RE: "-nobs"...

Message boards : Number crunching : RE: "-nobs"...
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile TimeLord04 Project Donor
Volunteer tester
Avatar

Send message
Joined: 9 Mar 06
Posts: 19813
Credit: 20,117,680
RAC: 48,474
United States
Message 1985723 - Posted: 18 Mar 2019, 3:47:45 UTC

3-17-2019 at 8:47 PM - Pacific

OK - I implemented the "-nobs" Command Line Parameter for the first time, today, on the Hackintosh. (iMac 18,3: Darwin 17.7.0)

Computer ID: 7952666


HOW do I know if this is working? I tried looking at the StdErr.txt Reports of SEVERAL work Units completed and Reported tonight between 18:00 and 20:00 Local Time, tonight. I can't find specifically where "-nobs" is or is NOT reported...

I've chosen a StdErr.txt Report from a recent Inconclusive Unit completed tonight. (See Below:)


Task 7512602855

Name 28dc06af.20147.33104.5.32.10.vlar_1
Workunit 3394630860
Created 17 Mar 2019, 8:43:26 UTC
Sent 17 Mar 2019, 13:22:41 UTC
Report deadline 9 May 2019, 18:22:23 UTC
Received 18 Mar 2019, 2:03:35 UTC <<--- (6:03 PM - 3-17-2019, Local Time.)
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x00000000)
Computer ID 7952666
Run time 2 min 40 sec <---[Are these Times good??? Is "-nobs" working???]
CPU time 2 min 36 sec <---[ (Ditto) ]
Validate state Checked, but no consensus yet
Credit 0.00
Device peak FLOPS 6,462.72 GFLOPS
Application version SETI@home v8
Anonymous platform (NVIDIA GPU)
Peak working set size 464.86 MB
Peak swap size 58,239.89 MB
Peak disk usage 0.20 MB


Stderr output

<core_client_version>7.8.6</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_CUDA: Found 2 CUDA device(s):
Device 1: GeForce GTX 1070, 8191 MiB, regsPerBlock 65536
computeCap 6.1, multiProcs 15
pciBusID = 7, pciSlotID = 0
Device 2: GeForce GTX 1050, 2047 MiB, regsPerBlock 65536
computeCap 6.1, multiProcs 5
pciBusID = 1, pciSlotID = 0
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 1070 is okay
SETI@home using CUDA accelerated device GeForce GTX 1070
Unroll autotune 15. Overriding Pulse find periods per launch. Parameter -pfp set to 15

setiathome v8 enhanced x41p_V0.97, Cuda 9.00 special
Modifications done by petri33, compiled by TBar

Detected setiathome_enhanced_v8 task. Autocorrelations enabled, size 128k elements.
Work Unit Info:
...............
WU true angle range is : 0.016776
Sigma 81
Sigma > GaussTOffsetStop: 81 > -17
Thread call stack limit is: 1k
Triplet: peak=11.0311, time=60.59, period=19.15, d_freq=1420099606.22, chirp=17.074, fft_len=32
Autocorr: peak=18.65217, time=60.4, delay=3.6337, d_freq=1420096506.29, chirp=-19.04, fft_len=128k
Pulse: peak=3.35284, time=53.71, period=7.871, d_freq=1420098882.04, score=1.008, chirp=-22.277, fft_len=512
Spike: peak=24.72502, time=60.4, d_freq=1420097605.98, chirp=25.808, fft_len=128k
Spike: peak=25.7455, time=60.4, d_freq=1420097605.98, chirp=25.812, fft_len=128k
Spike: peak=25.55931, time=60.4, d_freq=1420097605.98, chirp=25.816, fft_len=128k
Spike: peak=24.14177, time=60.4, d_freq=1420097605.98, chirp=25.819, fft_len=128k
Spike: peak=24.51218, time=73.82, d_freq=1420102010.43, chirp=28.986, fft_len=128k
Spike: peak=25.73269, time=73.82, d_freq=1420102010.45, chirp=28.994, fft_len=128k
Spike: peak=26.63828, time=73.82, d_freq=1420102010.44, chirp=28.995, fft_len=128k
Spike: peak=26.35579, time=73.82, d_freq=1420102010.44, chirp=28.996, fft_len=128k
Spike: peak=24.95525, time=73.82, d_freq=1420102010.43, chirp=28.997, fft_len=128k
Spike: peak=24.2127, time=73.82, d_freq=1420102010.46, chirp=29.004, fft_len=128k
Spike: peak=24.28826, time=73.82, d_freq=1420102010.45, chirp=29.005, fft_len=128k
Pulse: peak=2.768046, time=53.7, period=5.409, d_freq=1420097964.65, score=1.011, chirp=40.551, fft_len=256
Triplet: peak=10.51012, time=18.24, period=6.79, d_freq=1420099585.17, chirp=-44.819, fft_len=32
Pulse: peak=5.350749, time=53.9, period=14.16, d_freq=1420099196.08, score=1.012, chirp=-56.408, fft_len=4k
Pulse: peak=1.662429, time=53.74, period=3.15, d_freq=1420100447.95, score=1.028, chirp=59.225, fft_len=1024
Pulse: peak=4.235791, time=53.9, period=11.11, d_freq=1420097596.41, score=1.003, chirp=64.094, fft_len=4k
Triplet: peak=12.35086, time=59.36, period=26.17, d_freq=1420093878.07, chirp=66.161, fft_len=128
Pulse: peak=4.429358, time=53.79, period=11.92, d_freq=1420101937.81, score=1.022, chirp=-97.074, fft_len=2k

Best spike: peak=26.63828, time=73.82, d_freq=1420102010.44, chirp=28.995, fft_len=128k
Best autocorr: peak=18.65217, time=60.4, delay=3.6337, d_freq=1420096506.29, chirp=-19.04, fft_len=128k
Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.12e+11, d_freq=0,
score=-12, null_hyp=0, chirp=0, fft_len=0
Best pulse: peak=1.662429, time=53.74, period=3.15, d_freq=1420100447.95, score=1.028, chirp=59.225, fft_len=1024
Best triplet: peak=12.35086, time=59.36, period=26.17, d_freq=1420093878.07, chirp=66.161, fft_len=128

Spike count: 11
Autocorr count: 1
Pulse count: 6
Triplet count: 3
Gaussian count: 0

19:03:27 (40611): called boinc_finish(0)

</stderr_txt>
]]>
TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos
ID: 1985723 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 1277
Credit: 481,875,188
RAC: 2,066,734
United States
Message 1985726 - Posted: 18 Mar 2019, 4:11:27 UTC - in response to Message 1985723.  

the -nobs argument is not printed in the stderr.txt file.

the only way to see it, is to look at the run time vs cpu time for your tasks. with -nobs, the run times end up being roughly equal.

i looked at your most recent tasks and can tell you have -nobs added.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1985726 · Report as offensive
Profile TimeLord04 Project Donor
Volunteer tester
Avatar

Send message
Joined: 9 Mar 06
Posts: 19813
Credit: 20,117,680
RAC: 48,474
United States
Message 1985751 - Posted: 18 Mar 2019, 12:03:11 UTC - in response to Message 1985726.  

the -nobs argument is not printed in the stderr.txt file.

the only way to see it, is to look at the run time vs cpu time for your tasks. with -nobs, the run times end up being roughly equal.

i looked at your most recent tasks and can tell you have -nobs added.

Thank you. Just wanted to make sure that it was working. 😀


TL
TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos
ID: 1985751 · Report as offensive
Profile Cliff Harding
Volunteer tester
Avatar

Send message
Joined: 18 Aug 99
Posts: 1411
Credit: 99,700,565
RAC: 43,787
United States
Message 1985757 - Posted: 18 Mar 2019, 12:42:09 UTC - in response to Message 1985726.  

the -nobs argument is not printed in the stderr.txt file.

the only way to see it, is to look at the run time vs cpu time for your tasks. with -nobs, the run times end up being roughly equal.

i looked at your most recent tasks and can tell you have -nobs added.

Okay, I'll bite - what is the -nobs argument and if the run time (GPU) & CPU time are roughly equal, why use it? It seems to me that if the above is correct that's an awful lot of CPU cycles that could be used for something else or am I missing something?.


I don't buy computers, I build them!!
ID: 1985757 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 1277
Credit: 481,875,188
RAC: 2,066,734
United States
Message 1985759 - Posted: 18 Mar 2019, 12:49:37 UTC - in response to Message 1985757.  
Last modified: 18 Mar 2019, 12:49:57 UTC

it only applies to the linux special app. it forces an entire CPU thread to be used for feeding the GPU job, which speeds it up a little bit.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1985759 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 3013
Credit: 127,825,861
RAC: 816,041
United States
Message 1985772 - Posted: 18 Mar 2019, 13:50:19 UTC - in response to Message 1985759.  

it only applies to the linux special app. it forces an entire CPU thread to be used for feeding the GPU job, which speeds it up a little bit.


Or if the CPU is "feeling over-committed" it will sustain the speed of the gpu processing even though the CPU task are running much slower. This is from my AMD 2990wx experience. Otherwise, my GPUs on the 2990wx would slow down too.

Tom
I will stop procrastinating tomorrow.
\\// Live Long & Prosper (starting tomorrow ;)
ID: 1985772 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 12937
Credit: 136,780,131
RAC: 51,695
United Kingdom
Message 1985778 - Posted: 18 Mar 2019, 14:07:32 UTC - in response to Message 1985772.  

Or if the CPU is "feeling over-committed" it will sustain the speed of the gpu processing even though the CPU task are running much slower.
Yes, certainly, because the CPU part of GPU tasks run at a higher thread process priority than native CPU tasks.
ID: 1985778 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 7578
Credit: 463,617,128
RAC: 20,083
Panama
Message 1985779 - Posted: 18 Mar 2019, 14:08:57 UTC

From my experience -nobs while keeping 4 cores on the CPU running at 100% making almost nothing, produces a gain of about 2-3% in the GPU output but increases the CPU temp by few degrees (3 C) and make it use about 50W more of power from the outlet. That is why i stop to use it.
ID: 1985779 · Report as offensive

Message boards : Number crunching : RE: "-nobs"...


 
©2019 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.