Sieve in Production; Large and Solo deprecated
log in

Advanced search

Message boards : News : Sieve in Production; Large and Solo deprecated

Previous · 1 · 2 · 3 · 4 · 5 · Next
Author Message
Profile The Ancient One
Send message
Joined: 5 May 10
Posts: 23
Credit: 91,574,211
RAC: 403
Message 21512 - Posted: 13 Oct 2015, 13:26:16 UTC - in response to Message 21509.
Last modified: 13 Oct 2015, 13:27:18 UTC

Hi,
Drivers for AMD 13.1 will work on Win 10. I have ATI HD 4600 and runs fine. See my posts for further details of GPU.
____________

Helix Von Smelix
Send message
Joined: 2 Aug 10
Posts: 43
Credit: 10,196,603,866
RAC: 1,436,775
Message 21513 - Posted: 13 Oct 2015, 13:32:11 UTC - in response to Message 21505.

Looking good here. Up and down seems stable.
Many thanks for your efforts :-)

Tackleway
Send message
Joined: 29 Sep 13
Posts: 53
Credit: 1,745,324,487
RAC: 1,748,993
Message 21515 - Posted: 13 Oct 2015, 18:06:36 UTC - in response to Message 21513.

Same here, very many thanks :-)
____________

Profile Skivelitis2
Avatar
Send message
Joined: 28 Mar 15
Posts: 17
Credit: 250,891,877
RAC: 1,178,297
Message 21516 - Posted: 13 Oct 2015, 18:45:48 UTC

Yes, things are improving. Can actually access the server status page now.
Good work!
____________

cyrusNGC_224@P3D
Send message
Joined: 5 May 14
Posts: 7
Credit: 2,376,308,934
RAC: 836,775
Message 21517 - Posted: 13 Oct 2015, 19:16:56 UTC
Last modified: 13 Oct 2015, 19:29:47 UTC

Since the new sieve app all my Linux hosts failed with AMD GPUs.

Debian Jessie (8.2)
Linux Kernel 3.16.0-4-amd64
AMD fglrx driver 14.9+ga14.201-2

AMD Kalindi (657MB) OpenCL: 1.02

<core_client_version>7.4.23</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Collatz Conjecture Sieve 1.21 Linux x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
LLVM ERROR: Cannot select: 0x2d23990: i32 = setcc 0x2d46b50, 0x2d29b30, 0x2d23890 [ORD=14] [ID=50]
0x2d46b50: i64 = add 0x2d29b30, 0x2d2a830 [ORD=10] [ID=46]
0x2d29b30: i64 = AMDILISD::VEXTRACT 0x2d76c90, 0x2d76390 [ORD=9] [ID=36]
0x2d76c90: v2i64,ch = CopyFromReg 0x29ffa40, 0x2d76b90 [ORD=9] [ID=30]
0x2d76b90: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x2d2a830: i64 = add 0x2d2a530, 0x2d2a730 [ORD=8] [ID=45]
0x2d2a530: i64,ch = load 0x29ffa40, 0x2d2a030, 0x2d2a130<LD4[%arrayidx], zext from i32> [ORD=5] [ID=44]
0x2d2a030: i32 = add 0x2d76690, 0x2d29f30 [ORD=3] [ID=43]
0x2d76690: i32,ch = CopyFromReg 0x29ffa40, 0x2d76590 [ORD=3] [ID=29]
0x2d76590: i32 = Register %vreg35 [ORD=3] [ID=1]
0x2d29f30: i32 = shl 0x2d23390, 0x2d29e30 [ORD=3] [ID=42]
0x2d23390: i32 = AMDILISD::VEXTRACT 0x2d77090, 0x2d76390 [ORD=2] [ID=38]
0x2d77090: v4i32 = llvm.AMDIL.get.global.id 0x2d76f90 [ORD=1] [ID=32]
0x2d76f90: i32 = TargetConstant<2976> [ORD=1] [ID=4]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x2d29e30: i32 = Constant<2> [ORD=3] [ID=7]
0x2d2a130: i32 = undef [ORD=4] [ID=8]
0x2d2a730: i64 = shl 0x2d2a430, 0x2d2a630 [ORD=7] [ID=40]
0x2d2a430: i64 = zero_extend 0x2d76e90 [ORD=6] [ID=37]
0x2d76e90: i32,ch = CopyFromReg 0x29ffa40, 0x2d76d90 [ORD=6] [ID=31]
0x2d76d90: i32 = Register %vreg39 [ORD=6] [ID=3]
0x2d2a630: i32 = Constant<26> [ORD=7] [ID=9]
0x2d29b30: i64 = AMDILISD::VEXTRACT 0x2d76c90, 0x2d76390 [ORD=9] [ID=36]
0x2d76c90: v2i64,ch = CopyFromReg 0x29ffa40, 0x2d76b90 [ORD=9] [ID=30]
0x2d76b90: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
In function: __OpenCL_kernelSteps_kernel

</stderr_txt>
]]>


R9 280X series (Tahiti) (3072MB) driver: 1.4.1848 OpenCL: 1.02
<core_client_version>7.4.23</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Collatz Conjecture Sieve 1.21 Linux x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
LLVM ERROR: Cannot select: 0x21b34d0: i32 = setcc 0x1e129c0, 0x1e10fb0, 0x21b33d0 [ORD=14] [ID=50]
0x1e129c0: i64 = add 0x1e10fb0, 0x1e11cb0 [ORD=10] [ID=46]
0x1e10fb0: i64 = AMDILISD::VEXTRACT 0x1e1ff20, 0x1e1f620 [ORD=9] [ID=36]
0x1e1ff20: v2i64,ch = CopyFromReg 0x1dd1f50, 0x1e1fe20 [ORD=9] [ID=30]
0x1e1fe20: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x1e11cb0: i64 = add 0x1e119b0, 0x1e11bb0 [ORD=8] [ID=45]
0x1e119b0: i64,ch = load 0x1dd1f50, 0x1e114b0, 0x1e115b0<LD4[%arrayidx], zext from i32> [ORD=5] [ID=44]
0x1e114b0: i32 = add 0x1e1f920, 0x1e113b0 [ORD=3] [ID=43]
0x1e1f920: i32,ch = CopyFromReg 0x1dd1f50, 0x1e1f820 [ORD=3] [ID=29]
0x1e1f820: i32 = Register %vreg35 [ORD=3] [ID=1]
0x1e113b0: i32 = shl 0x21b2ed0, 0x1e112b0 [ORD=3] [ID=42]
0x21b2ed0: i32 = AMDILISD::VEXTRACT 0x1e20320, 0x1e1f620 [ORD=2] [ID=38]
0x1e20320: v4i32 = llvm.AMDIL.get.global.id 0x1e20220 [ORD=1] [ID=32]
0x1e20220: i32 = TargetConstant<2976> [ORD=1] [ID=4]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x1e112b0: i32 = Constant<2> [ORD=3] [ID=7]
0x1e115b0: i32 = undef [ORD=4] [ID=8]
0x1e11bb0: i64 = shl 0x1e118b0, 0x1e11ab0 [ORD=7] [ID=40]
0x1e118b0: i64 = zero_extend 0x1e20120 [ORD=6] [ID=37]
0x1e20120: i32,ch = CopyFromReg 0x1dd1f50, 0x1e20020 [ORD=6] [ID=31]
0x1e20020: i32 = Register %vreg39 [ORD=6] [ID=3]
0x1e11ab0: i32 = Constant<26> [ORD=7] [ID=9]
0x1e10fb0: i64 = AMDILISD::VEXTRACT 0x1e1ff20, 0x1e1f620 [ORD=9] [ID=36]
0x1e1ff20: v2i64,ch = CopyFromReg 0x1dd1f50, 0x1e1fe20 [ORD=9] [ID=30]
0x1e1fe20: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
In function: __OpenCL_kernelSteps_kernel

</stderr_txt>
]]>


edit: corrected output.

woohoo
Send message
Joined: 2 Oct 15
Posts: 8
Credit: 1,741,273,347
RAC: 1,354,172
Message 21518 - Posted: 13 Oct 2015, 19:18:23 UTC - in response to Message 21512.

i ended up going back to windows 8.1 from windows 10 on one of my computers because i wanted to keep using 15.7.1 but windows update would overwrite that with the latest beta driver which would cause a reboot every time i tried turning crossfire off.

if your computer is running 13.1 then the driver should show up as 1.4.1741. but in reality it's showing up as 1.4.1734 which would make that driver earlier than 12.6. so maybe you're having a similar problem to mine where you install 13.1 and later the OS overwrites that with a different driver.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 21519 - Posted: 13 Oct 2015, 20:29:42 UTC - in response to Message 21517.

Since the new sieve app all my Linux hosts failed with AMD GPUs.

Debian Jessie (8.2)
Linux Kernel 3.16.0-4-amd64
AMD fglrx driver 14.9+ga14.201-2

AMD Kalindi (657MB) OpenCL: 1.02
<core_client_version>7.4.23</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Collatz Conjecture Sieve 1.21 Linux x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
LLVM ERROR: Cannot select: 0x2d23990: i32 = setcc 0x2d46b50, 0x2d29b30, 0x2d23890 [ORD=14] [ID=50]
0x2d46b50: i64 = add 0x2d29b30, 0x2d2a830 [ORD=10] [ID=46]
0x2d29b30: i64 = AMDILISD::VEXTRACT 0x2d76c90, 0x2d76390 [ORD=9] [ID=36]
0x2d76c90: v2i64,ch = CopyFromReg 0x29ffa40, 0x2d76b90 [ORD=9] [ID=30]
0x2d76b90: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x2d2a830: i64 = add 0x2d2a530, 0x2d2a730 [ORD=8] [ID=45]
0x2d2a530: i64,ch = load 0x29ffa40, 0x2d2a030, 0x2d2a130<LD4[%arrayidx], zext from i32> [ORD=5] [ID=44]
0x2d2a030: i32 = add 0x2d76690, 0x2d29f30 [ORD=3] [ID=43]
0x2d76690: i32,ch = CopyFromReg 0x29ffa40, 0x2d76590 [ORD=3] [ID=29]
0x2d76590: i32 = Register %vreg35 [ORD=3] [ID=1]
0x2d29f30: i32 = shl 0x2d23390, 0x2d29e30 [ORD=3] [ID=42]
0x2d23390: i32 = AMDILISD::VEXTRACT 0x2d77090, 0x2d76390 [ORD=2] [ID=38]
0x2d77090: v4i32 = llvm.AMDIL.get.global.id 0x2d76f90 [ORD=1] [ID=32]
0x2d76f90: i32 = TargetConstant<2976> [ORD=1] [ID=4]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x2d29e30: i32 = Constant<2> [ORD=3] [ID=7]
0x2d2a130: i32 = undef [ORD=4] [ID=8]
0x2d2a730: i64 = shl 0x2d2a430, 0x2d2a630 [ORD=7] [ID=40]
0x2d2a430: i64 = zero_extend 0x2d76e90 [ORD=6] [ID=37]
0x2d76e90: i32,ch = CopyFromReg 0x29ffa40, 0x2d76d90 [ORD=6] [ID=31]
0x2d76d90: i32 = Register %vreg39 [ORD=6] [ID=3]
0x2d2a630: i32 = Constant<26> [ORD=7] [ID=9]
0x2d29b30: i64 = AMDILISD::VEXTRACT 0x2d76c90, 0x2d76390 [ORD=9] [ID=36]
0x2d76c90: v2i64,ch = CopyFromReg 0x29ffa40, 0x2d76b90 [ORD=9] [ID=30]
0x2d76b90: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x2d76390: i32 = TargetConstant<1> [ORD=2] [ID=21]
In function: __OpenCL_kernelSteps_kernel

</stderr_txt>
]]>


R9 280X series (Tahiti) (3072MB) driver: 1.4.1848 OpenCL: 1.02
<core_client_version>7.4.23</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Collatz Conjecture Sieve 1.21 Linux x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
LLVM ERROR: Cannot select: 0x21b34d0: i32 = setcc 0x1e129c0, 0x1e10fb0, 0x21b33d0 [ORD=14] [ID=50]
0x1e129c0: i64 = add 0x1e10fb0, 0x1e11cb0 [ORD=10] [ID=46]
0x1e10fb0: i64 = AMDILISD::VEXTRACT 0x1e1ff20, 0x1e1f620 [ORD=9] [ID=36]
0x1e1ff20: v2i64,ch = CopyFromReg 0x1dd1f50, 0x1e1fe20 [ORD=9] [ID=30]
0x1e1fe20: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x1e11cb0: i64 = add 0x1e119b0, 0x1e11bb0 [ORD=8] [ID=45]
0x1e119b0: i64,ch = load 0x1dd1f50, 0x1e114b0, 0x1e115b0<LD4[%arrayidx], zext from i32> [ORD=5] [ID=44]
0x1e114b0: i32 = add 0x1e1f920, 0x1e113b0 [ORD=3] [ID=43]
0x1e1f920: i32,ch = CopyFromReg 0x1dd1f50, 0x1e1f820 [ORD=3] [ID=29]
0x1e1f820: i32 = Register %vreg35 [ORD=3] [ID=1]
0x1e113b0: i32 = shl 0x21b2ed0, 0x1e112b0 [ORD=3] [ID=42]
0x21b2ed0: i32 = AMDILISD::VEXTRACT 0x1e20320, 0x1e1f620 [ORD=2] [ID=38]
0x1e20320: v4i32 = llvm.AMDIL.get.global.id 0x1e20220 [ORD=1] [ID=32]
0x1e20220: i32 = TargetConstant<2976> [ORD=1] [ID=4]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
0x1e112b0: i32 = Constant<2> [ORD=3] [ID=7]
0x1e115b0: i32 = undef [ORD=4] [ID=8]
0x1e11bb0: i64 = shl 0x1e118b0, 0x1e11ab0 [ORD=7] [ID=40]
0x1e118b0: i64 = zero_extend 0x1e20120 [ORD=6] [ID=37]
0x1e20120: i32,ch = CopyFromReg 0x1dd1f50, 0x1e20020 [ORD=6] [ID=31]
0x1e20020: i32 = Register %vreg39 [ORD=6] [ID=3]
0x1e11ab0: i32 = Constant<26> [ORD=7] [ID=9]
0x1e10fb0: i64 = AMDILISD::VEXTRACT 0x1e1ff20, 0x1e1f620 [ORD=9] [ID=36]
0x1e1ff20: v2i64,ch = CopyFromReg 0x1dd1f50, 0x1e1fe20 [ORD=9] [ID=30]
0x1e1fe20: v2i64 = Register %vreg38 [ORD=9] [ID=2]
0x1e1f620: i32 = TargetConstant<1> [ORD=2] [ID=21]
In function: __OpenCL_kernelSteps_kernel

</stderr_txt>
]]>


edit: corrected output.


The x86_64 Linux app for AMD had a 98.2% success rate in the last 24 hours with over 5,300 WUs completed successfully. Have you looked at the top computers for one running Linux w/ AMD GPUs and look at which driver versions that are working for them?

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 21520 - Posted: 13 Oct 2015, 20:32:22 UTC - in response to Message 21516.

Yes, things are improving. Can actually access the server status page now.
Good work!


I just delete a million spam profiles and users. That allowed me to shrink the database so that it now fits into RAM again. I also changed the profile settings so people can't create a profile until after they have completed a WU successfully. reCaptcha has also been enabled so it can't be done programatically anymore either.

Almo
Send message
Joined: 26 Apr 10
Posts: 2
Credit: 461,652
RAC: 0
Message 21522 - Posted: 13 Oct 2015, 22:30:55 UTC

GPU tasks aren't working for me. Here's the info...

Boinc 7.6.12
I don't know how it was installed... I just ran the OSX installer. It starts up automatically when I start my computer.
It's a MacPro 2013 with OSX 10.10.5 on it. I believe this is a 64-bit OS. 12 GB ram, AMD FirePro D300 2048 MB
3.7 GHz Quad-Core Xeon E5

The last lines of the log show what happens when it tries to do a GPU job... it just instantly finishes. The task display in the client shows "Computation error (0.719 CPUs + 1 AMD/ATI GPU)"

Log:
Tue 13 Oct 07:58:18 2015 | | cc_config.xml not found - using defaults
Tue 13 Oct 07:58:18 2015 | | Starting BOINC client version 7.6.12 for x86_64-apple-darwin
Tue 13 Oct 07:58:18 2015 | | log flags: file_xfer, sched_ops, task
Tue 13 Oct 07:58:18 2015 | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.5 c-ares/1.10.0
Tue 13 Oct 07:58:18 2015 | | Data directory: /Library/Application Support/BOINC Data
Tue 13 Oct 07:58:18 2015 | | OpenCL: AMD/ATI GPU 0: AMD Radeon HD - FirePro D300 Compute Engine (driver version 1.2 (Jul 29 2015 02:43:15), device version OpenCL 1.2, 2048MB, 2048MB available, 1360 GFLOPS peak)
Tue 13 Oct 07:58:18 2015 | | OpenCL: AMD/ATI GPU 1: AMD Radeon HD - FirePro D300 Compute Engine (driver version 1.2 (Jul 29 2015 02:43:15), device version OpenCL 1.2, 2048MB, 2048MB available, 240 GFLOPS peak)
Tue 13 Oct 07:58:18 2015 | | OpenCL CPU: Intel(R) Xeon(R) CPU E5-1620 v2 @ 3.70GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Tue 13 Oct 07:58:18 2015 | | Host name: SilentMac.local
Tue 13 Oct 07:58:18 2015 | | Processor: 8 GenuineIntel Intel(R) Xeon(R) CPU E5-1620 v2 @ 3.70GHz [x86 Family 6 Model 62 Stepping 4]
Tue 13 Oct 07:58:18 2015 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 x2apic popcnt aes pcid xsave osxsave tsctmr avx rdrand f16c
Tue 13 Oct 07:58:18 2015 | | OS: Mac OS X 10.10.5 (Darwin 14.5.0)
Tue 13 Oct 07:58:18 2015 | | Memory: 12.00 GB physical, 185.90 GB virtual
Tue 13 Oct 07:58:18 2015 | | Disk: 465.12 GB total, 185.65 GB free
Tue 13 Oct 07:58:18 2015 | | Local time is UTC -4 hours
Tue 13 Oct 07:58:18 2015 | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 431452; resource share 100
Tue 13 Oct 07:58:18 2015 | | No general preferences found - using defaults
Tue 13 Oct 07:58:18 2015 | | Reading preferences override file
Tue 13 Oct 07:58:18 2015 | | Preferences:
Tue 13 Oct 07:58:18 2015 | | max memory usage when active: 6144.00MB
Tue 13 Oct 07:58:18 2015 | | max memory usage when idle: 11059.20MB
Tue 13 Oct 07:58:18 2015 | | max disk usage: 185.59GB
Tue 13 Oct 07:58:18 2015 | | max CPUs used: 2
Tue 13 Oct 07:58:18 2015 | | don't use GPU while active
Tue 13 Oct 07:58:18 2015 | | suspend work if non-BOINC CPU load exceeds 25%
Tue 13 Oct 07:58:18 2015 | | max download rate: 102400 bytes/sec
Tue 13 Oct 07:58:18 2015 | | max upload rate: 51200 bytes/sec
Tue 13 Oct 07:58:18 2015 | | (to change preferences, visit a project web site or select Preferences in the Manager)
Tue 13 Oct 07:58:18 2015 | | Suspending GPU computation - computer is in use
Tue 13 Oct 07:58:28 2015 | | Suspending computation - CPU is busy
Tue 13 Oct 07:58:38 2015 | | Resuming computation
Tue 13 Oct 08:36:50 2015 | | Resuming GPU computation
Tue 13 Oct 08:36:50 2015 | Collatz Conjecture | Starting task collatz_sieve_2410893134167868964864_6597069766656_0
Tue 13 Oct 08:37:03 2015 | Collatz Conjecture | Computation for task collatz_sieve_2410893134167868964864_6597069766656_0 finished

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 21526 - Posted: 13 Oct 2015, 23:49:14 UTC - in response to Message 21522.

The short answer is that AMD on OS X is very buggy and the bugs are in the AMD code.

The error listed in the WU's error log is:

clBuildProgram() failed with error (-11)


I've given up on getting either Apple or AMD to fix it. The problem has been around for a couple years and the bug reports I've submitted have been ignored. The same exact app works fine for some AMD GPUs and it works for all nVidia GPUs. The AMD OpenCL runtime compiler is definitely the culprit.

Almo
Send message
Joined: 26 Apr 10
Posts: 2
Credit: 461,652
RAC: 0
Message 21530 - Posted: 14 Oct 2015, 0:06:19 UTC - in response to Message 21526.

Ok, thanks for the info. :)

adrian
Avatar
Send message
Joined: 17 Mar 12
Posts: 8
Credit: 287,464,567
RAC: 0
Message 21549 - Posted: 18 Oct 2015, 0:43:47 UTC

Have quite a few old AMD ATI Radeon HD 2300/2400/3200/4200 (RV610) (256MB) driver: 1.4.1703 which been ok with Solo and large before recent upgrades.
Getting nothing since, nVidias seems ok.
Any way to get it back working or forget it?

Profile The Ancient One
Send message
Joined: 5 May 10
Posts: 23
Credit: 91,574,211
RAC: 403
Message 21555 - Posted: 19 Oct 2015, 0:33:55 UTC

Sorry mate, You need to revise your program as it is not running as intended. Still not getting any WU. P.S. have 2008,2010,2012 and 2013 C++.
____________

Profile The Ancient One
Send message
Joined: 5 May 10
Posts: 23
Credit: 91,574,211
RAC: 403
Message 21556 - Posted: 19 Oct 2015, 0:33:59 UTC

Sorry mate, You need to revise your program as it is not running as intended. Still not getting any WU. P.S. have 2008,2010,2012 and 2013 C++.
____________

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 21562 - Posted: 19 Oct 2015, 15:06:05 UTC - in response to Message 21555.

Sorry mate, You need to revise your program as it is not running as intended. Still not getting any WU. P.S. have 2008,2010,2012 and 2013 C++.


You likely aren't getting work because there are no CAL/Brook+ apps for the sieve app. AMD switched from CAL/Brook+ to OpenCL back in 2011, having done no new development on it for since 2008 -- and not having fixed the bugs that were present since 2005.

Unlike nVidia which still provides OpenCL drivers for GPUs made since 2006, AMD did not bother with backwards compatibility when they switched to OpenCL. To continue using the old apps makes no sense because the new sieve app runs 70 times faster (or rather, doesn't 1/70 the work by eliminating the numbers via the sieve which allows it to check 70 times more numbers per WU) and the old app was producing invalid results because of a latent bug that only recently started showing up -- likely because the numbers being checked were greater than a given threshold.

If someone wants to fix the bugs in the existing Brook compiler and make it compatible with current development tools, I'd be willing adding a new Brook+ app. You can get the Brook+ source code at http://sourceforge.net/projects/brookplus/

Helix Von Smelix
Send message
Joined: 2 Aug 10
Posts: 43
Credit: 10,196,603,866
RAC: 1,436,775
Message 21565 - Posted: 19 Oct 2015, 18:46:20 UTC

looks like it has settled down very nicely now. Many thanks for putting up with our grumbling!

Profile BarryAZ
Send message
Joined: 21 Aug 09
Posts: 251
Credit: 13,343,709,803
RAC: 23,639,488
Message 21587 - Posted: 22 Oct 2015, 18:01:23 UTC - in response to Message 21562.
Last modified: 22 Oct 2015, 18:07:23 UTC

Slicker, I am still a bit at a loss with some systems working fine, and others sending off mass computation errors -- with seemingly the same environment.

I have some ATI video cards -- the 6750 and up run fine, and I realize the older legacy cards (4850's) are pretty much hopeless. Not an issue here

But I have a batch of Nvidia systems.

All of these computers are running Windows 7 64 bit.

All are running BOINC 7.6.9

All have Open CL 1.02

The video cards include 650TI, 750 and 750TI

The 650TI -- no problem

The problem I have is that *some* of the 750 and 750ti systems spin off fast computation errors. I've done both Collatz resets, and delete, reattach and reboot on these these systems to no avail.

I'd note the systems generating those errors are happily running GPUGrid work.

The confusing thing is that other systems, virtually identical are very happy with the Collatz application and sieve units and are steadily pumping away.

Is there something I should look for here? Again, all the Windows updates and deploying the latest BOINC, and Nvidia drivers, rebooting, etc. don't resolve those recalcitrant computers. I've simply removed Collatz from those computers -- though any time I try again with one of these bad boys, I generate a quick comp error, then halt processing and remove so as to not flood you with the rejects.

Thanks

Profile BarryAZ
Send message
Joined: 21 Aug 09
Posts: 251
Credit: 13,343,709,803
RAC: 23,639,488
Message 21613 - Posted: 27 Oct 2015, 0:30:07 UTC - in response to Message 21587.

OK -- I had thought I updated the C++ support -- I apparently did not have current versions on some of the computers.

Problem solved.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 21617 - Posted: 27 Oct 2015, 13:22:53 UTC - in response to Message 21565.

looks like it has settled down very nicely now. Many thanks for putting up with our grumbling!


One man's grumbling is another man's call to action. When one person complains, it is likely a setup issue on that machine. When lots of people complain, I need to get my act together because the problem is likely on my end. In this case, it was a conglomeration of things. Spam accounts and spam profiles increased the database size to the point where it would not fit in RAM. Add to that the thousands of WUs that were erroring out making it even larger. Finally, the logic in the work generator was such that when the database timed out, it was assumed more work was needed even when the timeout was caused because there was too much work already. Add them all together and things ground to a halt and it wasn't until all of them were fixed that things started getting back to normal.

cuBerBruce
Send message
Joined: 25 Aug 13
Posts: 4
Credit: 6,007,222
RAC: 0
Message 21653 - Posted: 2 Nov 2015, 18:27:42 UTC

Ever since solo_collatz was removed, I haven't been able to get any WUs. The System Requirements is now saying HD 4800 or later. Did you deliberately drop support of HD 4350? Why?

Previous · 1 · 2 · 3 · 4 · 5 · Next
Post to thread

Message boards : News : Sieve in Production; Large and Solo deprecated


Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.