1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Is this alright? [i7 4770k @ 4.5Ghz 1.3v 90c during high Intelburntest]

Discussion in 'Overclocking & Cooling' started by CjStaal, Oct 30, 2013.

  1. CjStaal

    Joined:
    Dec 14, 2005
    Messages:
    1,767 (0.44/day)
    Thanks Received:
    91
    Location:
    Calverton, New York
    Hey all. I just built a new rig and custom water cooling loop. I'm overclocking my 4770k but when I try anything other than intelburntest it stays at or below 72c. But when I try intelburntest it goes in to the low 90's on high and very high. Should I be worried? Why am I pulling some much heat when I have a 650 dollar custom loop setup? 45x100 1.3v vcore, 42x100 1.17 vring. 1.8v PLL or whatever that is, core input voltage or something? Any help would be appreciated
    Also, AI SUITE 3 shows my temp getting to 77, not 90. Untitled.jpg
     
    10 Year Member at TPU
  2. sneekypeet

    sneekypeet Unpaid Babysitter Staff Member

    Joined:
    Apr 12, 2006
    Messages:
    23,355 (6.00/day)
    Thanks Received:
    7,913
    IBT uses a new AVX instruction set that takes the CPU to a whole new level of heat! I would suggest not to run the proc at 90 for very long and move to something like Prime95 or the AIDA64 stability testing.

    Also go with core temp...AI shows the socket temp from what I saw.
     
    10 Year Member at TPU
  3. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    15,771 (4.05/day)
    Thanks Received:
    9,581
    Location:
    Beaumont, Alberta
    Because you have a $650 water-cooling loop on an Intel APU.

    Go by realtemp or try AIDA64 for temps. AIDA64 offers the most information, that I've seen.
     
    Vario says thanks.
    10 Year Member at TPU
  4. steponz

    Joined:
    Nov 5, 2013
    Messages:
    32 (0.03/day)
    Thanks Received:
    11
    How did you choose 1.3 vcore? Have you tested lower?

    Some chips need higher, some lower.. i would test.. This would bring temps down.

    4770K's run hot.. i lapped and delidded a couple and that helped tremendously. . The best method is trying to reduce voltages as much as you can..
     
  5. DarkStalker

    DarkStalker

    Joined:
    Nov 30, 2007
    Messages:
    77 (0.02/day)
    Thanks Received:
    15
    Location:
    Croatia
    hi guys! i also have built a new and very similar rig(see sys specs) witH aio coolermaster prestige 240l water colling solution.
    i currently oced 4770k to 4.5ghz usin 1.25v vcore, 1.15 cpu ring voltage, vrin to 1.8v, uncore multi x44 and all is fine but...temps are fine when gaming hard max at 65c but when i clock to 4.6(1.3vVcore) i can get to desktop and finish aida with very high max temps at 93c!!? and running 3dmark11 cpu physics test it freezes.
    DDR ram voltage is deffo 1.65v xmp profile 1 for my 2x8gb kit kingston beast 2400mhz default freq and latencies.
    TRYED HUNDREDS COMBINATIONS but no go for stable 4.6 ghz its tiring but ill figure it...
    my last cpu was core 2 quad q9550 and that was a heaven for ocing from 2.8ghz to 3.8 on air WITH MAX TEMPS 65c..good old days
    any advice would be great
     
    Last edited: Dec 12, 2013
  6. Live OR Die

    Live OR Die

    Joined:
    May 19, 2007
    Messages:
    4,201 (1.20/day)
    Thanks Received:
    475
    Probably just poor contact in between the DIE and IHS so heat transfer wont be very good so your custom loop wont be able to cool the DIE correctly. if you do use AIDA64 and just do the FPU test you will even see higher temps as IBT is more of a bench test not a stress test.
     
  7. TheHunter

    TheHunter

    Joined:
    Apr 10, 2012
    Messages:
    1,332 (0.78/day)
    Thanks Received:
    639
    Location:
    78°55' N, 11°56' E
    Did you disable VR fault management?

    Also try lower cache freq, maybe 41x @ 1.13v, LLC6, cpu current 120%. And do you really need 1.30v for 4.5Ghz?
     
  8. vega22

    vega22

    Joined:
    Jul 18, 2007
    Messages:
    2,303 (0.67/day)
    Thanks Received:
    601
    i need over 1.3vcore for 4.5ghz.

    but i also run much cooler as mine is delidded so max i see at 4.5ghz is low 60s.
     
  9. kn00tcn

    kn00tcn

    Joined:
    Feb 9, 2009
    Messages:
    1,184 (0.41/day)
    Thanks Received:
    280
    Location:
    Toronto
    it's called burnintest for a reason, but i didnt know AVX was added... is there a way to turn that off? i used it for my OC stability testing, 5min crash meant 2hr crash in TF2 in my case, so i found it much faster than doing some 24h prime test
     
  10. Octopuss

    Octopuss

    Joined:
    Aug 13, 2009
    Messages:
    1,768 (0.66/day)
    Thanks Received:
    507
    Location:
    Czech republic
    90°C is too much even for IBT.
    1.3V is way over the line imo for this frequency. I know chips are different one from another, but this is just too much.
    For example mine runs at tiny bit over 1.1V at 4.4GHz.
     
  11. vega22

    vega22

    Joined:
    Jul 18, 2007
    Messages:
    2,303 (0.67/day)
    Thanks Received:
    601
    congratz, you have a good chip.

    some of us are not so lucky.

    needs 1.38vcore for 100% stability.
     

    Attached Files:

  12. Ed_1

    Joined:
    Dec 14, 2006
    Messages:
    359 (0.10/day)
    Thanks Received:
    51
    OCCT has AVX option on its linpack test , with LinX and IBT I think you have to use older version with no AVX support .
     
  13. Octopuss

    Octopuss

    Joined:
    Aug 13, 2009
    Messages:
    1,768 (0.66/day)
    Thanks Received:
    507
    Location:
    Czech republic
    Or you can use Prime95, which supports all these latest instructions for a long time.
     
  14. MxPhenom 216

    MxPhenom 216 Corsair Fanboy

    Joined:
    Aug 31, 2010
    Messages:
    11,838 (5.16/day)
    Thanks Received:
    3,566
    Location:
    Seattle, WA
    No it hasn't. At least not AVX2 which is the new instruction on Haswell. Also Prime95 takes way to long to get any usable information out of it in terms of system stability.

    OP do like Cadaveca said, use AIDA for testing Haswell. He easily has more Knowledge on Haswell then anyone else who has replied in this thread.
     
  15. Live OR Die

    Live OR Die

    Joined:
    May 19, 2007
    Messages:
    4,201 (1.20/day)
    Thanks Received:
    475
    Or use Asus real bench that what i normally do and set the stress test on loop leave it going for a few hours.
     
  16. Jetster

    Jetster

    Joined:
    Jan 17, 2010
    Messages:
    8,638 (3.43/day)
    Thanks Received:
    4,901
    Location:
    Oregon
    Intel Burn test gets too hot. Dont use it
     
  17. Ed_1

    Joined:
    Dec 14, 2006
    Messages:
    359 (0.10/day)
    Thanks Received:
    51
    The problem with Haswell is if you do test with AVX the voltage goes up more than in the i5-35xx/i7-37xx so heat goes way up . Even with Ivy they get pretty hot with AVX enabled .

    Only issue with AIDA , is seems it can pass stable in that but not stable in other apps , which is always a possibility even testing in Prime95/Linpack .

    [H]OCP ran into this when first testing haswell , in one of there reviews .
     
  18. Live OR Die

    Live OR Die

    Joined:
    May 19, 2007
    Messages:
    4,201 (1.20/day)
    Thanks Received:
    475
    Just use real bench and be done with it its all you need.
     
  19. Octopuss

    Octopuss

    Joined:
    Aug 13, 2009
    Messages:
    1,768 (0.66/day)
    Thanks Received:
    507
    Location:
    Czech republic
    You're right about Haswell. That's currently in some test builds only. Normal AVX is supported just fine though.
     
  20. TheHunter

    TheHunter

    Joined:
    Apr 10, 2012
    Messages:
    1,332 (0.78/day)
    Thanks Received:
    639
    Location:
    78°55' N, 11°56' E
    That only happens if you use auto voltage which uses offset or adaptive voltage.

    By fixed 1.30v it will be fixed 1.30 even by those avx2 apps, btw aida64 raises the voltage as well, not just by fpu.:)
     
  21. Ed_1

    Joined:
    Dec 14, 2006
    Messages:
    359 (0.10/day)
    Thanks Received:
    51
    Sure , but many don't use fixed (especially if its stock) and most want offset for 24/7 use , but yes fixed will of course give fixed voltage all time no matter load :D

    Here link from HOCP on AIDA with stability . (read down last part of page)

    http://www.hardocp.com/article/2013...74770k_ipc_overclocking_review/7#.Uq8t605QalY
     
    Last edited: Dec 16, 2013
  22. TheHunter

    TheHunter

    Joined:
    Apr 10, 2012
    Messages:
    1,332 (0.78/day)
    Thanks Received:
    639
    Location:
    78°55' N, 11°56' E
    I mean if you test such apps then test with fixed voltage, otherwise yes use adaptive and dont run such tests since it spikes +0.09v.

    x264 encoders have occasional spikes as well, but it wont heat that bad..


    Im also at adaptive, i think its the best setting ever, atm in idle :)
    [​IMG]
     
    Last edited: Dec 16, 2013
    MxPhenom 216 says thanks.
  23. Vario

    Vario

    Joined:
    Oct 21, 2005
    Messages:
    3,394 (0.83/day)
    Thanks Received:
    1,288
    When stability testing a new core clock with Aida, stock ram, should I run memory testing or just cpu? After removing my ram overclock I am able to hit new levels of multiplier, currently exploring voltages for 46-48x... It seems to need between 1.25 and 1.35 core voltage for this range. I don't want to kill my ram testing the chip etc.

    Also aida crashed last night, but kept running stability tests somehow because the cores were fully loaded and hot until I rebooted this morning when I woke up. I might stick to XTU and Prime95.
     
    10 Year Member at TPU
  24. Live OR Die

    Live OR Die

    Joined:
    May 19, 2007
    Messages:
    4,201 (1.20/day)
    Thanks Received:
    475
    Whats the ambient temp for my chip to run that cool its needs to be below 25c.
     
  25. TheHunter

    TheHunter

    Joined:
    Apr 10, 2012
    Messages:
    1,332 (0.78/day)
    Thanks Received:
    639
    Location:
    78°55' N, 11°56' E
    idk 19-22C? Its winter here (atm -2C), I have a small central heating radiator in the room but its not opened to the max..


    lol one time i had balcony door open for 1-2 min was gonna go on smoke outside and temps dropped to 10-13C across all 4 cores



    EDIT: but there is one thing, if i disable c-states it will be in 30-34C area, 0.7v 800mhz vs 23-27C..
     
    Last edited: Dec 17, 2013

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)