Message boards :
Number crunching :
AP validaters working???? (RED ALERT)
Message board moderation
Previous · 1 · 2 · 3
Author | Message |
---|---|
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Assimilator queue is currently at 38,245. But the main thing is that validation is keeping up with the inbound. So, crunch yer APs off....LOL. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Yeah Sten, I was looking at the wrong numbers. The same thing was pointed out to me on the Lunatics forum. I was looking at the CPU time instead of the run time. Why he took so much more CPU time than I did is something else that has to be looked at. Your CPU times a re way to high Sunny. I have a 5850 running and usually between 1300 and 2400 seconds CPU time. Try 11.2 or 11.3 drivers to see if it changes something. Did you downclock your card its running at 700 MHZ ? With each crime and every kindness we birth our future. |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Your CPU times a re way to high Sunny. i'm currently using the 11.3 drivers. also, i have no idea why the Stderr outputs of my validated tasks are reporting a 700mhz GPU clock frequency, as i've never downclocked the card. its been running without a problem at the factory clock of 850mhz - and i know its not dropping down to 700mhz while crunching S@H AP tasks b/c i monitor the GPU w/ MSI Afterburner. i just thought of something - while i split my GPU's resources between 2 projects (sometimes MW@H, and sometimes S@H, though never simultaneously) on the GPU, i'm always crunching E@H on the CPU. when i crunch MW@H on the GPU, i'm able to run E@H on all 6 CPU cores without a substantial increase in the run times of either the MW@H GPU tasks or the E@H CPU tasks (despite the facts that i'm using 100% of the CPU for E@H, and that each MW@H GPU task still requires 5% of a CPU core). yet when i crunch S@H AP tasks on the GPU (each of which only requires 4% of a CPU core), my E@H CPU task run times go up substantially. while crunching MW@H & E@H at the same time, my E@H run times are in the 4.5-hour neighborhood. while crunching S@H and E@H at the same time, my E@H run times are closer to 8 hours. if i set BOINC to use only 85% of the CPU, E@H will crunch on only 5 CPU cores, leaving one free for other instructions & calculations. this helps my E@H run times by bringing them down to ~6.5 hours, but its still nowhere near as low as they are when i crunch MW@H on the GPU...and yet both MW@H tasks and S@H AP tasks use just about the same amount of CPU power. but while i know E@H CPU task run times are affected by concurrently running either MW@H or S@H AP on the GPU, i've never really paid attention to the CPU run times of my S@H AP GPU tasks, nor have i monitored them to see how they vary as i manipulate the number of CPU cores working on E@H. perhaps i need to do that to see if something is going on. or perhaps i should just suspend E@H for a bit, just run S@H AP on the GPU, and see if my CPU run times come down any... another thing i thought of is the fact that i'm running Windows XP Pro SP3 32-bit, which only recognizes ~3.25GB of the 4GB of system memory i have installed...perhaps that has something to do with high CPU run times, especially considering that the CPU is working on 2 different DC projects at any given time on my machine... |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Update on the update - 45,590 I have notice 4-6 keep going offline, but fix_ap_assimilator1keps running. Maybe they worked out some automated process to clear the problematic tasks? SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Your CPU times a re way to high Sunny.i just thought of something - while i split my GPU's resources between 2 projects (sometimes MW@H, and sometimes S@H, though never simultaneously) on the GPU, i'm always crunching E@H on the CPU. when i crunch MW@H on the GPU, i'm able to run E@H on all 6 CPU cores without a substantial increase in the run times of either the MW@H GPU tasks or the E@H CPU tasks (despite the facts that i'm using 100% of the CPU for E@H, and that each MW@H GPU task still requires 5% of a CPU core). yet when i crunch S@H AP tasks on the GPU (each of which only requires 4% of a CPU core), my E@H CPU task run times go up substantially. while crunching MW@H & E@H at the same time, my E@H run times are in the 4.5-hour neighborhood. while crunching S@H and E@H at the same time, my E@H run times are closer to 8 hours. if i set BOINC to use only 85% of the CPU, E@H will crunch on only 5 CPU cores, leaving one free for other instructions & calculations. this helps my E@H run times by bringing them down to ~6.5 hours, but its still nowhere near as low as they are when i crunch MW@H on the GPU...and yet both MW@H tasks and S@H AP tasks use just about the same amount of CPU power. I also find E@H CPU tasks runtimes are affected by OpenCL AP running, certainly with the ATI variant, not noticed if the Nvidia variant affects it too, the x38 app was also effected by ATI OpenCL AP running, which is why it ended up with a Normal Priority, since reversed with the x39 app, Claggy |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Just a bit of FYI... The x38g app is a bit faster than the x39e, which has some troubleshooting thingys built into it that slow it down just a tad. You can run the x38g app and use Fred's priority tool to force it to below normal priority. I am doing this on my daily driver, and it has eliminated the stalling I was getting when x38e was running at normal priority (which it does out of the box). "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
hey everyone, i don't want to hijack the thread anymore than it has been (regarding the unusually long CPU run times associated with S@H AP GPU tasks that some of us are seeing), so i'm gonna start a new thread soon dedicated to the topic. this morning i suspended E@H, leaving S@H AP as the only project running at the moment. i'm still running the older Lunatics v0.37 installer, but i want to see first if running E@H and S@H AP concurrently had anything to do with the unusually long CPU run times i was getting for both projects. i'll let some S@H AP tasks run today and study the results. then i'll post the info in the new thread and provide a link to it here so that any interested parties don't have to go searching for it... ...ok, back on topic regarding the validators ;-) |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
i'm still running the older Lunatics v0.37 installer, Saying you're running the older 0.37 Lunatics Installer doesn't make sense, all the Installer does is install the apps selected, and the only two apps that you are running weren't in the 0.37 installer, and had to be installed manually, (ATI MB r177_HD5 and ATI AP r521) Claggy |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
i'm still running the older Lunatics v0.37 installer, Thats correct Sten. He has onboard graphics activated also. I watched his results 5870 is running. With each crime and every kindness we birth our future. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
No usually the first device is shown. On discrete GPUs all are shown. Also i dont think its important so long it works. With each crime and every kindness we birth our future. |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Saying you're running the older 0.37 Lunatics Installer doesn't make sense, all the Installer does is install the apps selected, yes, sorry for the confusion Claggy. you're right, and i think its all starting to come back to me now...you see, some time ago i had installed v0.37 to run the optimized S@H CPU apps. then i got the 5870, and i must have manually installed the apps for it (ATI MB r177_HD5 and ATI AP r521), and removed the CPU apps that came w/ the v0.37 package. obviously my memory is horrendous LOL...and moreover, i see now that, having installed v0.37 at some point in time and having since removed it has nothing to do with my current dilemma. thank you for providing me with that moment of clarity, or else i might have wasted even more time exploring the possibility that not having run the v0.38 installer yet might be part of my "high CPU run times" problem... |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
thank you for providing me with that moment of clarity, or else i might have wasted even more time exploring the possibility that not having run the v0.38 installer yet might be part of my "high CPU run times" problem... Have you tried disabling the onboard GPU incase that's taking resources? then there Cat 11.2/SDK2.3 to try, I'm running Cat 11.6/SDK2.4 at the moment, will be soon downgrading to Cat 11.2/SDK2.3 soon to test something, Claggy |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Have you tried disabling the onboard GPU incase that's taking resources? then there Cat 11.2/SDK2.3 to try, i'll give those things a try when i get a chance and report back in the new thread i plan on starting. |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
ok, i started a new thread regarding unusually high CPU run times. you can find it here: http://setiathome.berkeley.edu/forum_thread.php?id=64845 |
©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.