EVGA

Hot!Low GFlops in Linpack

Page: 123 > Showing page 1 of 3
Author
kelkel1
Superclocked Member
  • Total Posts : 156
  • Reward points : 0
  • Joined: 2019/05/18 09:52:47
  • Status: offline
  • Ribbons : 0
2019/10/19 12:54:02 (permalink)
Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
 
The best I have done is low 300GFlops at 5200 MHz.
 
Any advice?

Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
#1

68 Replies Related Threads

    HeavyHemi
    Omnipotent Enthusiast
    • Total Posts : 14027
    • Reward points : 0
    • Joined: 2008/11/28 20:31:42
    • Location: Western Washington
    • Status: offline
    • Ribbons : 96
    Re: Low GFlops in Linpack 2019/10/20 14:52:04 (permalink)
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.

    EVGA X99 FTWK / i7 6850K 4.5ghz / GTX1080 Ti FE Hybrid / 32GB Corsair LPX 3200mhz / Samsung 850Pro 256GB / Corsair AX1200 / Window 10 Pro
    Fire Strike                        24,163
    Fire  Strike Extreme        14,452 
    Fire Strike Ultra                7,711 
    Time Spy                        10,357

    #2
    Cool GTX
    EVGA Forum Moderator
    • Total Posts : 18113
    • Reward points : 0
    • Joined: 2010/12/12 14:22:25
    • Location: Folding for the Greater Good
    • Status: online
    • Ribbons : 114
    Re: Low GFlops in Linpack 2019/10/20 21:48:18 (permalink)
    unstable OC or RAM ?
     
    background activity ?
     
    Clean Boot

    RTX Project EVGA X99 FTWK-5930K,EVGA2080Ti EVGA-1200P2  Nibbler EVGA X99 3-GPU Water Loop 1600P2  AIO Folding  1080Ti FTW3 Hybrid, 1200P
    Thank You for Your Support


    #3
    qswaaa
    New Member
    • Total Posts : 7
    • Reward points : 0
    • Joined: 2019/04/14 06:20:32
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/21 03:08:21 (permalink)
    make Ram 4500 cl17
    #4
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/21 14:36:25 (permalink)
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #5
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/21 14:40:57 (permalink)
    Cool GTX
    unstable OC or RAM ?

     
    RAM passed 1300% HCI memtest.
    OC passes LinX with no errors and identical residuals.
     
    Cool GTX
    background activity ?

     
    Possibly, but I tried turning off everything I could.
     
    Cool GTX
    Clean Boot



    Have not tried that yet.
     
    OK, clean boot made no difference.
     
    This is strange, because I get pretty good results in other benchmarks like Cinebench and RealBench.
    post edited by kelkel1 - 2019/10/21 15:16:29

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #6
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/21 14:42:13 (permalink)
    qswaaa
    make Ram 4500 cl17



    Not sure it can run that tight, it is 19-19-19-39.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #7
    HeavyHemi
    Omnipotent Enthusiast
    • Total Posts : 14027
    • Reward points : 0
    • Joined: 2008/11/28 20:31:42
    • Location: Western Washington
    • Status: offline
    • Ribbons : 96
    Re: Low GFlops in Linpack 2019/10/21 18:58:28 (permalink)
    kelkel1
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.




    Then you're probably not using the same linpack libraries. They make a large difference in score. Even the same version number of AgentGodIBT for example will have different libraries based on when it was uploaded. I would not worry about it if every other test is in the ballbark. Stuff doesn't fail in a manner that it would only have a defect running one particular binary.
     
    Edit, had a senior moment... what are your temps under LinX? It is entirely possible to be throttling for temps and pass the bench.  I would not be surprised unless you're under a custom water loop, you're well over 90C at those clocks.
    post edited by HeavyHemi - 2019/10/21 23:02:39

    EVGA X99 FTWK / i7 6850K 4.5ghz / GTX1080 Ti FE Hybrid / 32GB Corsair LPX 3200mhz / Samsung 850Pro 256GB / Corsair AX1200 / Window 10 Pro
    Fire Strike                        24,163
    Fire  Strike Extreme        14,452 
    Fire Strike Ultra                7,711 
    Time Spy                        10,357

    #8
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/23 10:13:37 (permalink)
    kelkel1
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.




    Please use LinX 0.9.5 or 0.9.6.  Not 0.6.5.
    If you were the same person posting on OCN, that was 0.6.5 which doesn't even support AVX2, much less any modern processor correctly !
    #9
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/23 13:50:57 (permalink)
    Kylearan
    kelkel1
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.




    Please use LinX 0.9.5 or 0.9.6.  Not 0.6.5.
    If you were the same person posting on OCN, that was 0.6.5 which doesn't even support AVX2, much less any modern processor correctly !




    Using LinX 0.9.5 or Linpack Extreme 1.1.1.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #10
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/23 14:58:22 (permalink)
    kelkel1
    Kylearan
    kelkel1
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.




    Please use LinX 0.9.5 or 0.9.6.  Not 0.6.5.
    If you were the same person posting on OCN, that was 0.6.5 which doesn't even support AVX2, much less any modern processor correctly !




    Using LinX 0.9.5 or Linpack Extreme 1.1.1.



    What cpu voltage (and vdroop) do you have set in your BIOS?
    How much RAM is installed?
    Did you use 35000 sample size?
     
    I think you're being throttled.  I can't even pass LinX at 5 ghz on mine!
      
    Attempting to do so, with auto vcore and intel default (max) vdroop and cpu load voltage @ 1.225v measured on the die is unstable and the test crashes (if it even completes one loop the residuals are totally wrong anyway.  Before it crashes, amps is over 200 amps(!) (max amps specification for 9900k is 193) and temps are 105C.
     
    I can pass 0.9.5 with same residuals (5 loops) at 4.9 ghz at the same settings, and over 100C core temps, but 1.1.1 gives residual errors (mismatch).  Of course 1.1.1 gets higher gflops also (like 520) vs 500.
     
    Are you delidded?  If not there's no way you're passing that with reasonable temps.
     
    What happens if you use a smaller sample size, like 10,000 ?
    #11
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/24 13:34:56 (permalink)
    Kylearan
    kelkel1
    Kylearan
    kelkel1
    HeavyHemi
    kelkel1
    Looking around the internet, I see others with 9900K scoring well over 500GFlops in Linpack at 5000MHz.
     
    The best I have done is low 300GFlops at 5200 MHz.
     
    Any advice?


    Are you using the exact same linpack libraries they are and are you using an AVX offset in the BIOS? Both of those will significantly impact the score.




    Not sure about the binaries, but same linpack version.
     
    My AVX offset is 0.




    Please use LinX 0.9.5 or 0.9.6.  Not 0.6.5.
    If you were the same person posting on OCN, that was 0.6.5 which doesn't even support AVX2, much less any modern processor correctly !




    Using LinX 0.9.5 or Linpack Extreme 1.1.1.



    What cpu voltage (and vdroop) do you have set in your BIOS?
    How much RAM is installed?
    Did you use 35000 sample size?
     
    I think you're being throttled.  I can't even pass LinX at 5 ghz on mine!
      
    Attempting to do so, with auto vcore and intel default (max) vdroop and cpu load voltage @ 1.225v measured on the die is unstable and the test crashes (if it even completes one loop the residuals are totally wrong anyway.  Before it crashes, amps is over 200 amps(!) (max amps specification for 9900k is 193) and temps are 105C.
     
    I can pass 0.9.5 with same residuals (5 loops) at 4.9 ghz at the same settings, and over 100C core temps, but 1.1.1 gives residual errors (mismatch).  Of course 1.1.1 gets higher gflops also (like 520) vs 500.
     
    Are you delidded?  If not there's no way you're passing that with reasonable temps.
     
    What happens if you use a smaller sample size, like 10,000 ?





    Various VCore, including Auto.
    16GB RAM.
    In Linx, it was 40K something sample problem size.
    Linpack Extreme 1.1.1 only gives ~170GFlops.
    Delidded, direct die. At 5200 MHz, CPU Package hits 78°C with 28°C ambient.
    Have not tried smaller sample size.
    post edited by kelkel1 - 2019/10/25 13:05:10

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #12
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/24 23:55:51 (permalink)
    kelkel1
     
     
    Various VCore, including Auto.
    16GB RAM.
    In Linx, it was 40K something sample size.
    Linpack Extreme 1.1.1 only gives ~170GFlops.
    Delidded, direct die. At 5200 MHz, CPU Package hits 78°C with 28°C ambient.
    Have not tried smaller sample size.




    Try running it at 4.5 ghz and 4.2 ghz cache.
    Test for gflops there.
    I am 100% sure you are being throttled.
    But I don't have a dark, so i have no idea why.
    The reason you get much higher Gflops with HT disabled in 0.9.5 and 1.1.1 is because it draws less current.
     
    There was one person with some Z390 Gigabyte board who had the same problem.  
    #13
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/25 13:10:16 (permalink)
    Just ran at 4500MHz and 4200MHz, as you suggested. 299 GFlops.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #14
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/25 15:39:00 (permalink)
    kelkel1
    Just ran at 4500MHz and 4200MHz, as you suggested. 299 GFlops.



    Yeah something's really wrong.
    At 4.7 ghz (4.4 ghz cache) I get 480-487 gflops in LinX 0.9.5 and 500 in Linpack Extreme 1.1.1 (this needs more vcore to remain stable than LinX 0.9.5)
    9900K with HT enabled.  I'm on a Z390 Aorus master.
    Note: AVX was NOT used in the old legacy versions (like the Agentg0d version 0.6.4, but I think there was a custom AVX version released for 0.6.4).
    There was no talk about people getting "Low" gflops like you, on 0.9.5 (if you translate to english in chrome): 
    Just people having problems getting same residuals, when they got same residuals in 0.9.3 or legacy (legacy=no AVX).
     
    There has to be a setting in your BIOS which is throttling you.  Or something in windows.  I honestly don't know what is wrong.
     
    I found a link with someone with a Z390 dark and 9900k.  A korean user got proper gflops here.
     
     
     
    Anyway I think you need to check your BIOS settings.
     
    *Edit* i can't post links for some reason.
    try (sorry I can't give you the link!! it wont let me.  you're on your own here.
    Just google LinX 0.9.6 and read the comments).
     
    This is the user with the Dark and 9900k that got full gflops.
    The incorrect residual description was on the Korean comments page for LinX 0.9.6 (just google it).
     
    post edited by Kylearan - 2019/10/25 15:42:14
    #15
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/25 17:21:23 (permalink)
    Kylearan
    kelkel1
    Just ran at 4500MHz and 4200MHz, as you suggested. 299 GFlops.



    Yeah something's really wrong.
    At 4.7 ghz (4.4 ghz cache) I get 480-487 gflops in LinX 0.9.5 and 500 in Linpack Extreme 1.1.1 (this needs more vcore to remain stable than LinX 0.9.5)
    9900K with HT enabled.  I'm on a Z390 Aorus master.
    Note: AVX was NOT used in the old legacy versions (like the Agentg0d version 0.6.4, but I think there was a custom AVX version released for 0.6.4).
    There was no talk about people getting "Low" gflops like you, on 0.9.5 (if you translate to english in chrome): 
    Just people having problems getting same residuals, when they got same residuals in 0.9.3 or legacy (legacy=no AVX).
     
    There has to be a setting in your BIOS which is throttling you.  Or something in windows.  I honestly don't know what is wrong.
     
    I found a link with someone with a Z390 dark and 9900k.  A korean user got proper gflops here.
     
     
     
    Anyway I think you need to check your BIOS settings.
     
    *Edit* i can't post links for some reason.
    try (sorry I can't give you the link!! it wont let me.  you're on your own here.
    Just google LinX 0.9.6 and read the comment
     
    This is the user with the Dark and 9900k that got full gflops.
    The incorrect residual description was on the Korean comments page for LinX 0.9.6 (just google it).
     




    Only thing I find regarding LinX 0.9.6 is in Korean(I believe), but since I cannot read Korean (or whatever language it is) that does not help.
     
    You could always just type out the link...
     
    Does Windows version have any effect? I am using W7.
    post edited by kelkel1 - 2019/10/25 17:24:10

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #16
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/25 17:22:49 (permalink)
    Anyone from evga care to comment on the possibility of a BIOS setting?

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #17
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/25 18:03:59 (permalink)
    kelkel1
    Anyone from evga care to comment on the possibility of a BIOS setting?




    It's here.
    And chrome can instantly translate Korean to english.  That's what I used.
    https://hwtips.tistory.com/1611
     
    http://cooln.kr/bbs/overclock/704572
    The screenshot is there. you can see he is using a Dark also, but he has 460 gflops and you have 300.  Therefore your bios settings must be wrong.
    That's still low (it should be 480) but its a lot higher than yours :).  I'm not sure if he's using an AVX offset or if he's using the turbo ratios separately for 1 to 8 cores (4700 mhz on 8 cores).

    Direct picture is here:
    http://cooln.kr/view_image?fn=http%3A%2F%2F112.172.165.208%2Fdata%2Feditor%2F1907%2FBimg_23c7fd3be87ed3f149f070d3ff301007_v4f4.png\
     
    translation from the second page:

    [<font][<font]cpu 9900k r0 stepping
    [<font]Board evga z390 dark 
    [<font]Ram gskill trident royal 3200/14
     
    [<font]We are testing the yield while over assembly after today.
    [<font]I'm new to evga board, so I can't adapt to BIOS.
     
    [<font]It's not a difference but the jeep value is a little lower than others.
     
    [<font]Over value is approximately
    [<font]CPU 50x / 45 Cache Override Voltage 1.2v / v Drop -50 / avx 3 / sa 1.15v / io 1.15v
    [<font]We tightened the tFAW to 16 on the RAM 4000 / 17-19-19-38-560-2t.
     
    [<font]What should we do beyond the 500 value of the jeep.
    [<font]

    post edited by Kylearan - 2019/10/25 18:11:51
    #18
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/25 18:42:07 (permalink)
    Thanks.
     
    That VCore is crazy low.
     
    I need 1.33 VCore to run LinX 0.9.5 at 5000MHz.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #19
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/25 19:25:17 (permalink)
    kelkel1
    Thanks.
     
    That VCore is crazy low.
     
    I need 1.33 VCore to run LinX 0.9.5 at 5000MHz.




    Remember he's using an AVX offset of -3.
    LinX is running at 4.7 ghz on his system.  Then 5 ghz when its finished.
    #20
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/26 06:45:16 (permalink)
    Kylearan
    kelkel1
    Thanks.
     
    That VCore is crazy low.
     
    I need 1.33 VCore to run LinX 0.9.5 at 5000MHz.




    Remember he's using an AVX offset of -3.
    LinX is running at 4.7 ghz on his system.  Then 5 ghz when its finished.




    Right, OK.

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #21
    tekleadm
    New Member
    • Total Posts : 15
    • Reward points : 0
    • Joined: 2008/06/06 09:54:35
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/26 11:16:44 (permalink)
    As others have mentioned, you have a throttling issue most likely, be it from power or thermal.
     
    First, some background. The linpack benchmark is a good choice for stress testing, but most of these repackaged front-end tools either use outdated Intel Linpack executables, wrong configuration, or a poor implementation. On Intel cpu's, your best choice is to download Intel's linpack program directly from their site (https://software.intel.com/en-us/articles/intel-mkl-benchmarks-suite). These are continuously updated for each new generation of Intel cpus and highly tuned for them. They either perform poorly or not at all on AMD cpus. When you run it, the proper values for AVX testing are to set the "number of equations" and "leading dimension" to an odd multiple of 16 and have "leading dimension" >= "number of equations", that is, greater than or equal to. For 8th and 9th gen Intel, the number of vector units and width result in an ideal GFLOPs of "core speed" x 16 per core (not thread). For example, at 5 GHz, you would expect an ideal of 80 GFLOPs per core. Intel's linpack benchmark fully utilizes each core so hyperthreading may provide little or even negative benefit. If I remember correctly, by default, it will run threads up to the number of cores, and will not utilize hyperthreads.
     
    Second, actual usage and testing. Intel linpack benchmark is a command line tool. It takes a few parameters as input while running it, but does not ask for the number of threads. Instead, you must set the number of threads using an environment variable. On Windows, just execute "set OMP_NUM_THREADS=X" where "X" is the number of threads, and then run the "linpack_xeon64.exe" (assuming you are on 64-bit). Here are my results on an i5-8600k at 5 GHz (0 AVX offset):
     
    Single-threaded performance:
    Average = 74.5295 (93% of ideal)
    Size   LDA    Align. Time(s)    GFlops   Residual     Residual(norm) Check
    20016  20016  4      71.746     74.5263  3.712881e-10 3.285072e-02   pass
    20016  20016  4      71.752     74.5198  3.712881e-10 3.285072e-02   pass
    20016  20016  4      71.730     74.5423  3.712881e-10 3.285072e-02   pass


    Multi-threaded performance:
    Average = 374.8176 (78% of ideal, ratio 5.03x ratio)
    Size   LDA    Align. Time(s)    GFlops   Residual     Residual(norm) Check
    20016  20016  4      14.273     374.6117 3.547637e-10 3.138867e-02   pass
    20016  20016  4      14.268     374.7390 3.547637e-10 3.138867e-02   pass
    20016  20016  4      14.255     375.1020 3.547637e-10 3.138867e-02   pass
     
    Third, interpreting these results. To ensure stability, the check column must say pass, which signals that the error value is low. Another thing to check is that each result generated from each trial is identical, since it is repeating the exact same problem. Now, running the single-threaded test is just to set a baseline for performance. You should not be power-limited in this scenario, possibly thermal-limited depending on cooling. My 8600k was around 70c for these tests and performed near ideal. (You would see better performance in this case depending on cache overclock and RAM speed since the 20016 problem size uses about 3.2 GB of RAM.) Next, in the multi-threaded test, the performance increased, however relative to the single-threaded test, a lot of performance was lost in scaling. I only receive 78% of the ideal performance, and even in scaling from 1 to 6 threads, the performance only increased by a little over 5x. While there is some overhead in multithreading, this much lost means that I too am power-limited. My overclock is still stable however, and at 1.4V, I am reaching about 85c during that test so I have little room to increase it.
     
    Now the 9900k has 8 cores, so assuming similar scaling in multithreading, 78% of (8 x 80 = 640 GFLOPS) is 500 GFLOPS. However, the 9900k has more cache so and usually is paired with higher speed RAM (versus mine which is 3000MHz - CL15), so a value over 500 GFLOPs is highly expected.
     
    The last bit I will say is that tuning for Linpack performance means tuning for a full AVX heavy workload. Unless you are constantly rendering, encoding, or running scientific programs, you will be leaving performance that would be stable for gaming on the table since the AVX units are power-hungry and will limit your overclock ceiling. Using them will increase the power usage as well as the thermal load versus sticking to the scalar and SSE units.
    #22
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/26 13:25:27 (permalink)
    tekleadm
    As others have mentioned, you have a throttling issue most likely, be it from power or thermal.
     
    First, some background. The linpack benchmark is a good choice for stress testing, but most of these repackaged front-end tools either use outdated Intel Linpack executables, wrong configuration, or a poor implementation. On Intel cpu's, your best choice is to download Intel's linpack program directly from their site (https://software.intel.com/en-us/articles/intel-mkl-benchmarks-suite). These are continuously updated for each new generation of Intel cpus and highly tuned for them. They either perform poorly or not at all on AMD cpus. When you run it, the proper values for AVX testing are to set the "number of equations" and "leading dimension" to an odd multiple of 16 and have "leading dimension" >= "number of equations", that is, greater than or equal to. For 8th and 9th gen Intel, the number of vector units and width result in an ideal GFLOPs of "core speed" x 16 per core (not thread). For example, at 5 GHz, you would expect an ideal of 80 GFLOPs per core. Intel's linpack benchmark fully utilizes each core so hyperthreading may provide little or even negative benefit. If I remember correctly, by default, it will run threads up to the number of cores, and will not utilize hyperthreads.
     
    Second, actual usage and testing. Intel linpack benchmark is a command line tool. It takes a few parameters as input while running it, but does not ask for the number of threads. Instead, you must set the number of threads using an environment variable. On Windows, just execute "set OMP_NUM_THREADS=X" where "X" is the number of threads, and then run the "linpack_xeon64.exe" (assuming you are on 64-bit). Here are my results on an i5-8600k at 5 GHz (0 AVX offset):
     
    Single-threaded performance:
    Average = 74.5295 (93% of ideal)
    Size   LDA    Align. Time(s)    GFlops   Residual     Residual(norm) Check
    20016  20016  4      71.746     74.5263  3.712881e-10 3.285072e-02   pass
    20016  20016  4      71.752     74.5198  3.712881e-10 3.285072e-02   pass
    20016  20016  4      71.730     74.5423  3.712881e-10 3.285072e-02   pass


    Multi-threaded performance:
    Average = 374.8176 (78% of ideal, ratio 5.03x ratio)
    Size   LDA    Align. Time(s)    GFlops   Residual     Residual(norm) Check
    20016  20016  4      14.273     374.6117 3.547637e-10 3.138867e-02   pass
    20016  20016  4      14.268     374.7390 3.547637e-10 3.138867e-02   pass
    20016  20016  4      14.255     375.1020 3.547637e-10 3.138867e-02   pass
     
    Third, interpreting these results. To ensure stability, the check column must say pass, which signals that the error value is low. Another thing to check is that each result generated from each trial is identical, since it is repeating the exact same problem. Now, running the single-threaded test is just to set a baseline for performance. You should not be power-limited in this scenario, possibly thermal-limited depending on cooling. My 8600k was around 70c for these tests and performed near ideal. (You would see better performance in this case depending on cache overclock and RAM speed since the 20016 problem size uses about 3.2 GB of RAM.) Next, in the multi-threaded test, the performance increased, however relative to the single-threaded test, a lot of performance was lost in scaling. I only receive 78% of the ideal performance, and even in scaling from 1 to 6 threads, the performance only increased by a little over 5x. While there is some overhead in multithreading, this much lost means that I too am power-limited. My overclock is still stable however, and at 1.4V, I am reaching about 85c during that test so I have little room to increase it.
     
    Now the 9900k has 8 cores, so assuming similar scaling in multithreading, 78% of (8 x 80 = 640 GFLOPS) is 500 GFLOPS. However, the 9900k has more cache so and usually is paired with higher speed RAM (versus mine which is 3000MHz - CL15), so a value over 500 GFLOPs is highly expected.
     
    The last bit I will say is that tuning for Linpack performance means tuning for a full AVX heavy workload. Unless you are constantly rendering, encoding, or running scientific programs, you will be leaving performance that would be stable for gaming on the table since the AVX units are power-hungry and will limit your overclock ceiling. Using them will increase the power usage as well as the thermal load versus sticking to the scalar and SSE units.



    The Intel binaries for download here:
    https://software.intel.com/file/835980/downloadAe2Lr-De
    are the same ones for Linx 0.9.6.
    https://hwtips.tistory.com/1611
     
    So Linx 0.9.6 should be used.
    And yes, LinX is truly a 'worst case' torture test.
    Passing 35,000 sample size is harder than passing FMA3 small FFT prime95 (29.8 build 6!).
    Passing this should only be used for testing rock solid stability at base clocks.  Most users without golden chips or delidded won't be able to pass this overclocked at 5 ghz.
    #23
    tekleadm
    New Member
    • Total Posts : 15
    • Reward points : 0
    • Joined: 2008/06/06 09:54:35
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/26 14:20:51 (permalink)
    Yes, the Intel linpack binary used in "LinX v0.9.6 for Intel (한국어)" is the exact same as the current version posted by Intel.
     
    sha1sum linx.exe intl.exe
    1f8849ba277c1808112451fab4328cc498dd74bb  linx.exe
    1f8849ba277c1808112451fab4328cc498dd74bb  intl.exe
     
    Using that is probably fine unless the frontend GUI eats excessive cpu time. Also, still be sure to use odd multiples of 16 (for AVX) to ensure optimal performance.
     
    I would argue that being unable to pass Intel linpack at stock speeds/settings would necessitate a return of that cpu as defective.
    #24
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/26 15:30:55 (permalink)
    I can pass 10 runs of LinX 0.9.5, maximum memory used (14-odd GB, and problem size +40K) with no errors and identical residuals.
     
    I keep hearing the term 'throttling'. Watching hwinfo while running LinX, I see no indication of thermal throttling, CPU temps never reach 80°C. How do I check power throttling?
     
    Also, what BIOS setting would cause the low score?
     
    Really would like some info, please.
     
     

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #25
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/26 16:08:54 (permalink)
    What were your Gflops?
     
    The one problem with the Z390 Dark is it doesn't support current monitoring.  If the VRM's supported monitoring, then HWinfo64 should also show the values.  So there isn't a clear way to determine how much current LinX is pulling.
    At 35000 sample size and 5 ghz, you should be pulling between 160 to 200 amps, depending on vcore.  That would get you about 500 to 510 gflops, and good luck cooling that processor with that current load.
    300 gflops is like you are not even using AVX.
    I'm not familiar with the Dark Bios.  Weren't there some settings for current protection or VRM load capacity or something?
    #26
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/26 16:11:25 (permalink)
    Kylearan
    What were your Gflops?
     
    The one problem with the Z390 Dark is it doesn't support current monitoring.  If the VRM's supported monitoring, then HWinfo64 should also show the values.  So there isn't a clear way to determine how much current LinX is pulling.
    At 35000 sample size and 5 ghz, you should be pulling between 160 to 200 amps, depending on vcore.  That would get you about 500 to 510 gflops, and good luck cooling that processor with that current load.
    300 gflops is like you are not even using AVX.
    I'm not familiar with the Dark Bios.  Weren't there some settings for current protection or VRM load capacity or something?


    hwinfo shows watts, could that not be extrapolated?


     
    I see nothing in BIOS regarding VRM.
     
    Do I need to submit a support ticket to get some input from evga?
    post edited by kelkel1 - 2019/10/26 16:46:37

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #27
    Kylearan
    New Member
    • Total Posts : 29
    • Reward points : 0
    • Joined: 2013/12/26 04:04:40
    • Status: offline
    • Ribbons : 1
    Re: Low GFlops in Linpack 2019/10/26 16:35:50 (permalink)
    kelkel1
     
    hwinfo shows watts, could that not be extrapolated?




    Watt is VID * Amps.
    Unless the MSR (CPU_PACKAGE_POWER_MSR) were rerouted to be read from the VRM, then this could be inaccurate as VID is not the same as vcore measured from the CPU die.
    #28
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/27 05:23:36 (permalink)
    Still waiting for an evga rep...
     
    While I am waiting, Kylearan mentioned something about VRMs.
     
    I presume that meant the VRMs were overheating? Touching the heatsinks on the board during stress testing, they do not seem very warm. Is that normal? If the VRMs are overheating, does that cause throttling, and would that cause the low GFlops?
     
    I need to find my infrared thermometer to verify heatsink temps.
     
     
    post edited by kelkel1 - 2019/10/27 05:50:45

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #29
    kelkel1
    Superclocked Member
    • Total Posts : 156
    • Reward points : 0
    • Joined: 2019/05/18 09:52:47
    • Status: offline
    • Ribbons : 0
    Re: Low GFlops in Linpack 2019/10/27 15:42:24 (permalink)
    Been trying to sort this out for over a week now.
     
    Tried installing my old Trident Z 3600 RAM, changing the Windows Power Plan to Power Saver.
     
    C-States disabled. AVX offset at 0. HT enabled. Other settings on Auto.
     
    Linpack Extreme 1.1.1, 3 passes using 14GB memory.
     
    CPU at 3600MHz, RAM using XMP1 at 3600MHz, GFlops averaged ~339.
    CPU at 5000MHz, RAM using XMP1 at 3600MHz, GFlops averaged ~332.
     
    Switched back to Trident Z 4500 kit.
     
    CPU at 3600MHz, RAM using XMP1 at 4500MHz, GFlops averaged ~341.
    CPU at 5000MHz, RAM using XMP1 at 4500MHz, GFlops averaged ~339.
     
    Both runs using BIOS 1.04.
     
    W T H?
     
    RMA time?
    post edited by kelkel1 - 2019/10/27 16:34:05

    Z390 DARK, 9900K, 2080 FTW3 ULTRA, GSKILL 4500, 960EVO M.2
    #30
    Page: 123 > Showing page 1 of 3
    Jump to:
  • Back to Mobile