CUDA for Milkyway@Home

Author	Message
Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25013 - Posted: 11 Jun 2009 \| 18:17:36 UTC Last modified: 11 Jun 2009 \| 18:18:29 UTC
Because it looks like the server can handle a bit more work, I've uploaded the CUDA code for plain milkyway@home, and it's in the code release directory. Hopefully we'll be able to just use CUDA code for regular milkyway@home and then we wouldn't even need to start up milkywayGPU@home. I've also added two binaries, one single and one double precision for intel-based macs running OS X. These are also in the code release directory. If anyone wants to try and make some linux and windows binaries for people to use that would be appreciated. ____________
ID: 25013 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 25026 - Posted: 11 Jun 2009 \| 18:43:04 UTC - in response to Message 25013.
How will that work with the single precision cards? Won't the searches eventually reach a point where they won't be accurate enough anymore?
ID: 25026 \| Rating: 0 \| rate: / Reply Quote

ztmike Send message Joined: 4 Jun 09 Posts: 45 Credit: 447,355 RAC: 0	Message 25029 - Posted: 11 Jun 2009 \| 18:47:55 UTC
I noticed a problem in the binaries and code I put on the code download page. I removed them to do some fixes, but they should be available tonight or at the latest tomorrow morning. ^^ From the front page.. Does this mean when you put it back up, that we Cuda users can download the app and try crunching on that? The download link you posted http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Is a little confusing ..to say the least.
ID: 25029 \| Rating: 0 \| rate: / Reply Quote

Brian Silvers Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0	Message 25030 - Posted: 11 Jun 2009 \| 18:48:16 UTC - in response to Message 25013.
Hopefully we'll be able to just use CUDA code for regular milkyway@home and then we wouldn't even need to start up milkywayGPU@home. I'm very confused by this... I thought you were wanting to issue much larger tasks to GPUs so that you could do things that CPUs wouldn't be able to handle quickly enough? What happened with that? Also, unless you're seeing a relatively large stockpile of tasks waiting to be sent, and I don't think < 1000 qualifies as "large", then I think that there is a high probability that people will figure out that things are working better here now, bring more processing power over, and then we'll be back in the same situation... Make sure that you're planning for a longer time horizon than just a few weeks....
ID: 25030 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25033 - Posted: 11 Jun 2009 \| 18:49:42 UTC - in response to Message 25013. Last modified: 11 Jun 2009 \| 18:52:37 UTC
Because it looks like the server can handle a bit more work, I've uploaded the CUDA code for plain milkyway@home, and it's in the code release directory. Hopefully we'll be able to just use CUDA code for regular milkyway@home and then we wouldn't even need to start up milkywayGPU@home. I've also added two binaries, one single and one double precision for intel-based macs running OS X. These are also in the code release directory. If anyone wants to try and make some linux and windows binaries for people to use that would be appreciated. I would be ready. I can release a single precision ATI version today if you want. But be warned, SP is about a factor 4 or more faster than DP per WU! The server should be prepared for that combined rush (a lot of CUDA GPUs + lower end ATI cards adding to the higher end ATIs already in use). Have you thought about the credit issue? Just giving the single precision clients half that of the DP ones isn't going to cut it. The single precision variant actually does quite a bit less operations than the double precision one because all used functions have hardware implementations. For instance one doesn't have to do the quite expensive exp() in software (like in DP or also on CPUs [done in the math library]). All in all the number of really executed operations for the SP version (have to thoroughly count it, will tell you the result) is almost exactly halved. I think it would be best to use exactly the same credit multiplier as SETI which also GPUGrid converges to (2.72 credits per SP TeraFlop, here at MW the current value for DP is 7.5, one may think about lowering it to 5.44, i.e. twice the SP value). Together with the lower count of the operations for SP it will most likely result in about the same credit/time rate as when using DP on the same GPU (at least on ATI, for nvidias GTX2xx cards SP would yield most likely a higher credit rate than DP). But how do you differentiate between the SP and DP versions? By using the applications version string? Or should one introduce an additional SP marker to the out file?
ID: 25033 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25034 - Posted: 11 Jun 2009 \| 18:50:07 UTC - in response to Message 25029.
I noticed a problem in the binaries and code I put on the code download page. I removed them to do some fixes, but they should be available tonight or at the latest tomorrow morning. ^^ From the front page.. Does this mean when you put it back up, that we Cuda users can download the app and try crunching on that? The download link you posted http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Is a little confusing ..to say the least. Yeah, I'll have a binary for OS X that you can try and crunch with. ____________
ID: 25034 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25035 - Posted: 11 Jun 2009 \| 18:51:41 UTC - in response to Message 25033. Last modified: 11 Jun 2009 \| 18:55:34 UTC
I would be ready. I can release a single precision ATI version today if you want. But be warned, SP is about a factor 4 or more faster than DP per WU! The server should be prepared for that combined rush (a lot of CUDA GPUs + lower end ATI cards adding to the higher end ATIs already in use). Talking to my advisors it seems they don't want us to release a single precision application for GPUs for milkyway@home just yet. I think that will have to wait until we get milkyway GPU up and running. The single precision binaries I'm going to release are just for testing really (i'd like to see how many machines my code works on). Have you thought about the credit issue? Just giving the single precision clients half that of the DP ones isn't going to cut it. The single precision variant actually does quite a bit less operations than the double precision one because all used functions have hardware implementations. For instance one doesn't have to do the quite expensive exp() in software (like in DP or also on CPUs [done in the math library]). That almost exactly halves the amount of really executed operations for the SP version (have to thoroughly count it, will tell you the result). Yeah we're going to have to come up with a credit scheme for milkywaygpu that takes into account floating point and double precision apps (double precision getting more credit because their results are more accurate and useful. I think it would be best to use exactly the same credit multiplier as SETI which also GPUGrid converges to (2.72 credits per SP TeraFlop, here at MW the current value for DP is 7.5, one may think about lowering it to 5.44, i.e. twice the SP value). Together with the lower count of the operations for SP it will most likely result in about the same credit/time rate as when using DP on the same GPU (at least on ATI, for nvidias GTX2xx cards SP would yield most likely a higher credit rate than DP). But how do you differentiate between the SP and DP versions? By using the applications version string? Or should one introduce an additional SP marker to the out file? ____________
ID: 25035 \| Rating: 0 \| rate: / Reply Quote

ztmike Send message Joined: 4 Jun 09 Posts: 45 Credit: 447,355 RAC: 0	Message 25036 - Posted: 11 Jun 2009 \| 18:55:25 UTC - in response to Message 25034.
I noticed a problem in the binaries and code I put on the code download page. I removed them to do some fixes, but they should be available tonight or at the latest tomorrow morning. ^^ From the front page.. Does this mean when you put it back up, that we Cuda users can download the app and try crunching on that? The download link you posted http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Is a little confusing ..to say the least. Yeah, I'll have a binary for OS X that you can try and crunch with. So no support for Windows yet? All my machines run Windows.. :(
ID: 25036 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25037 - Posted: 11 Jun 2009 \| 18:55:26 UTC - in response to Message 25035.
Talking to my advisors it seems they don't want us to release a single precision application for GPUs just yet. I'm probably not going to release one for CUDA either (i've just been using single precision because thats what my GPU supports). We've ordered some double precision nvidia GPUs to play around with so once we get those it'll probably be double precision from here on out. So no chance for any GPU older than the GTX2xx line of nvidia or the mainstream series of ATI (besides the HD4700 series which supports DP)?
ID: 25037 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25038 - Posted: 11 Jun 2009 \| 18:56:19 UTC - in response to Message 25037.
Talking to my advisors it seems they don't want us to release a single precision application for GPUs just yet. I'm probably not going to release one for CUDA either (i've just been using single precision because thats what my GPU supports). We've ordered some double precision nvidia GPUs to play around with so once we get those it'll probably be double precision from here on out. So no chance for any GPU older than the GTX2xx line of nvidia or the mainstream series of ATI (besides the HD4700 series which supports DP)? I edited my post :) We're going to have floating point on milkywayGPU whenever I get the backend up and running, but not here for the time being. ____________
ID: 25038 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25039 - Posted: 11 Jun 2009 \| 18:56:20 UTC - in response to Message 25037.
Talking to my advisors it seems they don't want us to release a single precision application for GPUs just yet. I'm probably not going to release one for CUDA either (i've just been using single precision because thats what my GPU supports). We've ordered some double precision nvidia GPUs to play around with so once we get those it'll probably be double precision from here on out. So no chance for any GPU older than the GTX2xx line of nvidia or the mainstream series of ATI (besides the HD4700 series which supports DP)? I edited my post :) We're going to have floating point on milkywayGPU whenever I get the backend up and running, but not here for the time being. ____________
ID: 25039 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25042 - Posted: 11 Jun 2009 \| 18:57:36 UTC - in response to Message 25036.
I noticed a problem in the binaries and code I put on the code download page. I removed them to do some fixes, but they should be available tonight or at the latest tomorrow morning. ^^ From the front page.. Does this mean when you put it back up, that we Cuda users can download the app and try crunching on that? The download link you posted http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Is a little confusing ..to say the least. Yeah, I'll have a binary for OS X that you can try and crunch with. So no support for Windows yet? All my machines run Windows.. :( I don't have a windows box to compile and test it on, so no. Another student who should be taking over most of the GPU work, Anthony, does I think. So we should have those when he gets up to speed (or when one of our users releases one). ____________
ID: 25042 \| Rating: 0 \| rate: / Reply Quote

borandi Send message Joined: 21 Feb 09 Posts: 180 Credit: 26,221,261 RAC: 0	Message 25045 - Posted: 11 Jun 2009 \| 19:05:33 UTC - in response to Message 25034.
I noticed a problem in the binaries and code I put on the code download page. I removed them to do some fixes, but they should be available tonight or at the latest tomorrow morning. ^^ From the front page.. Does this mean when you put it back up, that we Cuda users can download the app and try crunching on that? The download link you posted http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Is a little confusing ..to say the least. Yeah, I'll have a binary for OS X that you can try and crunch with. Travis I think you misunderstood ztmike here. The CUDA code release can only be tested on the test WUs ztmike, not for the main project itself (yet). Also Travis, if you are going to deploy the CUDA DP at normal MW, will this not require a BOINC upgrade server side? This would be so people can select 'Use CUDA app' in their preferences, akin to SETI, rather just randomly issue CUDA Wus to CUDA machines, like in Aqua@Home. It would be interesting if MW_GPU changed its name to MW_SP :) Such that the older cards (and some of the new cards) could go over there. ____________
ID: 25045 \| Rating: 0 \| rate: / Reply Quote

Phil Rosa-Leeke Send message Joined: 30 Mar 09 Posts: 1 Credit: 385,419 RAC: 0	Message 25160 - Posted: 12 Jun 2009 \| 15:59:03 UTC - in response to Message 25045. Last modified: 12 Jun 2009 \| 16:01:51 UTC
Hi there, Right, first things first, I'm no computer buff so all this tech talk to me is gibberish. All I want to know is if the up-coming CUDA thingy will work on my computer. What I have is the following: Hardware Overview: Model Name: iMac Model Identifier: iMac7,1 Processor Name: Intel Core 2 Duo Processor Speed: 2.4 GHz Number Of Processors: 1 Total Number Of Cores: 2 L2 Cache: 4 MB Memory: 1 GB Bus Speed: 800 MHz ATI Radeon HD 2600 Pro: Chipset Model: ATI,RadeonHD2600 Type: Display Bus: PCIe PCIe Lane Width: x16 VRAM (Total): 256 MB Vendor: ATI (0x1002) Device ID: 0x9583 Revision ID: 0x0000 ROM Revision: 113-B2250F-212 EFI Driver Version: 01.00.212 Displays: iMac: Resolution: 1680 x 1050 Depth: 32-Bit Color Core Image: Hardware Accelerated Main Display: Yes Mirror: Off Online: Yes Quartz Extreme: Supported Built-In: Yes Display Connector: Status: No Display Connected Just a simple Yes or No will do thanks. I'm running the 6.2.18 version of BOINC and it's pushing out 1280+ Avg. Work Units at the mo. If the new fandangled CUDA will work on my system, will it be up to doing more or will I be better off leaving the BOINC running and forgetting the CUDA? All the best and keep up the good work, Phil PORTUGAL ____________
ID: 25160 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25162 - Posted: 12 Jun 2009 \| 16:05:47 UTC - in response to Message 25160. Last modified: 12 Jun 2009 \| 16:06:41 UTC
CUDA is for NVidia graphics cards. You have an ATI graphics card. There might already be a GPU application for you (unless it doesn't support double precision). When OpenCL comes out, your card should be supported by that. But I don't think OpenCL comes out until Apple releases Snow Leopard. ____________
ID: 25162 \| Rating: 0 \| rate: / Reply Quote

Glenn Rogers Send message Joined: 4 Jul 08 Posts: 165 Credit: 363,844 RAC: 0	Message 25163 - Posted: 12 Jun 2009 \| 16:11:33 UTC - in response to Message 25033.
Yes Please Cluster, can you make it if possible so it works on X series ATI cards? It would be greatly appreciated... Cheers Glenn ____________
ID: 25163 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25168 - Posted: 12 Jun 2009 \| 16:57:21 UTC - in response to Message 25163.
Yes Please Cluster, can you make it if possible so it works on X series ATI cards? It would be greatly appreciated... All ATI HD series will work with the MW GPU project (most likely even the integrated chipsets like 780G), where single precision will be enough. In the moment it is simply not possible, because they don't support the required precision. But for the X series there may be no simple solution as they are not supported by the Stream SDK. One would have to write some DirectX shaders (or use the old BrookGPU) to get it to work and I don't know if they support the required features.
ID: 25168 \| Rating: 0 \| rate: / Reply Quote

arkayn Send message Joined: 14 Feb 09 Posts: 914 Credit: 74,781,320 RAC: 294	Message 25190 - Posted: 12 Jun 2009 \| 23:26:50 UTC - in response to Message 25162.
Apple is only going to support OpenCL on a very limited amount of GPU's. Graphics and OpenCL One of the big features in Snow Leopard is the ability to tap into many more computing resources at once, and part of this is the inclusion of OpenCL, which allows the system to make use of the graphics processor for everyday tasks when it's not in use by other programs. According to the Leopard technical specifications page, this ability will require a computer with at least the following graphics processors: NVIDIA Geforce 8600M GT, GeForce 8800 GT, GeForce 8800 GTS, Geforce 9400M, GeForce 9600M GT, GeForce GT 120, GeForce GT 130. ATI Radeon 4850, Radeon 4870 Most of these processors were released in the last year, which means that while Snow Leopard will run on any Intel system, its full speed and computing potential will only happen on the following models released so far: MacBook Pro: Mid-2007 and later MacBooks: Late 2008 models and later iMacs: Only the 2009 models and later Mac Mini: Only the 2009 models and later Mac Pro: All models with upgrades, though only the 2009 models will support it out of the box The following models will run the OS, but will not benefit from the enhanced power available from the GPU: MacBook Pro: Prior to 2007 MacBook: Prior to 2008 iMacs: Prior to 2009 Mac Mini: All except the 2009 models ____________
ID: 25190 \| Rating: 0 \| rate: / Reply Quote

ztmike Send message Joined: 4 Jun 09 Posts: 45 Credit: 447,355 RAC: 0	Message 25213 - Posted: 13 Jun 2009 \| 0:39:01 UTC
Is anybody even compiling/testing M@H Cuda app for Windows? Not sure why Windows was not supported first, as most people have that OS..
ID: 25213 \| Rating: 0 \| rate: / Reply Quote

SATAN Send message Joined: 27 Feb 09 Posts: 45 Credit: 305,963 RAC: 0	Message 25243 - Posted: 13 Jun 2009 \| 11:01:39 UTC
Well i've finally managed to get CUDA working properly on the Mac Pro. Not bad considering it's a slow old 8800GT. So looking forward to the next code release, to see if I can finally get Milkyway GPU working on this damn thing. ____________ Mars rules this confectionery war!
ID: 25243 \| Rating: 0 \| rate: / Reply Quote

Glenn Rogers Send message Joined: 4 Jul 08 Posts: 165 Credit: 363,844 RAC: 0	Message 25245 - Posted: 13 Jun 2009 \| 12:22:36 UTC - in response to Message 25168.
Thanks for the info not such an easy task it would appear...... ____________
ID: 25245 \| Rating: 0 \| rate: / Reply Quote

SATAN Send message Joined: 27 Feb 09 Posts: 45 Credit: 305,963 RAC: 0	Message 25251 - Posted: 13 Jun 2009 \| 13:04:43 UTC
Glenn, it may well be very easy. It's been almost 10 years since I did anything related to Unix commands. So getting CUDA itself to work is probably far easier than I found it. I found an idiots guide. I take my hate of to Travis and CP for building any GPU app. ____________ Mars rules this confectionery war!
ID: 25251 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25264 - Posted: 13 Jun 2009 \| 15:37:04 UTC - in response to Message 25243.
Well i've finally managed to get CUDA working properly on the Mac Pro. Not bad considering it's a slow old 8800GT. Do you have any performance figures to share? trisf told us a 9600GT on a C2D 6750 took about 15 minutes for the wedge 20 test unit. These test WUs are quite small so the execution time may be somehow limited by the CPU and all the calling overhead for the GPU stuff. Nevertheless it would be interesting to have a comparison with the 8800GT.
ID: 25264 \| Rating: 0 \| rate: / Reply Quote

trisf Send message Joined: 30 Nov 08 Posts: 11 Credit: 25,658 RAC: 0	Message 25275 - Posted: 13 Jun 2009 \| 17:34:42 UTC
I tried to run ps_sgr_214F_2s* on my 9600gt and self compiled linux64 binary... 1) insane desktop performance slowdown 2) after running 3hours i have to kill it 3) CPU load 100%
ID: 25275 \| Rating: 0 \| rate: / Reply Quote

Glenn Rogers Send message Joined: 4 Jul 08 Posts: 165 Credit: 363,844 RAC: 0	Message 25284 - Posted: 13 Jun 2009 \| 18:42:53 UTC - in response to Message 25251.
Gday Satan, I dont have any code writing experience or i would have a go at it myself and my ATI X1300 only handles single precision so it looks like i have to upgrade my graphics card...May have go trolling for some info on what my card is actually capable of... Absolutely hats off to Cluster and Travis they have done an outstanding job getting the app up and running....... Glenn ____________
ID: 25284 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25294 - Posted: 13 Jun 2009 \| 19:25:14 UTC - in response to Message 25275.
I tried to run ps_sgr_214F_2s* on my 9600gt and self compiled linux64 binary... 1) insane desktop performance slowdown 2) after running 3hours i have to kill it 3) CPU load 100% Yes, the production WUs are quite bit larger than the test WUs. As the code for MW_GPU does quite a bit more with one WU as the legacy MW@home code (roughly 300 or 400 times as much for the WU you tried to run, would have to check it to give an exact number), it is normal for them to take several hours. The fastest GPUs out there complete these WUs in about 50 seconds with the "classic" algorithm, albeit in double precision. Multiplying that time with 400 equals 5.5 hours. Such long WU were one of the goals of MW_GPU actually. That slow and sluggish behaviour of the GUI is a side affect of GPU apps with a very high utilization of the GPU. The ATI app also suffered (and still does to some extent) from this. One has to limit the duration of the GPU kernels somehow. That creates short opportunities for other tasks (like the screen refresh) to execute which will result in a smoother experience. The high CPU load should be easy to cure. One only have to send the application to sleep (a millisecond is enough) when it busy waits for the completion of a GPU kernel. That should be one line in the code (at least I hope so).
ID: 25294 \| Rating: 0 \| rate: / Reply Quote

SATAN Send message Joined: 27 Feb 09 Posts: 45 Credit: 305,963 RAC: 0	Message 25325 - Posted: 13 Jun 2009 \| 22:04:10 UTC
Cluster, I haven't dared mess with the Milkyway stuff, Gave me a big enough head ache just making sure CUDA was installed correctly. Will have a go over the next couple of days. I keep screwing something up because it keeps telling me that not target has been set. Will need to go through take a slow careful look at what i'm screwing up. I doubt i'll notice a slowdown with the desktop though as I run the 8800 purely on its own without a monitor connected. Will post back if/when I finally get the damn thing working properly. ____________ Mars rules this confectionery war!
ID: 25325 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25329 - Posted: 13 Jun 2009 \| 22:39:22 UTC - in response to Message 25325.
Will have a go over the next couple of days. I keep screwing something up because it keeps telling me that not target has been set. Will need to go through take a slow careful look at what i'm screwing up. I doubt i'll notice a slowdown with the desktop though as I run the 8800 purely on its own without a monitor connected. That could be the problem. Don't know how it works on a Mac, but under Win and Linux you have to attach a monitor to the card. Otherwise it is not active and one can't run anything on the GPU.
ID: 25329 \| Rating: 0 \| rate: / Reply Quote

SATAN Send message Joined: 27 Feb 09 Posts: 45 Credit: 305,963 RAC: 0	Message 25374 - Posted: 14 Jun 2009 \| 6:23:46 UTC Last modified: 14 Jun 2009 \| 6:50:39 UTC
I had no trouble getting it to run under BootCamp without a display connected. I don't know whether it is something in the Apple drivers or not, but I can run the CUDA examlples such as oceanFFT no problems and they show perfectly fine. Arkayn might have a better idea of why it works. [img=http://img44.imageshack.us/img44/7240/cudascreenshot.th.png] ____________ Mars rules this confectionery war!
ID: 25374 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 25375 - Posted: 14 Jun 2009 \| 7:08:22 UTC
According to Nvidia the requirement of having to attach a monitor is a strange Microsoft requirement that they could work around - but not without breaking WHQL certification. I don't know the deal with Linux though.
ID: 25375 \| Rating: 0 \| rate: / Reply Quote

arkayn Send message Joined: 14 Feb 09 Posts: 914 Credit: 74,781,320 RAC: 294	Message 25376 - Posted: 14 Jun 2009 \| 7:12:16 UTC - in response to Message 25374.
I had no trouble getting it to run under BootCamp without a display connected. I don't know whether it is something in the Apple drivers or not, but I can run the CUDA examlples such as oceanFFT no problems and they show perfectly fine. Arkayn might have a better idea of why it works. [img=http://img44.imageshack.us/img44/7240/cudascreenshot.th.png] Not really, I don't know hardly anything about software/driver developing. I am pretty good on app_info's up to when they added all that fplops to the mix. ____________
ID: 25376 \| Rating: 0 \| rate: / Reply Quote

verstapp Send message Joined: 26 Jan 09 Posts: 585 Credit: 464,286,454 RAC: 730	Message 25380 - Posted: 14 Jun 2009 \| 8:15:50 UTC Last modified: 14 Jun 2009 \| 8:18:15 UTC
Or even... Though you may have to shrink the image to make it fit. Not all of us have wide screens. :) ____________ Cheers, PeterV .
ID: 25380 \| Rating: 0 \| rate: / Reply Quote

borandi Send message Joined: 21 Feb 09 Posts: 180 Credit: 26,221,261 RAC: 0	Message 25407 - Posted: 14 Jun 2009 \| 12:05:13 UTC
There is a way around the monitor bug thing in windows without using a second monitor or a dummy plug. Go to your display settings, enable the second monitor as an extention of your desktop, AND as the primary monitor. When you click apply, you'll be left with a screen which is just your background. Now unplug the monitor cable from it's current graphics card, into the one you just enabled. You should be back to your desktop, albeit able to move your mouse off to the left. This enables both cards. The one drawback is that sometimes (not often) windows will pop up on the other screen - I had it with my MSN messenger, until I dragged the window over and then it was fine. ____________
ID: 25407 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 25537 - Posted: 15 Jun 2009 \| 15:38:40 UTC - in response to Message 25329.
Will have a go over the next couple of days. I keep screwing something up because it keeps telling me that not target has been set. Will need to go through take a slow careful look at what i'm screwing up. I doubt i'll notice a slowdown with the desktop though as I run the 8800 purely on its own without a monitor connected. That could be the problem. Don't know how it works on a Mac, but under Win and Linux you have to attach a monitor to the card. Otherwise it is not active and one can't run anything on the GPU. On the new macbook pros, you need to go into system preferences -> energy saver then select higher performance to use the other (faster) GPU. If you don't want to use that there's a line in evaluation_gpuX.cu which sets the device (it's at 1, i think it should be changed to 0 to use the on-chip GPU). ____________
ID: 25537 \| Rating: 0 \| rate: / Reply Quote

trisf Send message Joined: 30 Nov 08 Posts: 11 Credit: 25,658 RAC: 0	Message 25827 - Posted: 17 Jun 2009 \| 19:53:44 UTC
Trying to obtain results for linux_x86_64 cuda gpu http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=81905297 after ~4hours got this out result ps_sgr_214F5_2s_hiw_470211_1245248961_0_0 hessian [14 x 14]: 2.28519259071191482846 -0.39497154621000629682 -3.55474399915678374029 0.74480348740320800882 -0.10156406271555340481 -3.85943627057017080162 -0.81251291805806147295 -5.58602580857936370506 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -0.39497154621000629682 0.00175542913538606626 0.63195448873898374398 -0.05642450621960675566 -0.01579886098489345983 -0.01579886098489345983 -0.04513960563359534217 0.67709408450393004930 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -3.55474399915678374029 0.63195448873898374398 -12.18769307698152992714 0.94793169610104166534 -5.28133370369943122569 5.68759017660624976997 7.10948777626896344373 -48.75077341814914433371 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.74480348740320800882 -0.05642450621960675566 0.94793169610104166534 0.11736297489406412320 -0.31146327739151052905 -0.14896069933101330207 0.15573163869575526452 1.62502578060497171464 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -0.10156406271555340481 -0.01579886098489345983 -5.28133370369943122569 -0.31146327739151052905 0.02031282919645605034 0.08125129458136370886 0.12187693076981531703 1.42189759966271367375 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -3.85943627057017080162 -0.01579886098489345983 5.68759017660624976997 -0.14896069933101330207 0.08125129458136370886 3.14848748184104465508 -0.28437951993254273475 2.64066690736086684410 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -0.81251291805806147295 -0.04513960563359534217 7.10948777626896344373 0.15573163869575526452 0.12187693076981531703 -0.28437951993254273475 -0.10156411267558951295 3.65630814513906399199 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 -5.58602580857936370506 0.67709408450393004930 -48.75077341814914433371 1.62502578060497171464 1.42189759966271367375 2.64066690736086684410 3.65630814513906399199 16.25025891627273821882 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 0.00000000000000000000 gradient[14]: -0.47922010618095528534, 0.00001805584225343814, -0.12744264776820557472, -0.00211388771820253396, 0.00110095497385387375, -0.16134474836393408737, -0.00103189137901082972, 0.00658135446141017069, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000 initial_fitness: -3.12187372012968600288 inital_parameters[14]: 0.30045917696320600943, 30.00000000000000000000, -0.18211869995751800433, 162.48457188946628093618, 13.45735709916518807461, 6.27418922735829376336, 6.28318530717958623200, 8.36662928454264331890, -19.55595416138490350022, 218.02428246770318764902, 7.94448075244863360922, 5.46189868967828839885, 0.00000000000000000000, 18.87596924383203855768 result_fitness: 0.00000000000000000000 result_parameters[14]: 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000, 0.00000000000000000000 number_evaluations: 447 metadata: p: 27, v: 0.00005843793239637506 25.03408585491482796215 -0.01973286613357989189 9.72341257218874055468 0.99125759582857653207 0.07406591432380578433 5.64256730759030311617 0.22236400978163128883 0.02643427419835397973 -1.21804191646168202823 -12.80871199567133800201 -0.82128661750129838826 -3.16154559902874643385 2.90070763055914682127 and some stderr.txt APP: error reading hessian checkpoint file (for read): data_file == NULL shmget in attach_shmem: Invalid argument Can't set up shared mem: -1 Will run in standalone mode. APP: error reading hessian checkpoint file (for read): data_file == NULL APP: error reading hessian checkpoint file (for write): data_file == NULL called boinc_finish shmget in attach_shmem: Invalid argument Can't set up shared mem: -1 Will run in standalone mode. wu is still runnning
ID: 25827 \| Rating: 0 \| rate: / Reply Quote

SATAN Send message Joined: 27 Feb 09 Posts: 45 Credit: 305,963 RAC: 0	Message 25832 - Posted: 17 Jun 2009 \| 21:10:51 UTC
I'll give it ago when Travis posts the updated code files. I can't say that i will have any success though. ____________ Mars rules this confectionery war!
ID: 25832 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25894 - Posted: 18 Jun 2009 \| 10:11:25 UTC - in response to Message 25827.
Trying to obtain results for linux_x86_64 cuda gpu http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=81905297 after ~4hours got this out result ps_sgr_214F5_2s_hiw_470211_1245248961_0_0 [..] number_evaluations: 447 [..] As this is a 2 stream WU with a double sized wedge, I can give some comparison with a a HD3870 (overclocked to 860MHz). I've not run a whole WU yet (takes too long ;)), but I know the time for a single evaluation. As the the number of evaluations is given in the output file, I can say that HD3870 would take about 8000 seconds (2:15 hours) for the 447 evaluations (roughly 18 seconds per evaluation). This number is deduced from a normal sized wedge, so there is some uncertainty to it (maybe 20%). What graphics card do you use? Was it a 9600GT? It has 64 stream processors, opposed to the 112 to 128 of the 8800GT/GTX, 9800GT/GTX series. That would mean a G92 based graphics card is roughly as fast as a HD3870 with the current code, depending on the clock and the exact number of enabled units also a bit faster. The GT200 would battle it out with the HD4800 series then ;)
ID: 25894 \| Rating: 0 \| rate: / Reply Quote

trisf Send message Joined: 30 Nov 08 Posts: 11 Credit: 25,658 RAC: 0	Message 25921 - Posted: 18 Jun 2009 \| 15:15:26 UTC
Thanks CP. yes it was 9600gt strange behavior: when you stop project wus dont stop and continues to run only kill boinc helps
ID: 25921 \| Rating: 0 \| rate: / Reply Quote

Glenn Rogers Send message Joined: 4 Jul 08 Posts: 165 Credit: 363,844 RAC: 0	Message 25942 - Posted: 18 Jun 2009 \| 18:14:43 UTC - in response to Message 25921.
In the BOINC manager options menu check the enable manager exit menu check box then ok. Then file exit..the dialog box should have the checkbox stop science applications when exiting manager make sure this is checked click ok. That should be it ____________
ID: 25942 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 25954 - Posted: 18 Jun 2009 \| 20:23:37 UTC - in response to Message 25827. Last modified: 18 Jun 2009 \| 20:31:36 UTC
Trying to obtain results for linux_x86_64 cuda gpu http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=81905297 after ~4hours got this out result ps_sgr_214F5_2s_hiw_470211_1245248961_0_0 By the way, there may be a bug in the CUDA version when initializing the stream_c parameters. If one compares the init_constants function from the CPU version if (ap->sgr_coordinates == 0) { atGCToEq(ap->stream_parameters[i][0], 0, &ra, &dec, get_node(), wedge_incl(ap->wedge)); atEqToGal(ra, dec, &l, &b); } else if (ap->sgr_coordinates == 1) { gcToSgr(ap->stream_parameters[i][0], 0, ap->wedge, &lamda, &beta); //vickej2 sgrToGal(lamda, beta, &l, &b); //vickej2 } else { printf("Error: sgr_coordinates not valid"); } lbr[0] = l; lbr[1] = b; lbr[2] = ap->stream_parameters[i][1]; lbr2xyz(lbr, stream_c[i]); with the beginning of gpu__likelihood gc_to_gal(wedge, stream_parameters(i,0) * D_DEG2RAD, 0 * D_DEG2RAD, &(lbr[0]), &(lbr[1])); lbr[2] = stream_parameters(i,1); d_lbr2xyz(lbr, stream_c); one sees the CUDA version lacks the if statement for the SGR coordinates. Actually the CUDA version assumes that no SGR coordinates are used. At least this is how I read the code, the rotation matrix used in gc_to_gal is the same as in atEqToGal. I will stay with the CPU code version of that for the time being ;)
ID: 25954 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 26141 - Posted: 21 Jun 2009 \| 16:43:43 UTC
@Travis: I've posted the result file for the stripe 20 test WU in the other thread. The app for ATI is almost ready, only some "convenience features" are still missing: (correct) checkpointing, release of the CPU during the GPU computations (CPU load is really low as the likelihood is now done on the GPU too), higher responsivity of the GUI/OS by adaptive limiting of the execution domain size (only important for slower GPUs). The exactness of the single precision computation is astonishingly high. I've experimented with some tricks for the likehood stuff. Not only the usual Kahan summation in the likelihood kernel but also first order corrections for the exp and log functions extending them to more than 24bit precision. But frankly, it does not help at all, the resulting changes of the fitness are just at the noise level (10th decimal place). So I decided to take all that stuff out again (even not doing a Kahan summation within the convolve loop of the likelihood computation [but I do that in the integral part]) and restrict the Kahan summation to the GPU reduction (there it is required, as one won't get 9 decimal places precision without it). Maybe one has to do the more advanced corrections together in the integral and the likelihood parts, because the precision is limited by the worse of the two (or it does not matter at all), but actually I think the current state is already satisfying if you don't want to replace the double precision version completely ;)
ID: 26141 \| Rating: 0 \| rate: / Reply Quote

[P3D] Crashtest Send message Joined: 8 Jan 09 Posts: 58 Credit: 16,804,661 RAC: 12	Message 26754 - Posted: 30 Jun 2009 \| 15:39:44 UTC - in response to Message 26141.
Are there any news about the CUDA App for MW (running on Windows) ? I would like to compare the "Power" of the CUDA-Cards running MW unit with Double Prec. like the current ATI-Cards.
ID: 26754 \| Rating: 0 \| rate: / Reply Quote

Vincenzo Venuto Send message Joined: 21 Apr 09 Posts: 2 Credit: 120,747 RAC: 43	Message 27864 - Posted: 16 Jul 2009 \| 19:50:07 UTC
Hi all, i have a video card based on nvidia 8600GT After successfully compiled milkyway Cuda and installing, when MilkywayGPU elaborate a WU it's crash with "floating point exception". Why? Anyone compile and successfully execute milkyway CUDA? Thank's at all! Vincenzo
ID: 27864 \| Rating: 0 \| rate: / Reply Quote

Cappy [Team Musketeers] Send message Joined: 3 Oct 07 Posts: 71 Credit: 33,212,009 RAC: 0	Message 28386 - Posted: 26 Jul 2009 \| 10:41:31 UTC
is there ANY NEW NEWS on a CUDA app for MW????? its been talked about alot but havent really seen any updates on its progress.... is there or will there ever be a CUDA app for MW ???? BTW AQUA's MT app kicks ass!!!!!!!!! no need for GPU apps with the multi thread app,, and its a whole lot faster then GPU. ____________
ID: 28386 \| Rating: 0 \| rate: / Reply Quote

arkayn Send message Joined: 14 Feb 09 Posts: 914 Credit: 74,781,320 RAC: 294	Message 28407 - Posted: 26 Jul 2009 \| 16:59:32 UTC - in response to Message 28386.
Anthony Waters Posted this 18 Jul 2009 1:55:17 UTC Hello, This is Anthony, I'm working on the CUDA application with Travis. There are a few things that need to be resolved in order to get the CUDA app released, namely 1. Checking the accuracy of the double precision math 2. Integration with BOINC Anthony Waters Posted this 24 Jul 2009 21:58:58 UTC The GPU app will go up as a beta application on the regular Milkyway@Home site when it is ready. The final likelihood is accurate to about 12 decimal points, while we would like it higher, 12 is good enough. Therefore the next step is setting up the server, since it requires a major upgrade it will take a couple of weeks and will most likely be released with 0.19. ____________
ID: 28407 \| Rating: 0 \| rate: / Reply Quote

krahulik Send message Joined: 7 Nov 08 Posts: 14 Credit: 179,303,710 RAC: 3	Message 28427 - Posted: 27 Jul 2009 \| 7:29:58 UTC
Travis Posted 22 Jul 2009 5:02:47 UTC We actually have a working CUDA double precision app running right now, our new researcher Anthony is working on it. I think he's going to be releasing it as a beta application in the next week or so -- the only issue is that we're having an issue with the likelihood calculation that we want to resolve, when that happens we'll have to update all the applications.
ID: 28427 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28652 - Posted: 31 Jul 2009 \| 1:53:36 UTC Last modified: 31 Jul 2009 \| 2:12:05 UTC
Here is a new release of the GPU App source (v0.08) (Binaries for Windows and Linux included) http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.tar http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.zip please read the readme http://milkyway.cs.rpi.edu/milkyway/download/code_release/readme This will be the final release version that goes on BOINC pending any bugs and the new likelihood fixes. Also this release is more of a "does this work on other people's machines besides mine" so don't expect it to run with BOINC
ID: 28652 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 28654 - Posted: 31 Jul 2009 \| 3:31:01 UTC - in response to Message 28652.
(Binaries for Windows and Linux included) http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.tar http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.zip please read the readme http://milkyway.cs.rpi.edu/milkyway/download/code_release/readme from the readme: it takes around 90 seconds to execute on a GTX 285 By the way, the test WUs supplied in that package have the size of the (now quite rare) ~18 credits (stripes 79, 82, 86) or ~27 credit (stripes 20, 21) WUs.
ID: 28654 \| Rating: 0 \| rate: / Reply Quote

Divide Overflow Send message Joined: 16 Feb 09 Posts: 109 Credit: 11,089,510 RAC: 0	Message 28655 - Posted: 31 Jul 2009 \| 4:10:06 UTC
Cluster Physik, I do hope that you'll be able to work your magic and update the ATI application to be compatible with this. Is there any way to get it working better with the latest Catalyst / CAL versions? As always, your work is very much appreciated!
ID: 28655 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 28656 - Posted: 31 Jul 2009 \| 4:46:04 UTC - in response to Message 28655.
Cluster Physik, I do hope that you'll be able to work your magic and update the ATI application to be compatible with this. Is there any way to get it working better with the latest Catalyst / CAL versions? As always, your work is very much appreciated! In the moment it is compatible. The project has still to figure out what they want to change ;) And I really don't know what is up with the new Catalyst drivers. In some configurations under Vista it runs quite well, but WinXP is a no go with anything newer than 9.2. If I find the reason, I will of course update the apps.
ID: 28656 \| Rating: 0 \| rate: / Reply Quote

Vincenzo Venuto Send message Joined: 21 Apr 09 Posts: 2 Credit: 120,747 RAC: 43	Message 28665 - Posted: 31 Jul 2009 \| 12:10:54 UTC
Hi development, i try to run milkyway cuda gpu on my 8600GT, but is not supported (cuda 1.1) you have drop the support under 1.3? Thank's Vincenzo
ID: 28665 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28667 - Posted: 31 Jul 2009 \| 12:45:15 UTC
CUDA 1.3 is required for Double Precision support, single precision was tried, however, it only gave accuracy to 6/7 decimal places which is not enough.
ID: 28667 \| Rating: 0 \| rate: / Reply Quote

slavko.sk Send message Joined: 24 Dec 07 Posts: 2 Credit: 18,574,316 RAC: 74,574	Message 28672 - Posted: 31 Jul 2009 \| 15:12:01 UTC
So it will not work on my 8800 GTS as well? ____________ ALL GLORY TO THE HYPNOTOAD! Do You Dare?
ID: 28672 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 28673 - Posted: 31 Jul 2009 \| 15:43:59 UTC - in response to Message 28667.
CUDA 1.3 is required for Double Precision support, single precision was tried, however, it only gave accuracy to 6/7 decimal places which is not enough. then your volunteers just took a nose-dive... i only have one card 1.3 compatible which is a gtx 260.... my 2 other cards are 9400 gt which are 1.1.....
ID: 28673 \| Rating: 0 \| rate: / Reply Quote

Cluster Physik Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0	Message 28675 - Posted: 31 Jul 2009 \| 16:23:56 UTC - in response to Message 28673.
CUDA 1.3 is required for Double Precision support, single precision was tried, however, it only gave accuracy to 6/7 decimal places which is not enough. then your volunteers just took a nose-dive... i only have one card 1.3 compatible which is a gtx 260.... my 2 other cards are 9400 gt which are 1.1..... It is the same restriction as for ATI GPUs.
ID: 28675 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 28677 - Posted: 31 Jul 2009 \| 17:49:13 UTC - in response to Message 28675.
Yep, excludes a lot more people though due to Nvidia's reluctance to move to double precision. Are there still plans to make a single precision GPU sister project?
ID: 28677 \| Rating: 0 \| rate: / Reply Quote

BarryAZ Send message Joined: 1 Sep 08 Posts: 512 Credit: 222,955,456 RAC: 159,395	Message 28680 - Posted: 31 Jul 2009 \| 19:02:05 UTC - in response to Message 28675.
So this will only work on GTX cards? I have 9800GT cards -- sounds like I'm not in the lucky subset at the moment. CUDA 1.3 is required for Double Precision support, single precision was tried, however, it only gave accuracy to 6/7 decimal places which is not enough. then your volunteers just took a nose-dive... i only have one card 1.3 compatible which is a gtx 260.... my 2 other cards are 9400 gt which are 1.1..... It is the same restriction as for ATI GPUs. ____________
ID: 28680 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 28682 - Posted: 31 Jul 2009 \| 20:20:47 UTC - in response to Message 28680. Last modified: 31 Jul 2009 \| 20:29:35 UTC
gpugrid has currently total active Users 5,174 2,391 Hosts 8,322 3,575 source: Boincstats and this is a mixed bag of gpu... most have above 50 cores... ps3 is in this figure but a small percentage...8% i think... looking at boincstats... and to add things over, AQUA has a gpu app out. and seti lets any gpu cuda capable card run seti MB. so, your pretty much, asking the people from other projects with high-end cards to come over play.. not many of those... i myself have 1 "1.3" card. and lets mention the fact that, there are other projects with cuda app/gpu in the works as some of you know... i have a ati 2600 xt card setting in a box... ____________ I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
ID: 28682 \| Rating: 0 \| rate: / Reply Quote

Divide Overflow Send message Joined: 16 Feb 09 Posts: 109 Credit: 11,089,510 RAC: 0	Message 28683 - Posted: 31 Jul 2009 \| 21:31:55 UTC - in response to Message 28682.
Sorry zpm. You must be at least this tall to ride the attraction.
ID: 28683 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 28684 - Posted: 31 Jul 2009 \| 21:54:48 UTC - in response to Message 28683.
what, do i have to be as tall as Shaq. i'm 6'11" as is.... i'm just pointing out the fact that the volunteer base of gpu is very low, gpu computing with boinc is still in it's first year...
ID: 28684 \| Rating: 0 \| rate: / Reply Quote

Divide Overflow Send message Joined: 16 Feb 09 Posts: 109 Credit: 11,089,510 RAC: 0	Message 28685 - Posted: 31 Jul 2009 \| 22:10:54 UTC - in response to Message 28684.
Hehe, I sympathize with you. I have quite a few of those older cards available as well. I guess the project really needs the precision and those "older" GPU's aren't quite up to the task.
ID: 28685 \| Rating: 0 \| rate: / Reply Quote

James Nunley Send message Joined: 29 Nov 07 Posts: 39 Credit: 60,297,339 RAC: 31,051	Message 28699 - Posted: 1 Aug 2009 \| 6:55:05 UTC
Does this mean no single precision app for ati either?
ID: 28699 \| Rating: 0 \| rate: / Reply Quote

Travis Volunteer moderator Project administrator Project developer Project tester Project scientist Send message Joined: 30 Aug 07 Posts: 1976 Credit: 26,480 RAC: 0	Message 28700 - Posted: 1 Aug 2009 \| 8:20:59 UTC - in response to Message 28699.
Does this mean no single precision app for ati either? Thats correct. Milkyway@Home really requires double precision accuracy for what we're doing. There are too many numerical issues if we go to single precision. ____________
ID: 28700 \| Rating: 0 \| rate: / Reply Quote

TomaszPawel Send message Joined: 9 Nov 08 Posts: 41 Credit: 92,786,635 RAC: 0	Message 28701 - Posted: 1 Aug 2009 \| 10:47:03 UTC - in response to Message 28700.
-------------------------------------------------------------------------------- -------------------------------------------------------------------------------- NVIDIA CUDA Windows XP and Vista Release Notes Version 2.3 -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- New Features -------------------------------------------------------------------------------- Hardware Support o See http://www.nvidia.com/object/cuda_learn_products.html Platform Support o Additional OS support - Microsoft Windows 7 32-bit and 64-bit CUFFT Features o Performance enhancements o Double precision - CUFFT now supports double-precision transforms, with types and functions analagous to the existing single-precision versions. Similarly, the "cufftType" enumeration (used in calls like cufftPlan1d) has expanded to include double-precision identifiers: Precision: Single Double Type: cufftReal cufftDoubleReal Type: cufftComplex cufftDoubleComplex cufftType: CUFFT_R2C CUFFT_D2Z cufftType: CUFFT_C2R CUFFT_Z2D cufftType: CUFFT_C2C CUFFT_Z2Z Function: cufftExecC2C cufftExecZ2Z Function: cufftExecR2C cufftExecD2Z Function: cufftExecC2R cufftExecZ2D - The double-precision versions are invoked in an identical manner to the single-precision ones, obviously with arguments changed from the single- to the double-precision types. See "cufft.h" for exact definitions of the above. Cross-Compilation Support o Support compilation of 32bit applications on 64bit hosts. System Enhancements o SLI Device Enumeration - Applications can now access all devices in an SLI group individually. Double Handling by the Compiler o when a ptx file with an sm version prior to sm_13 contains double precision instructions, ptxas now emits a warning that double precision instructions are demoted to single precision. ptxas has a new option --suppress-double-demote-warning to suppress this warning :) ____________ A proud member of the Polish National Team COME VISIT US at Polish National Team FORUM
ID: 28701 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28729 - Posted: 1 Aug 2009 \| 21:51:17 UTC - in response to Message 28652. Last modified: 1 Aug 2009 \| 21:55:50 UTC
Here is a new release of the GPU App source (v0.08) (Binaries for Windows and Linux included) http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.tar http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.08.zip please read the readme http://milkyway.cs.rpi.edu/milkyway/download/code_release/readme This will be the final release version that goes on BOINC pending any bugs and the new likelihood fixes. Also this release is more of a "does this work on other people's machines besides mine" so don't expect it to run with BOINC Saw this in the ReadMe .. -Windows users need to use the 32 bit version 32bit version of what? BOINC? Because I downloaded the 64bit version..
ID: 28729 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28730 - Posted: 1 Aug 2009 \| 22:02:56 UTC - in response to Message 28729.
CUDA 2.2 Toolkit and appropriate driver listed on NVIDIA's website[1] -Windows users need to use the 32 bit version meaning the 32 bit version CUDA 2.2 Toolkit, sorry for the confusion
ID: 28730 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28733 - Posted: 2 Aug 2009 \| 0:50:25 UTC - in response to Message 28730.
meaning the 32 bit version CUDA 2.2 Toolkit, sorry for the confusion Wait..so I need Nvidas 32bit GPU drivers to be able to run this? Why would someone delete their 64bit GPU drivers to download 32bit drivers to run this on their spare time? Unless, I'm missing something here?
ID: 28733 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28734 - Posted: 2 Aug 2009 \| 0:51:25 UTC - in response to Message 28733. Last modified: 2 Aug 2009 \| 0:51:54 UTC
meaning the 32 bit version CUDA 2.2 Toolkit, sorry for the confusion Wait..so I need Nvidas 32bit GPU drivers to be able to run this? Why would someone delete their 64bit GPU drivers to download 32bit drivers to run this on their spare time? Unless, I'm missing something here? If that's how it is..it sounds like this is in Alpha stage.
ID: 28734 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28744 - Posted: 2 Aug 2009 \| 12:45:10 UTC - in response to Message 28734. Last modified: 2 Aug 2009 \| 12:46:28 UTC
32 bit version of the CUDA 2.2 Toolkit, you can use 32bit or 64bit drivers, sorry for the confusion Appropriate driver, just means the version number, I'll update the readme
ID: 28744 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28760 - Posted: 2 Aug 2009 \| 22:15:28 UTC
I clicked on the bat file..then clicked on the .exe file, and I got this: http://img195.imageshack.us/img195/1064/milkywayu.png Did I do it wrong?
ID: 28760 \| Rating: 0 \| rate: / Reply Quote

borandi Send message Joined: 21 Feb 09 Posts: 180 Credit: 26,221,261 RAC: 0	Message 28774 - Posted: 3 Aug 2009 \| 0:55:25 UTC - in response to Message 28760. Last modified: 3 Aug 2009 \| 0:56:24 UTC
Looks like you haven't installed the CUDA 2.2 Toolkit? Windows should automatically reference in the files like cudart.dll. Toolkit != drivers. You need both. ____________
ID: 28774 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28792 - Posted: 3 Aug 2009 \| 10:21:08 UTC - in response to Message 28774. Last modified: 3 Aug 2009 \| 10:33:24 UTC
Looks like you haven't installed the CUDA 2.2 Toolkit? Windows should automatically reference in the files like cudart.dll. Toolkit != drivers. You need both. Okay..now I downloaded the Cuda Toolkit and rebooted, now I get an error of: "The application was unable to start correctly (0xc000007b). Click OK to close the application." I am doing it right, right? Click on the Windows BAT file then click on the .exe? When I click on the BAT file 2 window things go onto my desktop, with the names of "Out" and "Stderr" ..I believe it shows them on my picture in my last post, are those files suppose to be on the desktop? I realize this is testing ..but these directions suck. I hope you guys aren't going to make it this hard for when it launches.
ID: 28792 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 28801 - Posted: 3 Aug 2009 \| 14:04:29 UTC - in response to Message 28792.
It'll only launch when the application has been fully integrated into BOINC.
ID: 28801 \| Rating: 0 \| rate: / Reply Quote

borandi Send message Joined: 21 Feb 09 Posts: 180 Credit: 26,221,261 RAC: 0	Message 28809 - Posted: 3 Aug 2009 \| 15:56:16 UTC - in response to Message 28652.
It'll only launch when the application has been fully integrated into BOINC. This will be the final release version that goes on BOINC pending any bugs and the new likelihood fixes. Also this release is more of a "does this work on other people's machines besides mine" so don't expect it to run with BOINC Emanuel, you seem to be contradicting Anthony? I realize this is testing ..but these directions suck. I hope you guys aren't going to make it this hard for when it launches. Yes this is testing - to see if it works on other CUDA developers machines. Of course it'll be more straight forward than this - BOINC has the CUDA infrastructure to cope for it just to be added. Then it's a case of making sure you have the right driver. As for the windows zip file, just loaded it on my work machine where I do most of my CUDA coding. Opened the .sln in VS08 and it's complaining it can't find the file Cuda.Rules: Custom build rules file 'mw_gpu_v0.08\milkyway\bin\Cuda.Rules' was not found or failed to load. App doesn't work here on a case of it's not a CUDA machine - was hoping to load it in VS then run in emu mode. I'll try on my CUDA machine at home in due course. ____________
ID: 28809 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28846 - Posted: 3 Aug 2009 \| 23:46:43 UTC - in response to Message 28792.
Okay..now I downloaded the Cuda Toolkit and rebooted, now I get an error of: "The application was unable to start correctly (0xc000007b). Click OK to close the application." I am doing it right, right? Click on the Windows BAT file then click on the .exe? When I click on the BAT file 2 window things go onto my desktop, with the names of "Out" and "Stderr" ..I believe it shows them on my picture in my last post, are those files suppose to be on the desktop? I realize this is testing ..but these directions suck. I hope you guys aren't going to make it this hard for when it launches. 1. Check the file stderr.txt for the detailed error message 2. Verify that in the same folder of the executable there is a file named astronomy_parameters.txt, stars.txt, and search_parameters.txt. If there is not then the set_parameters.bat file was run incorrectly, it needs to be executed from the command line and given a command line argument. Please see http://commandwindows.com/ if you are unfamiliar with the command line. Alternatively if you cannot get the command line to work you can copy and rename the file astronomy_parameters-20.txt to astronomy_parameters.txt, stars-20.txt to stars.txt and search_parameters-20.txt to search_parameters, this is effectively what the batch file does, except instead of using 20 it uses the number that was passed in as the command line argument. 3. From the looks of it your using Vista/Win7 so it might be an Operating System specific issue, I have personally only tested it on Windows XP x64 and Linux/GNU 64 bit When it launches through BOINC the application will download and automatically run without user intervention, unless an error occurs with initializing the GPU. As for the directions, any ideas as to what to include to make them better? I tried to write them as detailed as possible but it is always hard to imagine how so many different people can interpret a piece of writing, especially if they are unfamiliar with the subject.
ID: 28846 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28847 - Posted: 3 Aug 2009 \| 23:54:23 UTC - in response to Message 28809.
As for the windows zip file, just loaded it on my work machine where I do most of my CUDA coding. Opened the .sln in VS08 and it's complaining it can't find the file Cuda.Rules: Custom build rules file 'mw_gpu_v0.08\milkyway\bin\Cuda.Rules' was not found or failed to load. App doesn't work here on a case of it's not a CUDA machine - was hoping to load it in VS then run in emu mode. I'll try on my CUDA machine at home in due course. I have a lacking of expertise with respect to Visual Studio, especially the latest versions, so I'm not quite sure how to setup the solution and project files in a machine independent way. If you look in the same directory as the project file you will notice that the Cuda.Rules file is located in there, if you could update Visual Studio to use that file it should update the path and correctly load the file. I think the place where you specify the Build Rules is under the project settings menu (Right click on the project) and navigate to Custom Build Rules I realized that after I wrote the above paragraph that I neglected to include the actual file in the release, since the paragraph still sort of applies I chose not to delete it. Here is the Cuda.Rules file http://milkyway.cs.rpi.edu/milkyway/download/code_release/
ID: 28847 \| Rating: 0 \| rate: / Reply Quote

Mike Send message Joined: 9 Apr 09 Posts: 19 Credit: 544 RAC: 0	Message 28858 - Posted: 4 Aug 2009 \| 4:57:15 UTC
1. Check the file stderr.txt for the detailed error message Nothing in there but this: "" As far as doing number 2 of what you said..you completely lost me on that one, I think I'll just sit this one out. I never done this sort of thing, but I just wanted to help to get this Cuda project on its knees faster. Hopefully you guys have people testing this besides whats been said in this thread so far..because honestly its looking a little sparse for testers. I run Windows 7 64bit with a dual boot of Ubuntu 9.04 ..which I'm still learning, if I'm having this much trouble on Windows, I don't think I'm going to be trying it on Linux.
ID: 28858 \| Rating: 0 \| rate: / Reply Quote

borandi Send message Joined: 21 Feb 09 Posts: 180 Credit: 26,221,261 RAC: 0	Message 28863 - Posted: 4 Aug 2009 \| 8:31:58 UTC - in response to Message 28847. Last modified: 4 Aug 2009 \| 8:32:18 UTC
I realized that after I wrote the above paragraph that I neglected to include the actual file in the release, since the paragraph still sort of applies I chose not to delete it. Here is the Cuda.Rules file http://milkyway.cs.rpi.edu/milkyway/download/code_release/ Thanks Anthony :) ____________
ID: 28863 \| Rating: 0 \| rate: / Reply Quote

BulldogPO Send message Joined: 30 Jun 09 Posts: 17 Credit: 40,702,094 RAC: 0	Message 28927 - Posted: 5 Aug 2009 \| 14:28:06 UTC
Damn, project does not accept new users :(
ID: 28927 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 28936 - Posted: 5 Aug 2009 \| 16:12:35 UTC - in response to Message 28927.
Damn, project does not accept new users :( There is no project for the CUDA application, unless you are referring to something else? Right now the CUDA application is currently in alpha stage for community preview and will be joining the regular Milkyway@Home project as a Beta application once the likelihood issue is resolved.
ID: 28936 \| Rating: 0 \| rate: / Reply Quote

BulldogPO Send message Joined: 30 Jun 09 Posts: 17 Credit: 40,702,094 RAC: 0	Message 28938 - Posted: 5 Aug 2009 \| 16:44:17 UTC
Ah, so how do I join it? As I did try using this: http://milkyway.cs.rpi.edu/milkyway_gpu/
ID: 28938 \| Rating: 0 \| rate: / Reply Quote

arkayn Send message Joined: 14 Feb 09 Posts: 914 Credit: 74,781,320 RAC: 294	Message 28939 - Posted: 5 Aug 2009 \| 17:49:17 UTC - in response to Message 28938.
What that means is that CUDA will be on the regular MW project and the GPU one will most likely go away as they got the server straitened out. ____________
ID: 28939 \| Rating: 0 \| rate: / Reply Quote

Labbie Send message Joined: 29 Aug 07 Posts: 327 Credit: 116,463,193 RAC: 0	Message 28947 - Posted: 5 Aug 2009 \| 19:52:43 UTC - in response to Message 28939.
...will go away as they got the server straitened out. Until a gazillion new hosts attach here running CUDA. ;) ____________ Calm Chaos Forum...Join Calm Chaos Now
ID: 28947 \| Rating: 0 \| rate: / Reply Quote

banditwolf Send message Joined: 12 Nov 07 Posts: 2425 Credit: 295,133 RAC: 0	Message 28948 - Posted: 5 Aug 2009 \| 21:32:42 UTC - in response to Message 28947.
[quote]...will go away as they got the server straitened out. If they ever do. Each 'fix' tends to break things. Until a gazillion new hosts attach here running CUDA. ;) Listen carefully for the 'BOOM'. ;) ____________ Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it.
ID: 28948 \| Rating: 0 \| rate: / Reply Quote

Divide Overflow Send message Joined: 16 Feb 09 Posts: 109 Credit: 11,089,510 RAC: 0	Message 28959 - Posted: 6 Aug 2009 \| 1:24:57 UTC
What ever happened to the plan to substantially increase the science done for GPU tasks, providing much longer running work units? Will this still happen here at this project?
ID: 28959 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 28962 - Posted: 6 Aug 2009 \| 3:36:23 UTC - in response to Message 28947. Last modified: 6 Aug 2009 \| 3:42:15 UTC
...will go away as they got the server straitened out. Until a gazillion new hosts attach here running CUDA. ;) not going too happen b/c of single precision not being accepted...1.3 compatible only... and i don't see all the users at gpugrid which have 1.3 comp cards coming over here.. no offense to milkyway.. only reason they would come is if... A. work from the project they like, is fresh out of work.. and the secondary project that they use when this happens runs out of work.. B. Project they like goes down for some reason... C. problems with certain individual computers.
ID: 28962 \| Rating: 0 \| rate: / Reply Quote

BulldogPO Send message Joined: 30 Jun 09 Posts: 17 Credit: 40,702,094 RAC: 0	Message 28966 - Posted: 6 Aug 2009 \| 5:09:33 UTC
So how exactly i put my GF cards to work this on Vista X64?
ID: 28966 \| Rating: 0 \| rate: / Reply Quote

jedirock Send message Joined: 8 Nov 08 Posts: 178 Credit: 6,140,854 RAC: 0	Message 28984 - Posted: 6 Aug 2009 \| 18:50:32 UTC - in response to Message 28966.
So how exactly i put my GF cards to work this on Vista X64? Right now, you can't. Once the CUDA application is finished, you can attach to whatever project hosts it (we still don't know if it'll be Milkyway or Milkyway_gpu), and BOINC 6.4.x or later should pick up the GPU app automatically. ____________
ID: 28984 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 28994 - Posted: 7 Aug 2009 \| 8:56:58 UTC - in response to Message 28809. Last modified: 7 Aug 2009 \| 8:59:32 UTC
It'll only launch when the application has been fully integrated into BOINC. This will be the final release version that goes on BOINC pending any bugs and the new likelihood fixes. Also this release is more of a "does this work on other people's machines besides mine" so don't expect it to run with BOINC Emanuel, you seem to be contradicting Anthony? You're right, I think I misinterpreted what he said. You can understand "don't expect it to run with BOINC" as "BOINC wouldn't be able to run the program" or "BOINC can run the program but we're not sending out GPU-specific work-units yet" and I interpreted it as the former, but I guess it's the latter. That is, pending any bugs and the new likelihood fixes :) By the way, how would such GPU-specific WUs (that do more work) be sent out to the CAL app? Could it pretend to be a CUDA app, or would it be have to be a special case server-side?
ID: 28994 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29049 - Posted: 8 Aug 2009 \| 17:31:54 UTC - in response to Message 28994.
As I understand it BOINC does not support the notion of GPU-specific WUs. When the version of the milkyway application is updated the client asks the server what application to download. If there exists a CUDA 1.3 capable card the client will download the CUDA application, otherwise, the CPU version will be downloaded. Then this application is used for every WU that gets processed.
ID: 29049 \| Rating: 0 \| rate: / Reply Quote

Bymark Send message Joined: 6 Mar 09 Posts: 48 Credit: 385,976,815 RAC: 0	Message 29176 - Posted: 10 Aug 2009 \| 16:53:51 UTC - in response to Message 29049.
Confused, what Gpu card to by next, ati or nvidia, nvidia has more project accepting, but ati seems faster here, someone get me a answer, please? ____________
ID: 29176 \| Rating: 0 \| rate: / Reply Quote

TomaszPawel Send message Joined: 9 Nov 08 Posts: 41 Credit: 92,786,635 RAC: 0	Message 29203 - Posted: 11 Aug 2009 \| 6:04:59 UTC - in response to Message 29176. Last modified: 11 Aug 2009 \| 6:06:47 UTC
When MW will relased CUDA app, all will be clear, We will be able to do direct comparision ATI vs Nvidia.... However, making cuda app takes sooooooooooooooooooooo long..... ____________ A proud member of the Polish National Team COME VISIT US at Polish National Team FORUM
ID: 29203 \| Rating: 0 \| rate: / Reply Quote

refla Send message Joined: 2 Mar 09 Posts: 20 Credit: 10,718,572 RAC: 0	Message 29204 - Posted: 11 Aug 2009 \| 8:07:50 UTC - in response to Message 29203. Last modified: 11 Aug 2009 \| 8:08:23 UTC
When MW will relased CUDA app, all will be clear, We will be able to do direct comparision ATI vs Nvidia.... However, making cuda app takes sooooooooooooooooooooo long..... I attached MW_gpu 3 mouths ago. I have ever recieved any CUDA WU. And now My account of MW_gpu has been deleted by server. It denied my login and refuse to accept I reattach it....... I think CUDA is an impossible mission for MW_gpu......
ID: 29204 \| Rating: 0 \| rate: / Reply Quote

jedirock Send message Joined: 8 Nov 08 Posts: 178 Credit: 6,140,854 RAC: 0	Message 29211 - Posted: 11 Aug 2009 \| 12:12:59 UTC - in response to Message 29204.
When MW will relased CUDA app, all will be clear, We will be able to do direct comparision ATI vs Nvidia.... However, making cuda app takes sooooooooooooooooooooo long..... I attached MW_gpu 3 mouths ago. I have ever recieved any CUDA WU. And now My account of MW_gpu has been deleted by server. It denied my login and refuse to accept I reattach it....... I think CUDA is an impossible mission for MW_gpu...... For one, the CUDA application isn't even finished yet, so there's no way you could've downloaded CUDA work from it. As for the account problems, that was probably caused by the database crash not long ago that affected the main project too. ____________
ID: 29211 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29231 - Posted: 11 Aug 2009 \| 23:12:32 UTC - in response to Message 29204.
When MW will relased CUDA app, all will be clear, We will be able to do direct comparision ATI vs Nvidia.... However, making cuda app takes sooooooooooooooooooooo long..... I attached MW_gpu 3 mouths ago. I have ever recieved any CUDA WU. And now My account of MW_gpu has been deleted by server. It denied my login and refuse to accept I reattach it....... I think CUDA is an impossible mission for MW_gpu...... The Milkyway_GPU project has been closed because the CUDA application will be on the regular Milkyway@Home project. The only step left is updating the server side BOINC software, once that is done the CUDA application will be released on BOINC as a beta application. I am unable to give a time estimate for when the server side BOINC software will be upgraded because the upgrade cannot be done by me, it has to be done by the staff that controls the server that Milkyway@Home is hosted on. An alpha community preview of the CUDA application can be found in this post http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=906&nowrap=true#28652
ID: 29231 \| Rating: 0 \| rate: / Reply Quote

Emanuel Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 597	Message 29307 - Posted: 13 Aug 2009 \| 16:11:57 UTC - in response to Message 29049.
As I understand it BOINC does not support the notion of GPU-specific WUs. When the version of the milkyway application is updated the client asks the server what application to download. If there exists a CUDA 1.3 capable card the client will download the CUDA application, otherwise, the CPU version will be downloaded. Then this application is used for every WU that gets processed. Does that mean that any project utilizing CUDA won't run anymore CPU units alongside it? (as I understand it the GPU client does hundreds of times more work, but still)
ID: 29307 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29319 - Posted: 13 Aug 2009 \| 20:16:19 UTC - in response to Message 29307.
As I understand it BOINC does not support the notion of GPU-specific WUs. When the version of the milkyway application is updated the client asks the server what application to download. If there exists a CUDA 1.3 capable card the client will download the CUDA application, otherwise, the CPU version will be downloaded. Then this application is used for every WU that gets processed. Does that mean that any project utilizing CUDA won't run anymore CPU units alongside it? (as I understand it the GPU client does hundreds of times more work, but still) I'm not 100% sure but I think the way it works is as follows: (Hardware - Resulting Applications) 1 CPU - 1 CPU Application Runs 2 CPUs - 2 CPU Applications Run 1 CPU and 1 GPU - 1 GPU Application Runs 2 CPUs and 1 GPU - 1 GPU Application Runs and 1 CPU Application Runs
ID: 29319 \| Rating: 0 \| rate: / Reply Quote

DoctorNow Send message Joined: 28 Aug 07 Posts: 146 Credit: 5,183,509 RAC: 0	Message 29495 - Posted: 19 Aug 2009 \| 4:27:06 UTC - in response to Message 29319. Last modified: 19 Aug 2009 \| 4:30:53 UTC
2 CPUs and 1 GPU - 1 GPU Application Runs and 1 CPU Application Runs It depends on how strong the GPU-app uses a CPU, too. Regarding GPUGrid and Collatz Conjecture, the CPU-utilization is almost 0%, so that you really have 2 CPU apps and 1 GPU app running! The CUDA-Code over at AQUA however was bad, took too much CPU-time, and other apps were affected by that. ____________ Member of BOINC@Heidelberg and ATA! My BOINCstats
ID: 29495 \| Rating: 0 \| rate: / Reply Quote

Providence Christian School Send message Joined: 6 May 08 Posts: 7 Credit: 1,769,162 RAC: 189	Message 29502 - Posted: 19 Aug 2009 \| 15:04:44 UTC Last modified: 19 Aug 2009 \| 16:03:16 UTC
For those of you worrying about the double precision requirement, there are other projects which do not require double precision accuracy and you can put your less well endowed cards on. (I don't have a CC 1.3 card.) Einstein - CUDA app in BETA, compute capability >= 1.1 Collatz - CUDA in beta, ATI in beta, the ATI app will even run on slow integrated graphics chipsets. Milkyway - CUDA in beta, compute capability >= 1.3; ATI app requires double precision capable card, (and some zippy SSEx optimized cpu apps) AQUA- CUDA app has been decommissioned, but they do have a multicore CPU app. GPUGRID - CUDA, compute capability >= 1.1 SETI - CUDA, compute capability >= 1.1 non BOINC: folding@home offers CUDA, ATI (needs a reasonable card to meet deadlines), multithreaded CPU and most of these also have optimized CPU apps which can use SSE(x) for significant throughput improvements. It seems like the projects which are working on GPU apps are also investing into SIMD based optimizations. So there are plenty of projects out there, one of which will certainly fit your GPU and CPU. For the geekiness of it, I run them all.
ID: 29502 \| Rating: 0 \| rate: / Reply Quote

ztmike Send message Joined: 4 Jun 09 Posts: 45 Credit: 447,355 RAC: 0	Message 29506 - Posted: 20 Aug 2009 \| 1:48:51 UTC - in response to Message 29231.
The only step left is updating the server side BOINC software, once that is done the CUDA application will be released on BOINC as a beta application. I am unable to give a time estimate for when the server side BOINC software will be upgraded because the upgrade cannot be done by me, it has to be done by the staff that controls the server that Milkyway@Home is hosted on. Why hasn't this been done yet? How long is this Cuda app planned to be in Alpha stage?
ID: 29506 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29619 - Posted: 22 Aug 2009 \| 16:54:37 UTC - in response to Message 29506.
With the server upgrade now in place work has begun on setting up the server to be able to distribute the CUDA client to users that have access to a CUDA 1.3 card (one that supports double precision). Right now a Linux/GNU 64 Bit CUDA application is live, however, there are some issues with the NVIDIA drivers and BOINC so the application seg faults. This issue should be non-existant on Windows, therefore, over the next few days a Windows 32 bit CUDA client will be added to the server.
ID: 29619 \| Rating: 0 \| rate: / Reply Quote

[B^S] sledwina Send message Joined: 28 Aug 07 Posts: 16 Credit: 22,066,527 RAC: 0	Message 29620 - Posted: 22 Aug 2009 \| 17:19:18 UTC - in response to Message 29619.
Nice! Will there also be a 64bit Windows CUDA application? ____________ pixelicious.at - My little photoblog
ID: 29620 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 29637 - Posted: 22 Aug 2009 \| 18:38:13 UTC - in response to Message 29619.
With the server upgrade now in place work has begun on setting up the server to be able to distribute the CUDA client to users that have access to a CUDA 1.3 card (one that supports double precision). Right now a Linux/GNU 64 Bit CUDA application is live, however, there are some issues with the NVIDIA drivers and BOINC so the application seg faults. This issue should be non-existant on Windows, therefore, over the next few days a Windows 32 bit CUDA client will be added to the server. i'll be happy to do a test run for yall considering the new drivers suck for a lot of users and can't use them.. 190.62 is working fine for me on gpugrid, seti, and Einstein. ____________ I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
ID: 29637 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29662 - Posted: 23 Aug 2009 \| 16:40:38 UTC - in response to Message 29620.
Nice! Will there also be a 64bit Windows CUDA application? I'm having difficulties building 64bit Windows applications with Visual C++ Express Edition so there will only be a 32 bit application in the beginning.
ID: 29662 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 29664 - Posted: 23 Aug 2009 \| 18:29:23 UTC - in response to Message 29662.
Nice! Will there also be a 64bit Windows CUDA application? I'm having difficulties building 64bit Windows applications with Visual C++ Express Edition so there will only be a 32 bit application in the beginning. what exactly is the trouble? libraries?
ID: 29664 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29691 - Posted: 24 Aug 2009 \| 22:39:34 UTC - in response to Message 29664.
what exactly is the trouble? libraries? The CUDA compiler from NVIDIA is giving me some trouble with using the 64 bit binaries of the Visual C++ compiler (cl.exe). It complains of the wrong version, I'm researching the problem now.
ID: 29691 \| Rating: 0 \| rate: / Reply Quote

Crunch3r Volunteer developer Send message Joined: 17 Feb 08 Posts: 358 Credit: 256,958,531 RAC: 3,563	Message 29695 - Posted: 24 Aug 2009 \| 22:48:55 UTC - in response to Message 29691.
what exactly is the trouble? libraries? The CUDA compiler from NVIDIA is giving me some trouble with using the 64 bit binaries of the Visual C++ compiler (cl.exe). It complains of the wrong version, I'm researching the problem now. Visual C++ Express Edition doesn't include the 64 bit SDK. You need to download and install that to be able to compile and link a 64 bit app. Anyway, there's no use for an 64 bit CUDA app since you won't gain any speed by using 64 bits and any 64 bit windows can run 32 bit app. Problem solved, case closed :p ____________ Join BOINC United now!
ID: 29695 \| Rating: 0 \| rate: / Reply Quote

ztmike Send message Joined: 4 Jun 09 Posts: 45 Credit: 447,355 RAC: 0	Message 29696 - Posted: 24 Aug 2009 \| 23:15:45 UTC
So any news on status update for Cuda for Windows? ETA?
ID: 29696 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 29698 - Posted: 24 Aug 2009 \| 23:37:56 UTC - in response to Message 29696.
The Windows 32 bit application is complete and I have successfully downloaded it through BOINC and crunched about 5 WUs. I have deprecated it though because it is getting different results than older versions of the code, as a result of a few changes in the code repository. Right now I am waiting on a response from the team to see if it is okay if the code changes propagate to the users. As for an ETA, it will be any day now, pending on the conclusion of the code change. Work has been put off on the Windows 64 bit version and I am now concentrating on the 64/32 bit Linux version.
ID: 29698 \| Rating: 0 \| rate: / Reply Quote

zpm Send message Joined: 27 Feb 09 Posts: 41 Credit: 88,206 RAC: 0	Message 29699 - Posted: 25 Aug 2009 \| 0:08:21 UTC - in response to Message 29698.
i was looking at the apps page, what is this: MilkyWay@home Version 2 Beta Applications ____________ I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
ID: 29699 \| Rating: 0 \| rate: / Reply Quote

Bymark Send message Joined: 6 Mar 09 Posts: 48 Credit: 385,976,815 RAC: 0	Message 29719 - Posted: 25 Aug 2009 \| 19:42:19 UTC - in response to Message 29699.
What will be the supported gpu 260+ ? and what is the supported boinc version 6.47+ ? Regards Thomas ____________
ID: 29719 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 30525 - Posted: 12 Sep 2009 \| 0:14:37 UTC
The latest source code for CUDA is now up at http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v.zip
ID: 30525 \| Rating: 0 \| rate: / Reply Quote

twister@austria-national-team.at Send message Joined: 2 Aug 09 Posts: 10 Credit: 103,068,054 RAC: 0	Message 34299 - Posted: 6 Dec 2009 \| 8:10:14 UTC
I need help have problems since the restart in November and the long wu's I�m Work with 2 PC RAC 19,127.82 TOTAL 2,261,013 AuthenticAMD AMD Phenom(tm) 9950 Quad-Core Processor [AMD64 Family 16 Model 2 Stepping 3] (4 processors) [6] NVIDIA GeForce GTX 295 (895MB) driver: 19562 Microsoft Windows XP Professional x64 Edition, Service Pack 2, (05.02.3790.00) 3 Dec 2009 13:53:47 UTC and this PC: RAC 9,361.83 TOTAL 689,808 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [x86 Family 6 Model 15 Stepping 11] (4 processors) [2] NVIDIA GeForce GTX 295 (895MB) driver: 19107 Microsoft Windows XP Professional x86 Edition, Service Pack 2, (05.01.2600.00) 6 Dec 2009 7:40:27 UTC have now testet Nvidia Driver: 190.38_desktop_winxp_32bit_international_whql.exe 191.07_desktop_winxp_32bit_international_whql.exe 195.62_desktop_winxp_32bit_international_whql.exe Boinc Vers i have testet: boinc_6.6.36_windows_intelx86.exe boinc_6.10.17_windows_intelx86.exe boinc_6.10.18_windows_intelx86.exe and in the projectfolder are files: milkyway_0.21_windows_intelx86__cuda23.exe cutil32.dll cudart.dll parameter_222F5_3s_v2.txt stars222F5 All task finished, because: Completed, marked as invalid 5.30 0.02 0.00 Look here: explain Status Run time (sec) claimed credit granted credit 8529192 8194065 6 Dec 2009 7:22:41 UTC 6 Dec 2009 7:40:27 UTC Completed, marked as invalid 5.30 0.02 0.00 8500917 8167549 6 Dec 2009 6:06:22 UTC 6 Dec 2009 6:45:49 UTC Completed, marked as invalid 5.58 0.02 0.00 8500313 8166982 6 Dec 2009 6:05:11 UTC 6 Dec 2009 6:45:49 UTC Completed, marked as invalid 5.27 0.02 0.00 8495251 8162177 6 Dec 2009 5:51:48 UTC 6 Dec 2009 6:06:22 UTC Completed, marked as invalid 4.50 0.02 0.00 8494878 8161824 6 Dec 2009 5:50:41 UTC 6 Dec 2009 6:05:11 UTC Completed, marked as invalid 4.59 0.02 0.00 8490142 8157356 6 Dec 2009 5:37:56 UTC 6 Dec 2009 5:51:48 UTC Completed, marked as invalid 4.55 0.02 0.00 8489772 8157004 6 Dec 2009 5:36:49 UTC 6 Dec 2009 5:50:41 UTC Completed, marked as invalid 4.53 0.02 0.00 nice gretings from austria
ID: 34299 \| Rating: 0 \| rate: / Reply Quote

The Gas Giant Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,865,573 RAC: 0	Message 34300 - Posted: 6 Dec 2009 \| 8:19:07 UTC Last modified: 6 Dec 2009 \| 8:22:02 UTC
There appears to be a memory fault with the longer wu's with the app using over 300MB of video RAM, instead of ~30MB. If you only run 1 wu at a time it seams to help.
ID: 34300 \| Rating: 0 \| rate: / Reply Quote

twister@austria-national-team.at Send message Joined: 2 Aug 09 Posts: 10 Credit: 103,068,054 RAC: 0	Message 34303 - Posted: 6 Dec 2009 \| 9:05:06 UTC - in response to Message 34300.
There appears to be a memory fault with the longer wu's with the app using over 300MB of video RAM, instead of ~30MB. If you only run 1 wu at a time it seams to help. thanks So I set up incorrectly nothing that pleases me ;-) nice greetings
ID: 34303 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 34343 - Posted: 7 Dec 2009 \| 5:01:06 UTC
This issue should be resolved in the latest version (0.24) available on BOINC.
ID: 34343 \| Rating: 0 \| rate: / Reply Quote

cenit Send message Joined: 16 Mar 09 Posts: 58 Credit: 1,122,610 RAC: 0	Message 34398 - Posted: 9 Dec 2009 \| 0:36:53 UTC - in response to Message 34343.
This issue should be resolved in the latest version (0.24) available on BOINC. which is not available in code repository...
ID: 34398 \| Rating: 0 \| rate: / Reply Quote

Anthony Waters Send message Joined: 16 Jun 09 Posts: 85 Credit: 172,476 RAC: 0	Message 34419 - Posted: 9 Dec 2009 \| 22:48:37 UTC - in response to Message 34398.
This issue should be resolved in the latest version (0.24) available on BOINC. which is not available in code repository... Thanks for the reminder http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.24.tar http://milkyway.cs.rpi.edu/milkyway/download/code_release/mw_gpu_v0.24.zip
ID: 34419 \| Rating: 0 \| rate: / Reply Quote

Author

Message

Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 30 Aug 07
Posts: 1976
Credit: 26,480
RAC: 0

Because it looks like the server can handle a bit more work, I've uploaded the CUDA code for plain milkyway@home, and it's in the code release directory. Hopefully we'll be able to just use CUDA code for regular milkyway@home and then we wouldn't even need to start up milkywayGPU@home.

I've also added two binaries, one single and one double precision for intel-based macs running OS X. These are also in the code release directory. If anyone wants to try and make some linux and windows binaries for people to use that would be appreciated.
____________

ID: 25013 | Rating: 0 | rate: