stable for over 14 hours under AIDA64's CPU/FPU/Cache test and hit a peak temp of 75°c (on air).  It generally floated around 65°c-70° on all cores.

     

    Following this I've just carried out an FPU only test under the same spec.  It peaked at 90°c on core 3, with the lowest peak on any other core being 81°c.  It didn't throttle (I believe the Haswell chips do this at about 95°c, TJMax is 100°c), but I stopped it after 15 minutes.

     

    Now obviously the intention of a stress test is to put the machine under more or at least as much continuous pressure as real world use would.  Although my CPU didn't throttle, those FPU test temps are a bit higher than I'm really comfortable with.  I think I read somewhere that the FPU test uses AVX instructions, but I'm not sure if that's correct - truth is I don't know how it works to reach it's temperatures.  But what I really want to know is how representative of real-world use is it?  What sort of applications might push the CPU in the same way?  Many?  A certain set or applications in a particular professional field?  Are the results of this test alone something I should concern myself with if I've so far considered the same clock to be stable and safe under the full SST for over 14 hours?

     

    If it helps, I'm running Windows 7 x64 and will be using the PC for music production, meaning it's heaviest load in real-world use will likely be Ableton Live plus a good handful of VST instruments and effects (creating and processing sound in real time).  Native Instruments VSTs will feature fairly heavily.

     

    Cheers :)

    Mike

  3. Hi Fiery

     

    Thanks for the fast response :)  OK, great, well at least I have something to go on.  It was the not knowing that was frustrating, but if it's a sign of instability, then at least I know.  So if the SST thread crashed, the SST itself would still look like it's ongoing, with the SST window open and the time ticking away?

     

    The PC has just passed x42 again on 1.171 VCore with a x37 cache multiplier... it ran for over 14 hours, AIDA64 was applying pressure to the CPU as expected in all of that time until I pressed stop :)

     

    I'll have a go at the 4.3Ghz clock again and see what happens.  I hope I can hit at least that level and get my chip stable to be honest!  Is there any log that's produced by AIDA64 that would show me details like when the SST thread crashed?  Can I tell it to create one if it doesn't by default?  A log of the actual stability test would be useful.

     

    Thanks a lot

    Mike

  4. Hi everyone.

     

    I hope this is in the right place - I'm new to AIDA64 and in need of some help.

     

    The title pretty much says it all really, but here's the specifics.  I'm an experienced PC builder, but a novice OC'er with some minor experience on a previous build.  I've just built myself a new system and  I'm currently going through the OC routine of trying to find the sweet-spot between stability, performance, and safety.  My core components are as follows,

     

    Intel i7 4770K (Haswell, 4C/8T)

    Asus Z87-PLUS (running UEFI BIOS 1504, released Oct 2013)

    16gb (2x4gb) Corsair XMS3 DDR3 1600mhz memory (currently operating at stock 1333mhz, XMP profile not yet applied)

    Phanteks PH-TC14PE cooler

    Samsung EVO 840 120gb SSD (Windows 7 x64 installed and up to date)

    Seasonic 520w Fanless Platinum series PSU

     

    The spec was all chosen with a specific use in mind (music production), but also with the intention of taking a shot at OCing it.  After reading up on Haswell I grabbed the trial version of AIDA64 since there seem to be plenty of recommendations stating that it's more compatible than competitors like Prime95 (due to AVX dangers concerning voltage increases in those but not in AIDA64?).

     

    I'm using the 'system stability test' to stress the CPU, FPU, and cache, but not the memory just yet.  After each successful SST of 7 hours upwards I've then gone back into the BIOS and tweaked my settings.  My last such stable config was core multi x42, core voltage 1.171v, with the cache mutli at x37 and voltage on auto.  This idles around 25°c-30°c and hits a peak 77°c under load, and the test ran at full load, all 8 threads, for around 8 hours before I stopped it.

     

    Now.... this is where it get's weird. After this success I increased the multiplier on all CPU cores to x43 and upped the cache multiplier to x38.  With the voltages the same as previously I got a BSOD when I attempted to run the SST, so I then upped the CPU core voltage to 1.195v.  This time it I was able to launch the SST fine.  Again, it seemed the be running as normal, so I left it through the day.

     

    However, when I checked on it after 10 hours it was still ticking over... but there was absolutely no load on any core - 0%.  AIDA64 showed the 'time elapsed' as ongoing - the sst was still running, but with no load being applied!

     

    Both Windows task manager and CoreTemp grapher also confirmed the load as 0%.  AIDA64 showed the status of the test as 'Stability Test: Started', with the time that I launched it - there was no mentioend anywhere on-screen, or any other evidence, of a problem.  What makes it more frustrating is that I have no idea when the load dropped to 0%.  My GF was at home and checked on it periodically for me through the day, but only knew to look for signs that numbers were changing and graphs were moving!  I didn't even notice myself that the load had dropped to 0% until looking at the screen a few times after getting home.

     

    Following this I rebooted and then tried to run sst again at the same spec... this time I was watching it when after just 5 minutes the load dropped to 0% again.  Temps dropped accordingly, and as before... AIDA64's test continued without any notifications or obvious issue.  It, again, simply seemed to stop applying any pressure to the CPU.

     

    So... I guess what I'd like to know is is this a known bug?  A sign that my OC simply isn't stable at the level tested?  I'd also like to know if there's any sort of log or report I can pull out of the software that might tell me when within those 10 hours AIDA64 stopped applying load via the SST.  Will a log like this exist?

     

    I'm now retesting my 'stable' 4.2Ghz OC, since I'm no longer confident that the lengthy test it passed was actually lengthy at all.  So far it's hit 4hrs 32mins as I type, and AIDA is still applying 100% load to all cores.

     

    Any advice will be appreciated! :)

     

    Cheers

    Mike

×
×
  • Create New...