SETI MB CUDA for Linux

[riofl]

Why play with your cache levels? Change from 6 to 10 and then back to 2, why? Pick a cache level and leave it there. I'm using 10 days. If you were using 6 days, it is fine also. 6 days cache means that the workunits downloaded now will be crunched in about 6 days, so you have 6 days to check for vlars. No need to run that script x times per hour or x times per day.

ahh yes except for one thing. i have seen even this new version of boinc obey the due dates and pick the next workunit from among the newly downloaded. if they stayed in ascending date order i would agree but it does not seem to work that way for me. at least 3 or 4 times so far i noticed a cuda and/or cpu workunit placed on hold to pick up one that had a closer due date that was recently downloaded. this means there is a danger the gpu app will reject a possible vlar before it can be flagged.

if this works out better which by the logic you presented makes a lot of sense and im sure it will, ill just leave things alone and if it rejects a few workunits before the script can run, oh well.

[macros]

Back in the days when cuda needed a whole core, I was running a 3+1 config in my quad core. All processes had the lowest priority (19) and I don't think I had any serious slowdown, maybe a minute or so, not more. And this was my everyday desktop so many things were running, firefox with many many tabs, full 3d compiz effects, everyday backups, etc.

Only now that cuda shares a core with the other seti@home tasks, I started renicing them only to make them higher priority than the other seti@home instances. I think -5 is not necessary.

Question regarding this. I am using the default settings in app_info.xml's <app_version> for cuda as follows:

Code: [Select]

<avg_ncpus>0.040000</avg_ncpus>
<max_ncpus>0.040000</avg_ncpus>
The problem is, that setiathome-CUDA process has demand obviously higher than that and is able to eat up CPU time of whole one core. That results in other (regular CPU) processes to fight over the CPU time, context switches, cache thrashing etc. ->

Code: [Select]

  PID  PR  NI  RES  SHR %CPU    TIME+  COMMAND
15538  39  19  48m 1472  101  13:44.03 AK_V8_linux64_s
15539  39  19  48m 1464  101  20:17.47 AK_V8_linux64_s
15540  39  19  48m 1468  101  20:25.35 AK_V8_linux64_s
15541  39  19  48m 1464   99  19:52.12 AK_V8_linux64_s
15544  39  19  48m 1484   99  20:30.54 AK_V8_linux64_s
15545  30  10 114m  10m   99  18:42.69 setiathome-CUDA
15546  39  19  48m 1488   94  19:55.04 AK_V8_linux64_s
16208  39  19  48m 1488   51  12:22.11 AK_V8_linux64_s
15542  39  19  48m 1472   46  12:14.44 AK_V8_linux64_s
Now to the question - am I doing something wrong and cuda does not behave correctly?
Or is this normal and I should just set avg_ncpus & max_ncpus to 1, and pin the process to some core + make it use it exclusively?

[pp]

Why play with your cache levels? Change from 6 to 10 and then back to 2, why? Pick a cache level and leave it there. I'm using 10 days.

ahh yes except for one thing. i have seen even this new version of boinc obey the due dates and pick the next workunit from among the newly downloaded.

Set cache to 10 days and let it fill up
Set BOINC to not receive more WUs
Run script
Let computer crunch for 10 days
Repeat

[pp]

Now to the question - am I doing something wrong and cuda does not behave correctly?
Or is this normal and I should just set avg_ncpus & max_ncpus to 1, and pin the process to some core + make it use it exclusively?

Are you still running CUDA 2.1? The 100% CPU was apparently a bug in those libraries. Upgrade CUDA to 2.3, nvidia-drivers to 190.xx  and replace your setiathome executable with the 2.2 version and optionally renice that process if you think it's too slow.

[sunu]

ahh yes except for one thing. i have seen even this new version of boinc obey the due dates and pick the next workunit from among the newly downloaded. if they stayed in ascending date order i would agree but it does not seem to work that way for me. at least 3 or 4 times so far i noticed a cuda and/or cpu workunit placed on hold to pick up one that had a closer due date that was recently downloaded. this means there is a danger the gpu app will reject a possible vlar before it can be flagged.

This happens only for vhar workunits, they have shorter deadlines than the rest. VLARs have "normal" deadlines and they are crunched when their time comes, about x(cache) days after they've been downloaded.

Macros, what pp says. Make sure you're using cuda 2.2 or later together with a compatible nvidia driver.

Shameless plug: I've reached #4 in the top hosts list. I don't know how long I can hold on there though. Attaching pdf for future proof.

[attachment deleted by admin]

[pp]

Congratulations sunu. My goal is a modest top 2000 like I managed in Seti Classic. I'm slowly reaching there and the upgrade to CUDA 2.3 pushed my RAC on the 9800GT over 5000 now which helps a lot 

[riofl]

ahh yes except for one thing. i have seen even this new version of boinc obey the due dates and pick the next workunit from among the newly downloaded. if they stayed in ascending date order i would agree but it does not seem to work that way for me. at least 3 or 4 times so far i noticed a cuda and/or cpu workunit placed on hold to pick up one that had a closer due date that was recently downloaded. this means there is a danger the gpu app will reject a possible vlar before it can be flagged.

This happens only for vhar workunits, they have shorter deadlines than the rest. VLARs have "normal" deadlines and they are crunched when their time comes, about x(cache) days after they've been downloaded.

Macros, what pp says. Make sure you're using cuda 2.2 or later together with a compatible nvidia driver.

Shameless plug: I've reached #4 in the top hosts list. I don't know how long I can hold on there though. Attaching pdf for future proof.

congrats!!! hehe mine is only a paltry 188

i have had bad luck with 2.3 and had to go back to 2.2 which is flawless for me.

you were right... rac gained 200 points in the last 8 hours. thanks! i didnt think frequent restarting made that huge a difference. still a bit odd with 30min workunits but that seems to be what things are now. not quite sure why. just gonna let it go and see where it falls. maybe in a week or 2 things will level out if its the wu being supplied. or maybe this is the norm for this distribution .. dunno..

after get another gtx285 to replace this tesla and i finish next year's crunching project i think i will get a different motherboard with more x16 slots and wind up with maybe 3 gtx285 in this machine. that should help things along. i think this tesla has seen better days. probably a bad ram chip.

[riofl]

Back in the days when cuda needed a whole core, I was running a 3+1 config in my quad core. All processes had the lowest priority (19) and I don't think I had any serious slowdown, maybe a minute or so, not more. And this was my everyday desktop so many things were running, firefox with many many tabs, full 3d compiz effects, everyday backups, etc.

Only now that cuda shares a core with the other seti@home tasks, I started renicing them only to make them higher priority than the other seti@home instances. I think -5 is not necessary.

Question regarding this. I am using the default settings in app_info.xml's <app_version> for cuda as follows:

Code: [Select]

<avg_ncpus>0.040000</avg_ncpus>
<max_ncpus>0.040000</avg_ncpus>
The problem is, that setiathome-CUDA process has demand obviously higher than that and is able to eat up CPU time of whole one core. That results in other (regular CPU) processes to fight over the CPU time, context switches, cache thrashing etc. ->

Code: [Select]

  PID  PR  NI  RES  SHR %CPU    TIME+  COMMAND
15538  39  19  48m 1472  101  13:44.03 AK_V8_linux64_s
15539  39  19  48m 1464  101  20:17.47 AK_V8_linux64_s
15540  39  19  48m 1468  101  20:25.35 AK_V8_linux64_s
15541  39  19  48m 1464   99  19:52.12 AK_V8_linux64_s
15544  39  19  48m 1484   99  20:30.54 AK_V8_linux64_s
15545  30  10 114m  10m   99  18:42.69 setiathome-CUDA
15546  39  19  48m 1488   94  19:55.04 AK_V8_linux64_s
16208  39  19  48m 1488   51  12:22.11 AK_V8_linux64_s
15542  39  19  48m 1472   46  12:14.44 AK_V8_linux64_s
Now to the question - am I doing something wrong and cuda does not behave correctly?
Or is this normal and I should just set avg_ncpus & max_ncpus to 1, and pin the process to some core + make it use it exclusively?

i think you will find best resonse setting your preferences to use 6 or 7 cpus instead of 8 leaving 1 for cuda and your desktop to use. i played around a bit with max_ncpus but did not find a huge difference. mine is set at 0.35.

absolutely if you do nothing else change your cuda tookit and sdk to 2.2 and get the 2.2 application. make sure your driver is at the minimum 185.14 or 185.29. i am using 185.29.

ver 2.1 had huge flaws in it . i have heard 2.3 is even better, however i have not had good luck with 2.3 so i went back to 2.2 until i can figure out what went wrong.

[sunu]

Thanks pp, riofl. For a long time I was the only one in the top 100 list using linux. Now I see we are three. I hope more will get in the top 100. We have to show that linux is at least equal to windows for seti crunching.

pp,compared to you I don't think I was that good in seti classic.

Small correction to riofl: The driver versions are 185.18.14 and 185.18.29. Latest is 185.18.31. Macros, if you go to cuda 2.3 you'll need 190.18.

Macros, what card are you using? Maybe that 99% is because your card goes out of memory?

[pp]

Thanks sunu. I crunched Classic from the start and had lots of PCs running but I quit when Seti switched to Boinc. I started again with Boinc only a month ago because one of my colleagues challenged me. I keep in front of him with only two GPUs running but I'm planning a Core i7/GTX 295 upgrade to crush him!  I never say no to a challenge...

[sunu]

Go for it pp, crush him!!! 

[pp]

With pleasure, although I can't stop looking at you guys at the top with hundreds of thousands of daily RAC... that's impressive! 

[sunu]

Tens of thousands not hundreds of thousands. You're probably looking at the Top participants list. I have only one pc so all I can manage is the Top hosts list, PCs with the highest RAC output.

Well, if you build that i7/GTX 295 maybe with a second 295 thrown in the mix you'll join the club. All I have is a Q6600 with a GTX295 and a GTX285 and I managed #4. But as I said, I don't know how long I'll last there and I probably won't upgrade for quite some time so sooner or later I'll start to go down the ladder.

[pp]

Good thing AtlasFolder doesn't do SETI: http://atlasfolding.com/?page_id=148
That's borderline... 

[sunu]

I don't think it is possible with BOINC to use all those motherboards as a single host, it would be good though. I think the best possible setup for a seti crunching host would be with this motherboard. 4 dual-slot or 7 single-slot graphics cards for uber RAC.