Message boards :
Number crunching :
Can someone explain this to me, PLEASE?
Message board moderation
Author | Message |
---|---|
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
I have 2 crunchers, essentially identical. One currently has about 1700 WUs on board (BOINCBox), the other, about 1100 (unimatrix001). unimatrix001 gets "reached a limit" in answer to WU requests, BOINCBox just gets the standard "no WUs available". How is this possible? Both machines are dual quads (AMD Opteron 2356) with 4 GT 240s. |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
The limits are 400 WU per GPU card and 40 per CPU core. In your case you can get a max of 1920 WUs but only 1600 for GPUs and only 320 for CPUs. If the scheduler asks for CPU tasks and you already has 320 WU pending then you will get the "limit reached" message, wich seems to be the case with unimatrix. Also, I think (but Im not sure about the details) that if BOINC wants more CPU work and it dosnt get them, it gets stucked claiming for CPU and it dosnt ask tasks for the GPUs... |
David S Send message Joined: 4 Oct 99 Posts: 18352 Credit: 27,761,924 RAC: 12 |
The limits are 400 WU per GPU card and 40 per CPU core. Then how can my i7 http://setiathome.berkeley.edu/show_host_detail.php?hostid=5947619 have 800? 8 CPU cores and 1 GPU card should come to a max of 720. David Sitting on my butt while others boldly go, Waiting for a message from a small furry creature from Alpha Centauri. |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
The limits are 400 WU per GPU card and 40 per CPU core. Because, you're lucky? :D If you hasnt changed the hardware I dont know. May be the limit is applied when you already reached it but it dosnt apply to limit the number of tasks assigned when you are under the max... (Or they changed the limits recently...) (EDIT: Or you used the Boinc rescheduler and fooled the limits... ;-b ) |
JohnDK Send message Joined: 28 May 00 Posts: 1222 Credit: 451,243,443 RAC: 1,127 |
I have a PC with 1 GPU and 2 CPU core, and I have 500 WUs in progress. Haven't used rescheduler, sounds more like 50 WUs per core. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I have a PC with 1 GPU and 2 CPU core, and I have 500 WUs in progress. Haven't used rescheduler, sounds more like 50 WUs per core. Horacio seems to be mistaken. The limits have been 50/CPU 400/GPU for sometime now. Not all machines will reach this limits with 10 day cache settings, but for the faster ones this is easily obtained. I'm not sure if jravin was looking at the In progress tasks for the machines or locally. If the server thinks that the limit is reached it could be due to some ghosties floating around. I see 1832 & 1190 for the two machines right now. If some CPU cores are being set aside to feed the GPUs then you would subtract them from the total. So instead of 2000 tasks it would be 1950 for 1 core reserved, or 1900 for 2 reserved. If you are maxing at 1850 I would guess 3 reserved. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
I have a PC with 1 GPU and 2 CPU core, and I have 500 WUs in progress. Haven't used rescheduler, sounds more like 50 WUs per core. You're right. Ive digged a bit on the forum and I was wrong, the limit is 50 per CPU core. |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Thanks for the info. On both machines, I have reserved exactly ONE core for GT 240 support. Way back when, I noticed that with 8 cores running WUs, I got no more work done than with 7, due to stretch out by competition of CPU vs. GPU. So that's where I sit now. On both machines, I looked at the number of files in the data directory and the number shown via the Account -> Computers -> Tasks, and they were consistent with each other. (i.e., no gross diffs, which there would be if I had a lot of "ghosts"). |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Another thing to consider with the ongoing shorties. You may have reached the daily limit for the machine instead of the "reached max number of tasks in progress". It wasn't clear from your original post which message you were seeing. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
David S Send message Joined: 4 Oct 99 Posts: 18352 Credit: 27,761,924 RAC: 12 |
The limits are 400 WU per GPU card and 40 per CPU core. When I get a new computer, I download the current version of Boinc straight from Berkeley, attach to my projects, and let it go. I don't use any 3rd party enhancers and seldom intervene (if I'm in the basement and have time, I'll check on all the computers and maybe kick them into reporting any tasks that are ready, or retry a stuck download, but that's about it). The corrected statement of limits explains how I got 800. If some CPU cores are being set aside to feed the GPUs then you would subtract them from the total. So instead of 2000 tasks it would be 1950 for 1 core reserved, or 1900 for 2 reserved. When I look at my task list, I usually find one that says it's running on a partial core and another that says it's on the rest of the core and the GPU, but there doesn't seem to be anything reserved. David Sitting on my butt while others boldly go, Waiting for a message from a small furry creature from Alpha Centauri. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
If some CPU cores are being set aside to feed the GPUs then you would subtract them from the total. So instead of 2000 tasks it would be 1950 for 1 core reserved, or 1900 for 2 reserved. BOINC doesn't reserve anything itself. You would manually tell BOINC to not use one or more of your CPU cores. You would do this by setting the "On multiprocessors, use at most x% of the processors" to a lower value such as 75% to leave one free core on a 4 core system. Some of the big GPU guys have found that to work better for them. What you see in the GUI "Running (0.04 CPUs + 1.00 NVIDIA GPUs)" just refers to how the task is treated for the scheduler. Not that it is using, in my example, 4% of the CPU to process the task. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
David S Send message Joined: 4 Oct 99 Posts: 18352 Credit: 27,761,924 RAC: 12 |
BOINC doesn't reserve anything itself. You would manually tell BOINC to not use one or more of your CPU cores. You would do this by setting the "On multiprocessors, use at most x% of the processors" to a lower value such as 75% to leave one free core on a 4 core system. Some of the big GPU guys have found that to work better for them. Oh yeah, now that you desceribe it, I've seen that in the settings. I'm pretty sure I have it at 100%. David Sitting on my butt while others boldly go, Waiting for a message from a small furry creature from Alpha Centauri. |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Good news - things seem to have fixed themselves (though why they broke is still up for dispute) - both machines are now near 2000 tasks in progress. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Good news - things seem to have fixed themselves (though why they broke is still up for dispute) - both machines are now near 2000 tasks in progress. Did you ever catch the exact message you were getting? I'm still guessing that it might have been the daily limit. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
©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.