Message boards :
Number crunching :
7970s Dont Run
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next
Author | Message |
---|---|
Send message Joined: 18 May 11 Posts: 8 Credit: 24,306,657 RAC: 0 |
I use opencl version on both my two ATI 7950 , gpu is at 96% each |
Send message Joined: 15 May 11 Posts: 1392 Credit: 5,774,953 RAC: 0 |
Hm. I took my Nvidia card out of the computer. Now I tried to use a only ati card. When I run the moo wrapper, the program will end immediately an error: distributed.net client has stopped working...etc. I use cat 13.1 drivers. It seems that when I fix one problem, I bump immediately into another problem. Problem signature: Problem Event Name: APPCRASH Application Name: dnetc.exe Application Version: 2.9111.520.0 Application Timestamp: 50902214 Fault Module Name: ntdll.dll Fault Module Version: 6.1.7601.17725 Fault Module Timestamp: 4ec49b8f Exception Code: c0000005 Exception Offset: 00038dc9 OS Version: 6.1.7601.2.1.0.256.1 Locale ID: 1035 Additional Information 1: 0a9e Additional Information 2: 0a9e372d3b4ad19135b953a78882e789 Additional Information 3: 0a9e Additional Information 4: 0a9e372d3b4ad19135b953a78882e789 I tried to get help google. I tried to run alternately two and one workunit but the result remained the same. Edit: I think i found the reason from stderr.txt <core_client_version>7.0.28</core_client_version> <![CDATA[ <message> - exit code 195 (0xc3) </message> <stderr_txt> 13:40:57 (3372): wrapper: starting v1.3.9.7 13:40:57 (3372): device: 4 x Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz [Family 6 Model 42 Stepping 7] (4.598067 GFLOPS) 13:40:57 (3372): checkpoint interval: 4875 min (task 1344000 GFLOPS, 4872 min) 13:40:57 (3372): wrapper: running dnetc.exe (-ini dnetc.ini -runoffline -multiok=1) - attempt 1/10 LLVM ERROR: Cannot select: 0x440ce78: i8 = setcc 0x440aef0, 0x440ac48, 0x440cd68 [ID=34] 0x440aef0: i32 = AMDILISD::ADD 0x440ac48, 0x440b5d8 [ID=32] 0x440ac48: i32 = AMDILISD::VEXTRACT 0x440abc0, 0x440b000 [ORD=4] [ID=23] 0x440abc0: v2i32,ch = load 0x3d8e990, 0x440a918, 0x440ab38<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x440a918: i32,ch = CopyFromReg 0x3d8e990, 0x440a890 [ORD=1] [ID=17] 0x440a890: i32 = Register %vreg14 [ORD=1] [ID=1] 0x440ab38: i32 = undef [ORD=2] [ID=2] 0x440b000: i32 = TargetConstant<2> [ORD=9] [ID=16] 0x440b5d8: i32 = srl 0x440a9a0, 0x440b550 [ORD=13] [ID=30] 0x440a9a0: i32 = AMDILISD::ADD 0x440acd0, 0x440b198 [ORD=12] [ID=27] 0x440acd0: i32 = AMDILISD::VEXTRACT 0x440b3b8, 0x440b4c8 [ORD=11] [ID=21] 0x440b3b8: v4i32,ch = llvm.AMDIL.get.global.id 0x3d8e990, 0x440b330 [ORD=10] [ID=18] 0x440b330: i32 = TargetConstant<2561> [ORD=10] [ID=6] 0x440b4c8: i32 = TargetConstant<1> [ORD=8] [ID=15] 0x440b198: i32 = AMDILISD::VEXTRACT 0x440abc0, 0x440b4c8 [ORD=3] [ID=24] 0x440abc0: v2i32,ch = load 0x3d8e990, 0x440a918, 0x440ab38<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x440a918: i32,ch = CopyFromReg 0x3d8e990, 0x440a890 [ORD=1] [ID=17] 0x440a890: i32 = Register %vreg14 [ORD=1] [ID=1] 0x440ab38: i32 = undef [ORD=2] [ID=2] 0x440b4c8: i32 = TargetConstant<1> [ORD=8] [ID=15] 0x440b550: i32 = Constant<8> [ORD=13] [ID=7] 0x440ac48: i32 = AMDILISD::VEXTRACT 0x440abc0, 0x440b000 [ORD=4] [ID=23] 0x440abc0: v2i32,ch = load 0x3d8e990, 0x440a918, 0x440ab38<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x440a918: i32,ch = CopyFromReg 0x3d8e990, 0x440a890 [ORD=1] [ID=17] 0x440a890: i32 = Register %vreg14 [ORD=1] [ID=1] 0x440ab38: i32 = undef [ORD=2] [ID=2] 0x440b000: i32 = TargetConstant<2> [ORD=9] [ID=16] Seems that the wrapper cannot find gpu though boinc find it. Now need to get the wrapper to recognize the graphics card. The problem is that I do not know how to do it. |
Send message Joined: 18 May 11 Posts: 8 Credit: 24,306,657 RAC: 0 |
it only works on Cat 12.10 drivers. |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,407,912 RAC: 3,236 |
it only works on Cat 12.10 drivers. I don't believe that's true, I have all/most of my machines running version 13.1 and getting no errors at all! I am running some 5770's and some 5870's and I believe ALL have been udpated to version 13.1 |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,407,912 RAC: 3,236 |
Hm. I took my Nvidia card out of the computer. Now I tried to use a only ati card. When I run the moo wrapper, the program will end immediately an error: distributed.net client has stopped working...etc. I am running Boinc version 7.0.40 or version 7.0.45 and am not seeing the same problems that you are. You can always get the latest Boinc version here: http://boinc.berkeley.edu/dl/?C=M;O=D While the change-log is here: http://boinc.berkeley.edu/dev/forum_thread.php?id=6698 |
Send message Joined: 18 May 11 Posts: 8 Credit: 24,306,657 RAC: 0 |
is to run the 7950 on Moo, I try all the version of catalyst, only the running 12.10. Checked http://moowrap.net/result.php?resultid=18814165 I am the only one to run Moo at 7950 ... for this moment :D |
Send message Joined: 15 May 11 Posts: 1392 Credit: 5,774,953 RAC: 0 |
is to run the 7950 on Moo, I try all the version of catalyst, only the running 12.10. Yes you are. I changed my cat 12.10 drivers, but I bump into problems: http://moowrap.net/result.php?resultid=18827961 When i start Moo wrapper, it says distributed.net exe has stopped working etc... Exit status 195 (0xc3) EXIT_CHILD_FAILED <core_client_version>7.0.28</core_client_version> <![CDATA[ <message> - exit code 195 (0xc3) </message> <stderr_txt> 12:18:18 (3660): wrapper: starting v1.3.9.7 12:18:18 (3660): device: 4 x Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz [Family 6 Model 42 Stepping 7] (4.551505 GFLOPS) 12:18:18 (3660): checkpoint interval: 5275 min (task 1440250 GFLOPS, 5274 min) 12:18:18 (3660): wrapper: running dnetc.exe (-ini dnetc.ini -runoffline -multiok=1) - attempt 1/10 LLVM ERROR: Cannot select: 0x46c7650: i8 = setcc 0x46c56c8, 0x46c5420, 0x46c7540 [ID=34] 0x46c56c8: i32 = AMDILISD::ADD 0x46c5420, 0x46c5db0 [ID=32] 0x46c5420: i32 = AMDILISD::VEXTRACT 0x46c5398, 0x46c57d8 [ORD=4] [ID=23] 0x46c5398: v2i32,ch = load 0x47c4ff8, 0x46c50f0, 0x46c5310<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x46c50f0: i32,ch = CopyFromReg 0x47c4ff8, 0x46c5068 [ORD=1] [ID=17] 0x46c5068: i32 = Register %vreg14 [ORD=1] [ID=1] 0x46c5310: i32 = undef [ORD=2] [ID=2] 0x46c57d8: i32 = TargetConstant<2> [ORD=9] [ID=16] 0x46c5db0: i32 = srl 0x46c5178, 0x46c5d28 [ORD=13] [ID=30] 0x46c5178: i32 = AMDILISD::ADD 0x46c54a8, 0x46c5970 [ORD=12] [ID=27] 0x46c54a8: i32 = AMDILISD::VEXTRACT 0x46c5b90, 0x46c5ca0 [ORD=11] [ID=21] 0x46c5b90: v4i32,ch = llvm.AMDIL.get.global.id 0x47c4ff8, 0x46c5b08 [ORD=10] [ID=18] 0x46c5b08: i32 = TargetConstant<2561> [ORD=10] [ID=6] 0x46c5ca0: i32 = TargetConstant<1> [ORD=8] [ID=15] 0x46c5970: i32 = AMDILISD::VEXTRACT 0x46c5398, 0x46c5ca0 [ORD=3] [ID=24] 0x46c5398: v2i32,ch = load 0x47c4ff8, 0x46c50f0, 0x46c5310<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x46c50f0: i32,ch = CopyFromReg 0x47c4ff8, 0x46c5068 [ORD=1] [ID=17] 0x46c5068: i32 = Register %vreg14 [ORD=1] [ID=1] 0x46c5310: i32 = undef [ORD=2] [ID=2] 0x46c5ca0: i32 = TargetConstant<1> [ORD=8] [ID=15] 0x46c5d28: i32 = Constant<8> [ORD=13] [ID=7] 0x46c5420: i32 = AMDILISD::VEXTRACT 0x46c5398, 0x46c57d8 [ORD=4] [ID=23] 0x46c5398: v2i32,ch = load 0x47c4ff8, 0x46c50f0, 0x46c5310<LD8[%arrayidx_v4]> [ORD=2] [ID=20] 0x46c50f0: i32,ch = CopyFromReg 0x47c4ff8, 0x46c5068 [ORD=1] [ID=17] 0x46c5068: i32 = Register %vreg14 [ORD=1] [ID=1] 0x46c5310: i32 = undef [ORD=2] [ID=2] 0x46c57d8: i32 = TargetConstant<2> [ORD=9] [ID=16] 12:18:35 (3660): input buffer 12 packets (2144 bytes), checkpoint file 1 packets (208 bytes), output buffer 0 packets (0 bytes) 12:18:35 (3660): premature exit detected, app exit status: 0xc0000005 |
Send message Joined: 17 Nov 11 Posts: 7 Credit: 502,178,011 RAC: 0 |
Will ATI HD7000 series cards ever work at moo? Any update on why the massive delay? |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
LLVM ERROR: Cannot select: 0x440ce78: i8 = setcc 0x440aef0, 0x440ac48, 0x440cd68 [ID=34] Actually, it's not the wrapper but the D.net Client that's failing for some reason. Looks like it can't compile the OpenCL Kernel so it can be a ATI driver problem (like the client doesn't support that exact version or there's a bug in driver or the client). If that still happens with latest beta client (and especially if you can reproduce it outside BOINC) you could file a bug at upstream for their devs to take a look (there might be a bug there open for such a thing). However, the wrapper does need some changes to know about OpenCL client types and that's where I've currently stalled due to simply failing to find any time/energy/something in addition to my daily job (and fixing some fires like DB crashing). :( -w |
Send message Joined: 3 May 11 Posts: 1 Credit: 1,102,650,910 RAC: 9,949 |
LLVM ERROR: Cannot select: 0x440ce78: i8 = setcc 0x440aef0, 0x440ac48, 0x440cd68 [ID=34] I am sorry Teemu, but I do not buy this. The problem with HD7970 not running on your project has existed now for over twelve months. I do not believe that somebody could not have asked other projects what to be done, since yours is the only one that do not run this GPU. As a volunteer that spent a lot of money on HD7970, I am disappointed, and please do not blame it on ATI or AMD as all other projects runs it quite fine...... |
Send message Joined: 25 Jan 13 Posts: 2 Credit: 377,241,815 RAC: 0 |
I am a total novice at GPU processing but from what I read here is that Teemu needs to get together with Benson and figure this thing out. His (Benson's) runs and if it takes someone to interrupt, well we have people from all over the world running around here. I am in the process of buying more GPU's and hate to buy things that will be extinct in a short period of time just so I can run Moo. |
Send message Joined: 18 May 11 Posts: 8 Credit: 24,306,657 RAC: 0 |
I am ready to help. if anyone wants more information ;) |
Send message Joined: 15 May 11 Posts: 1392 Credit: 5,774,953 RAC: 0 |
I try again and here is stderr.txt: 13:03:26 (3776): Can't open init data file - running in standalone mode 13:03:26 (3776): wrapper: starting v1.3.9.7 13:03:26 (3776): can't open job file job.xml 13:03:26 (3776): can't parse job file: -108 13:03:26 (3776): called boinc_finish 17:24:18 (4712): Can't open init data file - running in standalone mode 17:24:18 (4712): wrapper: starting v1.3.9.7 17:24:18 (4712): can't open job file job.xml 17:24:18 (4712): can't parse job file: -108 17:24:18 (4712): called boinc_finish 19:36:12 (3184): Can't open init data file - running in standalone mode 19:36:12 (3184): wrapper: starting v1.3.9.7 19:36:12 (3184): can't open job file job.xml 19:36:12 (3184): can't parse job file: -108 19:36:12 (3184): called boinc_finish one question: what is job.xml and where can I find it? |
Send message Joined: 18 May 11 Posts: 8 Credit: 24,306,657 RAC: 0 |
this is : "job-ati14-1.00.xml" the original is given by Moo wrapper and must be changed like this : <job_desc> <task> <application>dnetc.exe</application> <checkpoint_filename>chkpoint</checkpoint_filename> <command_line>-ini dnetc.ini -runoffline -multiok=1</command_line> </task> </job_desc> |
Send message Joined: 5 Oct 12 Posts: 1 Credit: 0 RAC: 0 |
It is a driver issue --> 13.4 work without a glitch. if you really want to crunch with a 7xxx go to http://www.distributed.net/Download_prerelease grab the open cl client and update your boinc client with this one like Benson did. like on any other project keep 0.5 or 1 core free for any open cl device. nice benefit since AMD 12.10 is the new driver is 15% faster and reaches the old stream client speed 1900 MKeys @ AMD 5870 stock clock. Not to mention this only works with windows 7 or linux as open cl support is dropped for windows xp somewhere around 12.x. |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,407,912 RAC: 3,236 |
this is : "job-ati14-1.00.xml" Is this a file used by all the workunits or must it be changed after every new download of units? |
Send message Joined: 2 May 11 Posts: 10 Credit: 4,408,435 RAC: 0 |
Any news how to run it on the 79xx serie? |
Send message Joined: 20 Jul 12 Posts: 1 Credit: 2,304 RAC: 0 |
Using the 13.4 driver, my 7870xt (rebranded 7950) errors out on all jobs within 35 seconds or so. I would have tried downloading the beta, but I'm crunching on 64 bit Windows, and that is apparently not a supported platform. |
Send message Joined: 10 May 11 Posts: 23 Credit: 293,912,578 RAC: 0 |
I've tried the suggestions posted but WU's just sit, ready to run. Anyone have success with the 79xx cards? |
Send message Joined: 2 May 11 Posts: 10 Credit: 4,408,435 RAC: 0 |
I have now successful run it on HD7950 with 13.4 driver but against 5xxx/6xxx serie poor speed. It must have 30%-40% more speed increase. The client still needs some optimizations with 7xxx serie. Get now working stream client. Download Download and copy files into project folder. Thx for the infos below. Good luck! |