Message boards :
Number crunching :
Monitoring inconclusive GBT validations and harvesting data for testing
Message board moderation
Previous · 1 . . . 3 · 4 · 5 · 6 · 7 · 8 · 9 . . . 36 · Next
Author | Message |
---|---|
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
You can use whatever you need to upload files, I use google drive as I pay for it, but if you have dropbox, box, amazon cloud, etc, etc then by all means. If you are unsure where they are located it is in: C:\ProgramData\BOINC\projects\setiathome.berkeley.edu on windows. There are 2 files to upload One will be named similar to this: blc5_2bit_guppi_57403_68833_HIP11048_OFF_0003.20883.416.21.44.64.vlar The other is smaller and will be similar to this: blc5_2bit_guppi_57403_68833_HIP11048_OFF_0003.20883.416.21.44.64.vlar_4_0 |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
JASON! What are you using for comparison? What program? |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
I have the datafile, I'll do a manual run of stock cpu, cuda50, and sog. SoG one will be done when I can figure out how to run that in offline mode |
W-K 666 Send message Joined: 18 May 99 Posts: 19309 Credit: 40,757,560 RAC: 67 |
You can use whatever you need to upload files, I use google drive as I pay for it, but if you have dropbox, box, amazon cloud, etc, etc then by all means. Cannot see this one, will it already have been uploaded, I only noticed it was a _4 after it had completed, uploaded but not reported. I've uploaded the blc5_2bit_guppi_57403_68833_HIP11048_OFF_0003.20883.416.21.44.64.vlar data file and I have added blc3_2bit_guppi_57449_49131_HIP83613_OFF_0028.4249.831.17.26.179.vlar_3_0 and its data file |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
Then the result file is gone. :( We got the datafile though Once uploaded the host will delete the file, as it is no longer needed and takes up space. How long did it take with you CPU? |
W-K 666 Send message Joined: 18 May 99 Posts: 19309 Credit: 40,757,560 RAC: 67 |
Then the result file is gone. :( We got the datafile though It took 00:46:48 I've not got any more inconclusives at the moment, the other two >1 are abandoned or timed out. Will now allow network and get some more tasks. |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
Could you rerun that workunit? I will describe what you need to do, if you need it. As we have the datafile it is safe to report it as complete |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
JASON! rescmpv4, from Lunatics Downloads benchmark suite, in its tool folder. I use an older version than exists there, often separately by command line when comparing results. The newer one which exists there I believe adds 'super' and and 'exact' categories when printing weak similarity table, which was added once Cuda and OpenCL builds because close to stock CPU it became difficult to improve them. I'll probably update once I get around to it, and the newer apps become close to what they should be again. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
At the moment I am running through a double inconclusive stock sog ati vs stock sog nv both running windows I am going to use v2.08 is that the version your using? |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
Shouldn't we really looking at both data sets instead of just gbt? I am seeing a bunch of arecibo ones go inconclusive as well |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
Result from my manual test: To this unit here C:\Users\James\Desktop\comparison>rescmpv4 result_from_cpu.sah result_from_sog.sah Original inconclusive is from the ATI coproc, reporting 1 less pulse than both cpu and nv sog app |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. Here Here is the datafile and result file from my host. |
Mike Send message Joined: 17 Feb 01 Posts: 34347 Credit: 79,922,639 RAC: 80 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. Stock cuda is a false overflow and petris found 2 pulses less. It might still validate. With each crime and every kindness we birth our future. |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. I would run the app manually, but alas I don't have petri's app on my laptop, so I will have to run cuda50 and compare them I guess. Unless someone wants to build petri's app for windows and send it my way...... |
Mike Send message Joined: 17 Feb 01 Posts: 34347 Credit: 79,922,639 RAC: 80 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. Its always a good idea to bench with official app first to see if the result matches. With each crime and every kindness we birth our future. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14672 Credit: 200,643,578 RAC: 874 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. Ideally, with three apps: The stock app (gold standard) The app you usually use on you own rig (as a double check) The app being used by the third-party wingmate that you suspect may be returning out-of-tolerance results Of course, test (3) is difficult when tasks are being run on the main project by apps which aren't yet even available to all testers. |
Kiska Send message Joined: 31 Mar 12 Posts: 302 Credit: 3,067,762 RAC: 0 |
here are the results from my manual run C:\Users\James\Desktop\comparison>rescmpv5 8034200_cpu.sah result_from_cpu.sah result_from_cuda42.sah is from the cuda42 app manually ran/ result_from_cpu.sah is the cpu app manually ran. 8034200_cpu.sah is the computer which had the original inconclusive thrown. Results are all in this folder |
Mike Send message Joined: 17 Feb 01 Posts: 34347 Credit: 79,922,639 RAC: 80 |
In this case its enough for me. 99.22% is not what i call strongly similar. Should be 99.5x%+ Thats what i call stromngly similar. MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe -device 0 -sbs 384 -tt 60 -spike_fft_thresh 2048 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 32 -oclfft_tune_cw 32 / blc3_2bit_guppi_57451_19958_HIP62472_0005.7585.831.17.26.109.wu : AppName: MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3500.exe AppArgs: -device 0 -sbs 384 -tt 60 -spike_fft_thresh 2048 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 32 -oclfft_tune_cw 32 TaskName: blc3_2bit_guppi_57451_19958_HIP62472_0005.7585.831.17.26.109.wu Started at : 14:09:50.756 Ended at : 14:21:04.311 672.880 secs Elapsed 199.588 secs CPU time Speedup : 13.22% Ratio : 1.15x R2: .\ref\ref-MB8_win_x64_AVX_VS2010_r3308.exe-blc3_2bit_guppi_57451_19958_HIP62472_0005.7585.831.17.26.109.wu.res Result [b] : Strongly similar, Q= 99.95%[/b] R2: .\ref\ref-MB8_win_x86_SSE2_OpenCL_ATi_HD5_r3330.exe-blc3_2bit_guppi_57451_19958_HIP62472_0005.7585.831.17.26.109.wu.res Result [b] : Strongly similar, Q= 100.0%[/b] With each crime and every kindness we birth our future. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
We have a triple inconclusive. What is more interesting is _2 which is a petri special, also included. No clue what caused the inconclusive. Broken/overheating 8400GS caused the initial inconclusive (no mystery). I already built a Windows version, however, as with the other platforms its validation has known problems such that if it escapes into the Windows world it would highly likely cripple the project (both with throughput and co-validation of inaccurate results). So far the validator appears to be coping with the limited 'unofficial' Mac and Linux builds in circulation, though if significant co-validations begin to occur I will likely have to raise a discussion with project staff. I highly appreciate the excitement/anticipation, though No-one needs to replicate the known faulty alpha application's results at this time. Patience is needed for the stage of development that one is at, rather than widescale testing. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
petri33 Send message Joined: 6 Jun 02 Posts: 1668 Credit: 623,086,772 RAC: 156 |
Hi, Thanks to jason_gee and everyone else I have now a valid guppi result that I can run my version against to with debugging options on. That is a 'Kiska guppi' that I'm using to find a/the bug(s) that are keeping the special version still in the alpha. Keep on waiting. For now I'm testing with missing pulses. To overcome Heisenbergs: "You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones |
©2024 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.