+- +-
Say hello if visiting :) by Gecko
11 Jan 2023, 07:43:05 pm

Seti is down again by Mike
09 Aug 2017, 10:02:44 am

Some considerations regarding OpenCL MultiBeam app tuning from algorithm view by Raistmer
11 Dec 2016, 06:30:56 am

Loading APU to the limit: performance considerations by Mike
05 Nov 2016, 06:49:26 am

Better sleep on Windows - new round by Raistmer
26 Aug 2016, 02:02:31 pm

Author Topic: Better sleep on Windows - new round  (Read 44142 times)

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #15 on: 19 Aug 2016, 03:18:42 pm »
Mike's results:

WU : AR075.wu
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe -verb -nog :
  Elapsed 474.039 secs
      CPU 228.042 secs
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3486.exe  -use_sleep :
  Elapsed 497.494 secs, speedup: -4.95%  ratio: 0.95x
      CPU 180.618 secs, speedup: 20.80%  ratio: 1.26x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe  -use_sleep :
  Elapsed 500.524 secs, speedup: -5.59%  ratio: 0.95x
      CPU 177.576 secs, speedup: 22.13%  ratio: 1.28x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep0.exe  -use_sleep :
  Elapsed 472.639 secs, speedup: 0.30%  ratio: 1.00x
      CPU 406.617 secs, speedup: -78.31%  ratio: 0.56x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep1.exe  -use_sleep :
  Elapsed 474.594 secs, speedup: -0.12%  ratio: 1.00x
      CPU 285.856 secs, speedup: -25.35%  ratio: 0.80x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_SwitchTothread.exe  -use_sleep :
  Elapsed 472.914 secs, speedup: 0.24%  ratio: 1.00x
      CPU 407.116 secs, speedup: -78.53%  ratio: 0.56x


GT720, CPU busy, use_sleep active results:

MB8_win_x86_SSE3_OpenCL_NV_SoG_Sleep0.exe  :
  Elapsed 3031.125 secs, speedup: 46.35%  ratio: 1.86x
      CPU 365.136 secs, speedup: 90.83%  ratio: 10.90x
MB8_win_x86_SSE3_OpenCL_NV_SoG_Sleep1.exe  :
  Elapsed 3016.956 secs, speedup: 46.60%  ratio: 1.87x
      CPU 324.747 secs, speedup: 91.84%  ratio: 12.26x
MB8_win_x86_SSE3_OpenCL_NV_SoG_STT.exe  :
  Elapsed 3037.066 secs, speedup: 46.24%  ratio: 1.86x
      CPU 348.428 secs, speedup: 91.25%  ratio: 11.42x
setiathome_8.16_windows_intelx86__opencl_nvidia_SoG.exe  :
  Elapsed 3012.764 secs, speedup: 46.67%  ratio: 1.88x
      CPU 1721.908 secs, speedup: 56.74%  ratio: 2.31x
setiathome_8.17_windows_intelx86__opencl_nvidia_SoG.exe  :
  Elapsed 3016.387 secs, speedup: 46.61%  ratio: 1.87x
      CPU 324.966 secs, speedup: 91.83%  ratio: 12.25x


So, for these places current choice of sleep(1) is optimal one even w/o high-prec timer activation.
I'll repreat test with -high_prec_timer now for GT720

And counters:
Sleep0: class SleepQuantum:      total=13556.229,   N=3065,   <>=4.4229134,   min=0.011274812   max=17.502548
Sleep1: class SleepQuantum:      total=3163.4568,   N=3153,   <>=1.0033165,   min=0.86198002   max=40.154495
STT:     class SleepQuantum:      total=16757.236,   N=2412,   <>=6.9474446,   min=0.011177354   max=18.476677

« Last Edit: 19 Aug 2016, 03:28:56 pm by Raistmer »

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #16 on: 20 Aug 2016, 01:42:08 pm »
binaries updated to fix newly introduced bug in signal logging.
WARNING: don't use binaries from V2 online.
« Last Edit: 22 Aug 2016, 05:29:50 am by Raistmer »

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #17 on: 21 Aug 2016, 10:20:07 am »
WU : AR075.wu
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe -verb -nog :
  Elapsed 474.039 secs
      CPU 228.042 secs
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe   :
  Elapsed 476.706 secs, speedup: -0.56%  ratio: 0.99x
      CPU 228.994 secs, speedup: -0.42%  ratio: 1.00x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep0.exe   :
  Elapsed 475.049 secs, speedup: -0.21%  ratio: 1.00x
      CPU 289.210 secs, speedup: -26.82%  ratio: 0.79x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep1.exe   :
  Elapsed 475.277 secs, speedup: -0.26%  ratio: 1.00x
      CPU 288.009 secs, speedup: -26.30%  ratio: 0.79x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_STT.exe   :
  Elapsed 474.973 secs, speedup: -0.20%  ratio: 1.00x
      CPU 288.415 secs, speedup: -26.47%  ratio: 0.79x
 
WU : PG1327_v7.wu
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe -verb -nog :
  Elapsed 62.481 secs
      CPU 36.145 secs
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe   :
  Elapsed 61.959 secs, speedup: 0.84%  ratio: 1.01x
      CPU 36.348 secs, speedup: -0.56%  ratio: 0.99x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep0.exe   :
  Elapsed 62.114 secs, speedup: 0.59%  ratio: 1.01x
      CPU 42.370 secs, speedup: -17.22%  ratio: 0.85x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep1.exe   :
  Elapsed 61.824 secs, speedup: 1.05%  ratio: 1.01x
      CPU 42.557 secs, speedup: -17.74%  ratio: 0.85x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_STT.exe   :
  Elapsed 62.313 secs, speedup: 0.27%  ratio: 1.00x
      CPU 42.604 secs, speedup: -17.87%  ratio: 0.85x
 
CPU consumption is higher on all versions.

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #18 on: 22 Aug 2016, 05:43:00 am »
I see no -use_sleep used.
Is it idle CPU? or busy CPU run?

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #19 on: 22 Aug 2016, 12:56:50 pm »
I see no -use_sleep used.
Is it idle CPU? or busy CPU run?

Well i better repeat.  :o

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #20 on: 22 Aug 2016, 04:28:26 pm »

Not much different.
Just slower.

WU : AR075.wu
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe -verb -nog :
  Elapsed 474.039 secs
      CPU 228.042 secs
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe   :
  Elapsed 543.290 secs, speedup: -14.61%  ratio: 0.87x
      CPU 194.861 secs, speedup: 14.55%  ratio: 1.17x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep0.exe   :
  Elapsed 495.577 secs, speedup: -4.54%  ratio: 0.96x
      CPU 414.791 secs, speedup: -81.89%  ratio: 0.55x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep1.exe   :
  Elapsed 492.114 secs, speedup: -3.81%  ratio: 0.96x
      CPU 297.541 secs, speedup: -30.48%  ratio: 0.77x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_STT.exe   :
  Elapsed 483.082 secs, speedup: -1.91%  ratio: 0.98x
      CPU 415.961 secs, speedup: -82.41%  ratio: 0.55x
 
WU : PG1327_v7.wu
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe -verb -nog :
  Elapsed 62.481 secs
      CPU 36.145 secs
MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe   :
  Elapsed 64.840 secs, speedup: -3.78%  ratio: 0.96x
      CPU 38.345 secs, speedup: -6.09%  ratio: 0.94x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep0.exe   :
  Elapsed 63.978 secs, speedup: -2.40%  ratio: 0.98x
      CPU 58.812 secs, speedup: -62.71%  ratio: 0.61x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_Sleep1.exe   :
  Elapsed 64.277 secs, speedup: -2.87%  ratio: 0.97x
      CPU 44.944 secs, speedup: -24.34%  ratio: 0.80x
MB8_win_x86_SSE2_OpenCL_ATi_HD5_STT.exe   :
  Elapsed 65.041 secs, speedup: -4.10%  ratio: 0.96x
      CPU 59.062 secs, speedup: -63.40%  ratio: 0.61x
 

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #21 on: 23 Aug 2016, 02:36:34 am »
r3500:class SleepQuantum:      total=2.8579862,   N=3,   <>=0.95266207,   min=0.93661302   max=0.97626472
Sleep0: class SleepQuantum:      total=4.8358912,   N=2704,   <>=0.0017884213,   min=0.00054984231   max=0.4228799
Sleep1: class SleepQuantum:      total=2148.8459,   N=1791,   <>=1.1998023,   min=0.86739361   max=3.0483601
STT: class SleepQuantum:      total=3.9076965,   N=2704,   <>=0.001445154,   min=0.0004952898   max=0.0027276319

The same question. CPU idle or busy? Or, maybe, single CPU core free only?
Sleep behavior strongly depends from host load that's I always ask for full description of test conditions.
And for prev run w/o sleep enabled - no explanation why these builds consume much more CPU  :o




Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #22 on: 23 Aug 2016, 04:39:29 am »
r3500:class SleepQuantum:      total=2.8579862,   N=3,   <>=0.95266207,   min=0.93661302   max=0.97626472
Sleep0: class SleepQuantum:      total=4.8358912,   N=2704,   <>=0.0017884213,   min=0.00054984231   max=0.4228799
Sleep1: class SleepQuantum:      total=2148.8459,   N=1791,   <>=1.1998023,   min=0.86739361   max=3.0483601
STT: class SleepQuantum:      total=3.9076965,   N=2704,   <>=0.001445154,   min=0.0004952898   max=0.0027276319

The same question. CPU idle or busy? Or, maybe, single CPU core free only?
Sleep behavior strongly depends from host load that's I always ask for full description of test conditions.
And for prev run w/o sleep enabled - no explanation why these builds consume much more CPU  :o

Yep 7 cores were in use.

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #23 on: 24 Aug 2016, 01:48:45 am »
That shows the need of fixed amount sleep in case of underloaded CPU.
GPU app has bigger priority so, if some free CPU resource awailable, it will be scheduled for exection there.
What strange is no differencies in STT and Sleep(0) behavior. From what I read on main forums Sleep(0) should return to the same process immediately so just spin with full CPU busy while STT should give up CPU slice always(wrong, only if there are ready threads on the same CPU). So, in SleepQuantum counter it should have bigger mean value (hard to imagine that with absolute most of 2704 occurencies process was exactly at the end of its current time slice). Nevertheless once can see VERY close mean times (<>) for Sleep(0) and STT. Strange. If so I don't see any advantage of STT at all  :-\
[NB: Windows time slice ~10-15 ms and STT mean is 0.0014 ms]
« Last Edit: 24 Aug 2016, 08:37:43 am by Raistmer »

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #24 on: 24 Aug 2016, 08:49:51 am »
I have to remove r3500 from this bench  because it doesn`t even start with all cores in use.

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #25 on: 24 Aug 2016, 08:52:43 am »
I have to remove r3500 from this bench  because it doesn`t even start with all cores in use.

Ok for now, there is separate issue we just discovered...

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #26 on: 24 Aug 2016, 08:56:46 am »
I have to remove r3500 from this bench  because it doesn`t even start with all cores in use.

Ok for now, there is separate issue we just discovered...

Also sleep versions doesn`t even start.
Zero CPU usage on GPU task so i aborted after 5 minutes.
Not even wisgen started.

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #27 on: 24 Aug 2016, 08:58:42 am »
I have to remove r3500 from this bench  because it doesn`t even start with all cores in use.

Ok for now, there is separate issue we just discovered...

Also sleep versions doesn`t even start.
Zero CPU usage on GPU task so i aborted after 5 minutes.
Not even wisgen started.

Please remove all wisgen tasks, run bench, await ~5mins, locate stderr.txt in ScienceApps folder and attach it as is.

Offline Mike

  • Alpha Tester
  • Knight who says 'Ni!'
  • ***
  • Posts: 2427
Re: Better sleep on Windows - new round
« Reply #28 on: 24 Aug 2016, 09:15:44 am »
I have to remove r3500 from this bench  because it doesn`t even start with all cores in use.

Ok for now, there is separate issue we just discovered...

Also sleep versions doesn`t even start.
Zero CPU usage on GPU task so i aborted after 5 minutes.
Not even wisgen started.

Please remove all wisgen tasks, run bench, await ~5mins, locate stderr.txt in ScienceApps folder and attach it as is.

Host reseted after 5 minutes.
FX can`t use permanent all 8 cores.
I told you that before.

stderr attached.

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Better sleep on Windows - new round
« Reply #29 on: 24 Aug 2016, 12:26:06 pm »
Host reseted after 5 minutes.
FX can`t use permanent all 8 cores.
Could it be power issues? Maybe more strong power supply needed?

stderr attached.
thanks. app processed OK some time, even found some spikes.

 

Welcome, Guest.
Please login or register.
 
 
 
Forgot your password?
Members
Total Members: 97
Latest: ToeBee
New This Month: 0
New This Week: 0
New Today: 0
Stats
Total Posts: 59559
Total Topics: 1672
Most Online Today: 23
Most Online Ever: 983
(20 Jan 2020, 03:17:55 pm)
Users Online
Members: 0
Guests: 12
Total: 12
Powered by EzPortal