Message boards :
Number crunching :
ATI OpenCL MultiBeam 6.10 problem..
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 . . . 7 · Next
Author | Message |
---|---|
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
do you uninstall your drivers before you install the new ones. I know its not necessary for everything but the video drivers definitely need to be removed before you can update them thats the point I was getting at In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Had you also tried the ATI uninstaller that cleans out any and everything ATI from the system & then reinstalling the driver with CCC? yes, i tried this as well, but to no avail. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
just by chance have you ran a malware scan and an AV scan lately. If the CCC is failing to install properly that might be a problem. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Jamie Send message Joined: 5 Apr 06 Posts: 162 Credit: 9,867,955 RAC: 0 |
Had the same issue recently Turns out that when CCC is being installed it tries to install to "C:\Program Files\ATI Technologies\". Only way I could get CCC installed correctly was to change the install path to be "C:\Program Files\ATI" Apparently this is an old bug that has been re-introduced in a couple of the newer versions |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
just by chance have you ran a malware scan and an AV scan lately. If the CCC is failing to install properly that might be a problem. i suppose i could give this a shot b/c i haven't scanned for malware in a while. i may just do it to make sure i don't have anything bad on the system, but i'm not sure if i'm going to try to install CCC afterwards b/c MSI Afterburner is now fulfilling all of my hardware monitoring and OCing needs. Had the same issue recently interesting to say the least...do you have any idea why a slight change in the name/location of the install directory would fix this problem? *EDIT* - by the way, the 5870 has been running flawlessly for the past 12 hours straight, crunching 2 MW@H tasks simultaneously in approx. 2:50 (so approx. 1:25 per task). GPU temps are steadily oscillating between 57°C and 60°C depending on the rise and fall of the ambient temps in my condo over the course of the day. initially on auto, the 5870's fan would only throttle up to ~33% or so, and temps would go as high as 77°C. at first i had no idea if this was safe or not, but then i stumbled across a post of yours in another thread Skildude, and realized that the 5870 really shouldn't be going above 60°C. when set to auto, the fan is at 20% when the card is idle, and i would say that it is audible, but not offensive by any means - it hardly made my system louder than it was before. when the auto function throttles the fan up to 33% under 100% GPU load, i'd say its very audible, but not necessarily loud. in order to maintain a max. of 60°C under 100% GPU load i have to manually set the fan to approx. 50%, at which point it sounds like a small vacuum cleaner...thank god this is my home office PC and not the one in my master bedroom, or else i probably wouldn't be GPU crunching at night. i know you keep your fan at 75% Skildude, but i haven't even attempted to see what the fan sounds like at anything over 50% yet. i suppose during the day when its warmer out (and consequently so is the ambient temp in my condo despite having the AC set to 73°F all the time), i could throttle the fan up higher than 50% since i'm at work and can't hear it. but is the general consensus to keep my GPU at or below 60°C? TIA, Eric |
Jamie Send message Joined: 5 Apr 06 Posts: 162 Credit: 9,867,955 RAC: 0 |
Not a clue, when installing I would either get a error saying "Installer package failure" or it would install but would never launch. Probably something to do with rights on that folder but didn't investigate too far after I got it working |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Not a clue, when installing I would either get a error saying "Installer package failure" or it would install but would never launch. Probably something to do with rights on that folder but didn't investigate too far after I got it working well if i care enough to get CCC working (which i don't really need anymore now that i have MSI Afterburner) i'll get this method a try. its interesting b/c those are the exact issues i ran into - "installer package failure" and "install package success" with no working CCC. i also got a third result during installation - "Warnings occured during installation." but upon completion i'd look at the installation log, and it would say that all components installed successfully (even though CCC still wouldn't work). |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
interesting to say the least...do you have any idea why a slight change in the name/location of the install directory would fix this problem? I'd guess it's the blank space in the name. If it doesn't get quoted correctly in all scripts, the result is "unpredictable". Gruß, Gundolf |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
ok everyone, i'll steer the thread back on topic again by posting some questions now that i have S@H AP 5.06 installed for the GPU. i decided not to run MB tasks after finding out that they're nothing more than S@H Enhanced CPU tasks optimized to run on the GPU instead. at any rate, i'm pretty sure i have all the right settings, but i'm not getting tasks from the server. i've checked the server status and it appears that there are plenty of AP tasks waiting to be sent out. perhaps something's amiss in my app_info.xml file? i'll post it up to see if anyone catches any mistakes, minor or glaring: <app_info> ...and here are my preferences just in case something isn't right there either... Computing preferences: SETI@Home preferences: any suggestions? TIA, Eric |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
you may end up having an idle GPU if you ignore the MB tasks. AP WU's are produced at a rate of 1 per 32 MB WU's. Which means you'll get 1 AP for every 32 MB WU's created. You do understand the AP and MB look for different things. In fact running MB on the GPU is much faster than on the CPU. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Your app_info looks fine. Just because there's 4,487 Astropulse tasks available doesn't mean they are available, the feeder only holds 100 tasks at a time, some will be MB, some be resends, those AP tasks need to filter down to the front of the queue before the feeder sends them out, to that end i recommend you set your 'Maintain enough work for an additional' setting to a couple of days, say 4 or 5, otherwise you'll only get ATI AP work infrequently, now you're just got to wait, if you see spikes in the cricket graph, those are probably AP tasks going out, Graphs for gigabitethernet2_3 Claggy |
Ethelred Send message Joined: 8 Jan 11 Posts: 4 Credit: 224,172 RAC: 0 |
the MB 6.10 are for ati and Fermi cards. You apparently have a pair of 6XXX series cards on board. Thank you for the information that MB 6.10 are for ati and Fermi cards. Yesterday i replaced my old Geforce GTS250 with the brand new (Gigabyte)Geforce 560 Ti card and for some reason my system now choose to dowload MB 6.10. As a result all my tasks are crashing and get the "Completed, validation inconclusive" message. Panic, i had to close down SETI@Home! Does anyone have an idea how to fix this problem? //Ethelred (team Sweden) |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
you may end up having an idle GPU if you ignore the MB tasks. AP WU's are produced at a rate of 1 per 32 MB WU's. Which means you'll get 1 AP for every 32 MB WU's created. You do understand the AP and MB look for different things. In fact running MB on the GPU is much faster than on the CPU. i guess i should have explained why i prefer AP tasks over classic/Enhanced/MB tasks when i first mentioned it earlier in the thread. i understand that, while AP and MB search more or less "search" for the same thing, they actually "look" for different kinds of signals. i prefer Astropulse b/c it not only has a chance of detecting possible signals of intelligent origin, but it also has the chance of detecting yet-to-be-discovered pulsars, black holes, and other astrophysical phenomena...whereas as Multibeam only looks for the traditional "little green men." don't get me wrong, i believe in the cause just as much as any other enthusiast here...its just that my primary interests in the cosmos are more physics/mathematics/relativity/quantum theory-related than they are related to the search for extraterrestrial intelligence. it is also the reason why i feel Einstein@Home is worth more of my computing time than any other DC project. i know its just an opinion, but those particular projects/applications are the ones that stand to benefit my personal favorite scientific fields of research the most. that being said, i think i'll go ahead and install Multibeam temporarily just to do some GPU testing, say, until i see an AP task or two come down the que. if i'm truly not getting work from the server right now b/c the AP tasks are that rare, then i'll ultimately do as Claggy suggested and cache 4-5 day's worth of AP tasks. |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Thank you for the information that MB 6.10 are for ati and Fermi cards. i'm having a similar problem now that i have the Multibeam 6.10 GPU app up and running for testing purposes. i wish i could help you, but i only know the ATI side of things when it comes to GPUs. so long as my 5870 GPU only runs 1 S@H MB task at a time, they don't crash and error out. however, it apparently wants to run 2 tasks at a time, and so one task after another is erroring out as it tries to add a 2nd task to the workload. the first task to start crunching goes from start to finish without error. its just the subsequent 2nd task that immediately errors out upon beginning to crunch simultaneously with the first task. if i don't suspend the S@H project, errored tasks start to cascade down the page at the pace of one every other second! i figured it might have to do with my app_info.xml file, so please take a look at my MB 6.10 section and tell me if any of you see anything out of the ordinary: <app> TIA, Eric |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Your ATI MB Section looks fine, but what GPU's are fitted to that host? Is it two HD5800's? or one HD5800 and a HD 2300/2400/3200? because you're erroring out Wu's, if you're still got the HD 2300/2400/3200 fitted you'll have eithier disable Boinc's use of it or remove it, as it isn't OpenCL capable, you can do this by making a cc_config.xml file (with Notepad) with the following in it: <cc_config> <options> <ignore_ati_dev>n</ignore_ati_dev> </options> </cc_config> where 'n' is for the ATI device number, should be 1 in your case, drop the file in your Boinc Data directory, and restart Boinc, this will disable Boinc's use of the 2nd ATI GPU, so you might want to put it in the other host. Claggy |
Sunny129 Send message Joined: 7 Nov 00 Posts: 190 Credit: 3,163,755 RAC: 0 |
Your ATI MB Section looks fine, but what GPU's are fitted to that host? the host is fitted with one HD 5870 (discrete) and one HD 3300 (integrated video). both are enabled b/c i use the 5870 to crunch tasks, while the 3300 is used for my actual display. however if you look at my computer details (http://setiathome.berkeley.edu/show_host_detail.php?hostid=5800349), you'll see that BOINC/the server side does not recognize the 3300 (which is good), but it seems to think there are 2 5870's in my rig. at any rate, i'll try what you suggested. i'm assuming that you suggested the device number N = 1 b/c it is the 2nd GPU (and that N = 0 is actually the main GPU)? |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
The tasks that are erroring show: <core_client_version>6.10.58</core_client_version> While the tasks that run O.K show: <core_client_version>6.10.58</core_client_version> I'm assuming because the HD5870 shows up first that it is device 0 (of all Cal devices), Claggy |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
It may be trying to run the work in "hybrid crossfire mode" I'd disable the 3200 and just use the 58XX to run work and use to browse. My 6970 was running into this on my backup machine. BOINC was trying to run off the onboard 4200 in combination with the 6970. I finally had to disable and uninstall the 4200 drivers. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14676 Credit: 200,643,578 RAC: 874 |
the host is fitted with one HD 5870 (discrete) and one HD 3300 (integrated video). both are enabled b/c i use the 5870 to crunch tasks, while the 3300 is used for my actual display. however if you look at my computer details (http://setiathome.berkeley.edu/show_host_detail.php?hostid=5800349), you'll see that BOINC/the server side does not recognize the 3300 (which is good), but it seems to think there are 2 5870's in my rig. at any rate, i'll try what you suggested. i'm assuming that you suggested the device number N = 1 b/c it is the 2nd GPU (and that N = 0 is actually the main GPU)? The BOINC server database doesn't store proper records for each and every GPU card in our systems - would that it did. All you get in the host_detail display is the "best" GPU of each type, and a count of the total number. You should see much better information in your local message log at startup - the BOINC client does keep track of the different cards, it just turns into a summary when it gets to the server. Actually, BOINC shouldn't be using the secondary HD 3300 card at all, unless you'ce consciously put in a <use_all_gpus> directive. I haven't read the whole thread, but from what Claggy says, you'd be better off taking that back out again. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
It's an HD 3300 (integrated video), so he can't remove it, he can eithier just disable Boinc's use of it, and still use it as his Main display, (as i suggested) or he disables it totally in the Bios and uses the HD5870 for the Display instead, (as Skildude suggested) Claggy |
©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.