Message boards :
Number crunching :
AMD ATI work Units finished after 0.018% work finished? That can't be right?
Message board moderation
Author | Message |
---|---|
Norm Hill Send message Joined: 6 Sep 99 Posts: 9 Credit: 10,396,539 RAC: 0 |
All, I need a little help here. I have a Intel i7-980x (6 cores, 12 threads)with 18GB RAM with an AMD Radeon HD 5800 Series video card with the latest drivers. The other day I was watching the BOINC Manager and noticed that the GPU thread was spinning through work. It turns out that it starts running a work unit with 0.403 CPU's and 1 ATI GPU. It says the remaining time is about 2:40:00 and it will go for around 20-25 secs, get up to 0.015 - 0.020% and then say it is complete and uploads the result of the work unit. And it can blow through 40-50 units an hour. That can't be right. Does anyone have any idea what is going on and how we can make the GPU work and not just failing (that is what I suspect is happening) the work units? BTW, all non-GPU packets process in the normal expect time frames of between 2 and 30 hours. Any thoughts would be greatly appreciated. Kind regards, Norm Hill |
Bernie Vine Send message Joined: 26 May 99 Posts: 9954 Credit: 103,452,613 RAC: 328 |
If you want to check what your results are doing click on "ACCOUNT" in the top bar, that will take you to your account page, about half way down you will see "Tasks" with a clickable "View" link, if you click this you will see your GPU has 300+ "Invalid" tasks. I am not an expert on AMD GPU, hopefully someone will be along to help you out. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
First try restarting the PC, then try suspending GPU usage, deleting the compilations the app made, similar to (depends on your app version, CPU model and GPU model): MultiBeam_Kernels_r1843.clHD5_Capeverde.bin_V7 The compilations will be redone when the app starts again. Claggy |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
But since the AMD OpenCL AP app is also producing inconclusives/invalids I'd try downgrading to the last driver, in combination with deleting the compilations of both the MB apps, and the AP apps, otherwise it'll likely be a weak PSU or a malfunctioning GPU, amongst other possible causes. Claggy |
Norm Hill Send message Joined: 6 Sep 99 Posts: 9 Credit: 10,396,539 RAC: 0 |
I tried all the suggestions but to of no avail. All GPU bound packets bomb out after less than a full percentage point completed. Any ideas? |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Check GPU memory. Maybe downclock GPU memory/GPU shaders. Looks like hardware malfunction. Also, 9 December you had many crashes on GPU binary generation. That's imply broken driver or some issues with system (not GPU but CPU) memory. Try to check system memory too, try to run other GPU projects, check clinfo output, repair driver installation. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Wiggo Send message Joined: 24 Jan 00 Posts: 34744 Credit: 261,360,520 RAC: 489 |
Check to make sure that the card is not getting clogged up with dust and do you have the fan speed turned up? Cheers. |
Mr. Kevvy Send message Joined: 15 May 99 Posts: 3776 Credit: 1,114,826,392 RAC: 3,319 |
What's your PSU like? I just had the same problem with a GTX480 which was working perfectly fine on a Corsair 750 but would do the same thing when I moved it to a different system with a Corsair 650. Because the card is verified fine, everything is up to date, and the problem went away when I dialed everything down to minimum (GPU and memory clocks as low as possible) it looks like the PSU... it's possible if it can't put out the power the GPU requires it may go flaky and error out. |
Norm Hill Send message Joined: 6 Sep 99 Posts: 9 Credit: 10,396,539 RAC: 0 |
It "used" to work fine until the last few revisions of the BOINC/Seti Software. Hopefully the folks (Matt et.al) might peruse this forum and see this and have a fix for it. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
|
Dimly Lit Lightbulb 😀 Send message Joined: 30 Aug 08 Posts: 15399 Credit: 7,423,413 RAC: 1 |
OP you are streaming invades. A polite PM to the OP informing them of their situation task wise would be far more effective than dragging a nearly two month old thread back to the front page. Member of the People Encouraging Niceness In Society club. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
It was a knee jerk reaction after checking my "Validation inconclusive" WUs. PMs are creepy anyway. |
Bernie Vine Send message Joined: 26 May 99 Posts: 9954 Credit: 103,452,613 RAC: 328 |
Actually I find a post with someone asking "Are you streaming invalids" BEFORE trying to contact me would be a lot more "creepy" If it was me I would want to know about it via PM BEFORE someone resorted to posting the fact that I was having a problem for everyone to see. |
Wiggo Send message Joined: 24 Jan 00 Posts: 34744 Credit: 261,360,520 RAC: 489 |
But then a lot do not respond to PM's (or anything else) at all I've found out over the years. :-( Cheers. |
Bernie Vine Send message Joined: 26 May 99 Posts: 9954 Credit: 103,452,613 RAC: 328 |
Indeed, but it doesn't hurt to try a PM first. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
Isn't the OP still having the same problem he started this thread about? I don't know what the problem is so I couldn't help with a PM. |
©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.