Message boards :
Number crunching :
blanked AP tasks
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
TBar, The values in the stderr_txt are as they say "Max clock frequency:" & not the devices current running clock rate. You can see this same information by running clinfo from the command line. The HD6570 showing 1000MHz is a bit weird, but the manufacture of the card probably just typed that in the max clock field. I would also expect your HD6570 to run a fair bit faster than ~15 hours. As my HD6370m runs AP tasks under 6 hours with no tuning options. Another way to reserve the CPU for your GPU would be with an app_config.xml For my cards where I run 1 AP at a time I use. <app_config> <app> <name>astropulse_v6</name> <gpu_versions> <gpu_usage>1.0</gpu_usage> <cpu_usage>1.0</cpu_usage> </gpu_versions> </app> </app_config> This way when no AP are running on the GPU all of my CPU cores are free to crunch. Also my HD5750 runs about 1 hr 20 min for AP tasks. So 3+ hours for a 7770 does seem high. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
TBar, Ah... 3 at a time? On a Turks? Well, your times are still slow, even for 3 at a time. Why someone would run 3 APs at a time on a low end card is beyond me. All you're doing is creating Data Swap and increasing Run-Times. You can use the Command Line settings I gave to increase GPU load and avoid Data Swap. Running 6 GPU APs would call for 6 Free CPU cores instead of 1 or 2. Don't know, OpenCL says your Turks is set at 1000. Have you restarted lately? My GDDR5 7750 completes an AP in under a hour, the 7770 should be faster. The Turks should be about the same as my DDR3 7750 which completes an AP in 2 hours. I'm running 1 task at a time with Command Line settings. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I run 3 at a time on the 6570 because I need 3 at a time on my 7770 so my app_config is something like (in abbreviated form): app gpu_versions gpu .33 cpu .1 This was how I had been told to do it. That I couldn't separately ask for 3 on one and 2 on the other. Confused a bit now????? --edit-- My machine we are talking about is shutdown right now. --edit-- I'm a tired old-timer, have to hit the hay now. I'll check in the am. Thanks much for your help. The command line thing might be what I need. |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
Sorry to but in guys but i have 1 question which i think i all ready know . I have noticed i'm not getting good GPU utilstion on the GPU and the tasks have been running a bit long so i'm assuming they must be highly blanket so my question is if they are the utilisation will go down on those units right ? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Here is the AP ReadMe from the latest Installer; App instances. The 7770 would be at the low end of the 'mid range cards'. I have Three 6850/6870s and haven't noticed any advantage in running 2 instances. Not only that, but, with a Quad core CPU running 6 Free CPU cores would be rather difficult. I do run with an unroll number of 16 though. Yes, the current GPU blanked tasks are run on the CPU leaving the GPUs to twiddle their thumbs. Best to toss them to the CPUs and let the GPUs continue with unblanked tasks. That is technical, and with AP_V7 just around the corner best left alone. |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
Yes, the current GPU blanked tasks are run on the CPU leaving the GPUs to twiddle their thumb Thank Tbar I thought that was what i was seeing i only asked because of the prob's i'm experiencing with the AMD rig and my brain is getting a bit lost with all the info i'm reading so thanks again it was a stupid question i spose seeing as i knew that was prob what i am seeing highly blanked units I'll crawl back down my hole now |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Here is the AP ReadMe from the latest Installer; One task at a time normally runs around 80%-90% on my 6870. I did get excited once when I saw it running around 30% with some tuning values, but that just turned out to be a highly blanked task. I decided it would either require borrowing Rasitmers brain foe a little while or a x900 series card to effectively run more than 1 at a time on my 6870. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Here is the AP ReadMe from the latest Installer; I was able to get my 6870 up to ~95% GPU usage using an unroll of 18. It seems to top out around there, further increases didn't seem to raise the usage. Here it is in Vista, http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=72451 Here it is in Ubuntu, http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=72013&offset=20&show_names=0&state=0&appid= If you can get it to 95% load, why even bother with running more instances? |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Here is the AP ReadMe from the latest Installer; I will see usage around 95-99% with unroll of 16, but it depends on the work running. I may even back it back down to 14. As I have not seen much difference in usage between 14 and 16. With anything over 80% usage I'm not sure running more than 1 instance is actually a net benefit. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
TBar, re. your comments: 1) Make sure the cards are not overheating, the Turks card has errors associated with overheating. 2) Update to the latest AMD Beta driver. I've have issues on different machines with 1348 & 1445. 14.6/14.7 works better on my machines. 3) Find the ap_cmdline_win_x86_SSE2_OpenCL_ATI.txt file in your setiathome.berkeley.edu folder and paste the following line in the file; -unroll 10 -ffa_block 4096 -ffa_block_fetch 2048 -sbs 256 What does this "-unroll 10 -ffa_block 4096 -ffa_block_fetch 2048 -sbs 256" do? |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
TBar, A description of what these functions do can be found in the ReadMe_AstroPulse_OpenCL_ATI.txt I think a basic description would be they are used change how data is loaded when processing. Which can be more effective than the default values the program uses for all cards to function. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
TBar, -unroll N :sets number of data chunks processed per kernel call in main application loop. N should be integer number, minimal possible value is 2. -sbs not necessary with AP thats done with unroll. -tune is more effective. With each crime and every kindness we birth our future. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
Thanks for all these comments. I have another problem coming up. I'm adding a r7 265 to this same machine which is going to further complicate things for me. I have a lot of studying to do. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Thanks for all these comments. I have another problem coming up. I'm adding a r7 265 to this same machine which is going to further complicate things for me. Interesting choice. I would have gone with the R9 270 myself. As it is the same power usage & about 25% faster. If you add a card like that I would remove the 6570. Having a card that much slower than two faster ones may throw off the estimates. Which will could cause your system to start throwing out errors instead of getting any work done. The general idea is to get as much done as possible, but you seem hell bent on going the other way. :P SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
Hal, I am just inexpereienced and over my head. I'm just trying my best. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Hal, I'm not sure most of us really know what we are doing. I know nothing & I can prove it. Just ask! SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I am just inexpereienced and over my head. I'm just trying my best. So... How are you making out? I was hoping to see your cards ripping through those choice APs by now. You have quite a few low-blanked APs there that will start expiring in a couple days. For now you can just alter the app_config file to run one at a time with 1 CPU core reserved. Hal's example will work for that; <app_config> <app> <name>astropulse_v6</name> <gpu_versions> <gpu_usage>1.0</gpu_usage> <cpu_usage>1.0</cpu_usage> </gpu_versions> </app> </app_config> Then add the Command Line setting to increase the GPU usage while just running one task at a time; 3) Find the ap_cmdline_win_x86_SSE2_OpenCL_ATI.txt file in your setiathome.berkeley.edu folder and paste the following line in the file; -unroll 10 -ffa_block 4096 -ffa_block_fetch 2048 -sbs 256 See how that works for now. |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
I still run BOINC 7.2.42 (64BIT) in a VM. 1 AP is stuck at 79.801% after 34:09 h. What's best todo ? Just wait or delete it? Also planning to run optimized? |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
1 AP is stuck at 79.801% after 34:09 h. How long has it been stuck at 79.801% if it been a few hours and hasn't moved 1) suspend set@home 2) shut Bonic down 3) restart VM 4) start Bonic 5) resume seti Then see what happens it mite not be Bonic or seti units it mite be a problem with the VM so start there and reboot it . If it's still stuck after a 30 mins abort it Edit Lunatics is faster worth a try |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Can anybody tell me.... Does the new .42 Lunatics AP GPU app for NV handle heavily blanked AP work more poorly than the last .41 version? Or are we just getting more heavily blanked tasks recently? Or has the blanking in the WUs changed? Just trying to get a handle on it. Meow? "Freedom is just Chaos, with better lighting." Alan Dean Foster |
©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.