Message boards :
Number crunching :
ATI AstroPulse released
Message board moderation
Previous · 1 · 2 · 3 · 4
Author | Message |
---|---|
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Yes, credit granting still weird. But it should not affect "science" as long as not too much peoples would leave because of credit weirdness. |
Oleg Send message Joined: 23 Jul 09 Posts: 18 Credit: 235,786 RAC: 0 |
Raistmer Ñчитает 30 минут, прогреÑÑ 0%, загрузка гпу 97%(Ñо Ñпадами(на 2-3 Ñекунды) каждые 4 минуты до 85%)- Ñто нормально или убивать процеÑÑ? ЗЫ: перезапуÑкал боинк - тоже Ñамое. ЗЗЫ: разобралÑÑ - перезагрузил комп и вÑе заработало нормально. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
I finally got a validated result and I think my wingmen are going to be disappointed That 72.85 Credits was from a Wu that only took 1,214.74 seconds (20 minutes) to complete on your GPU, so 72.85 Credits isn't too bad for that length of computation. Admittantly the first ten validated Wu's Credits suck, or paired with someone doing their first ten Wu's, which are both your wingmen in this case, but that improves once you're done at least 10 validated Wu's, or paired with someone who has had ten validated Wu's. Claggy |
Wandering Willie Send message Joined: 19 Aug 99 Posts: 136 Credit: 2,127,073 RAC: 0 |
I have not received any AP WU,s yet but credits look similar to the credits received on Seti@home Beta AP.s. Wich if you remember varied over a wide range. Michael |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Raistmer Стоит проверить, не было ли реÑтарта драйвера (в ÑиÑтемных логах). Я такой Ñитуации Ñо времен ранних верÑий у ÑÐµÐ±Ñ Ð½Ðµ вÑтречаю. "быть такого не должно" :) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I have not received any AP WU,s yet but credits look similar to the credits received on Seti@home Beta AP.s. Wich if you remember varied over a wide range. It's true, ATI AP should claim just same amount of credit as CPU AP does. It's server's credit granting weirdness, not app's one... |
Oleg Send message Joined: 23 Jul 09 Posts: 18 Credit: 235,786 RAC: 0 |
Стоит проверить, не было ли реÑтарта драйвера (в ÑиÑтемных логах). Ð’ логах ничего не нашел. Видуха Ñчитала милкивей, когда загрузилаÑÑŒ ВЮ аÑтропульÑа раÑчет переключилÑÑ Ð½Ð° него. |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
The credit granted to Skildude may not be as weird as it looks. I noticed this Found 30 single pulses and 30 repeating pulses, exiting.and maybe the 72 something credits is just where the lowest claiming one stopped. PROUD MEMBER OF Team Starfire World BOINC |
archae86 Send message Joined: 31 Aug 99 Posts: 909 Credit: 1,582,816 RAC: 0 |
I have an host which currently runs Einstein on CPU and this ATI Astropulse ap on a cheap ATI board (HD4550 512M--only 80 stream processors). At the moment I am typing the valid tasks list has unusually many results--as the ones returned when the servers came back up have not mostly be deleted. I've been getting credits in the 600 to 800 range for the ATI AP work, save for obvious premature terminations. I'm not aware of punishing my quorum partners in credit. For this 80-stream board the initial performance tuning options shown in the sample ap_info fragment work for crunching, but give a pretty bad user experience on almost any simple desktop interaction, and make, for example, Netflix streamed movies so jerky as to be unwatchable. This could be addressed by specifying no use of GPU for crunching during interaction, but hoping to preserve more crunch, I've been slowly backing down the performance tuning values, and am currently using this line: <cmdline>-ffa_block 512 -ffa_block_fetch 128</cmdline> These values are not likely ideal for this board, but seem to have lost not very much and perhaps no crunch performance, while giving a much better user experience, though I think the graphics are still probably compromised appreciably. I suspect the 80 stream boards want lower values here than the serious graphics 800 stream processor boards do. I'm looking forward to being able to try the lunatics ATI multibeam ap when it goes public. |
Les Send message Joined: 20 May 99 Posts: 53 Credit: 21,062,237 RAC: 18 |
I didn't see a specific mention of running down work units already in progress or previously downloaded. Is this something that is so obvious that I shouldn't be asking the question or can we safely set this up as long as we stop all science applications before making the changes to app_info? |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
Yes if you have any Astropulse onboard you can run the work down then install. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I didn't see a specific mention of running down work units already in progress or previously downloaded. Is this something that is so obvious that I shouldn't be asking the question or can we safely set this up as long as we stop all science applications before making the changes to app_info? With correct app_info editing one not need to empty cache, all just continue to work as before + new app will be available. Hence no mention of this action. But if you unsure the safe way to go is not to empty cache, but to make backup of BOINC data dir, disable network connection do upgrade and then, if all go fine, re-enable network connection. With empty cache and invalid app info you have perfect chances to trash much more tasks than your initial cache was. So safe way is just as I described, not to empty cache. |
Les Send message Joined: 20 May 99 Posts: 53 Credit: 21,062,237 RAC: 18 |
Thank You for the feedback. And would also like to say Thank You for the all of the hard work in creating GPU applications for those of us stuck with ATI cards. Right now I have eight AP tasks that have started so will play it safe because I do not want to waste all of the download time should I do something stupid (which happens every minute of every day). The extra four are in progress because of a shift to high priority mode. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
They're going to have to check the credit. I'm assuming the validator isn't giving full credit for the WU and is just looking at the completion times. I'm only getting 400-800 credits depending on the WU/blanking. These are valid results but getting substantially lower credit than expected In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
These are valid results but getting substantially lower credit than expected New credit system? Gruß, Gundolf |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
I just turned in an AP with zero blanking and got 665.16 credits for it. This was run on my CPU. http://setiathome.berkeley.edu/result.php?resultid=1724416785 PROUD MEMBER OF Team Starfire World BOINC |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
Perhaps we have a new credit system for the AP's. I'll go back to abusing animals and hiding for Johnny Law. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
©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.