Page 6 of 13
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 2:19 am
by k1wi
-platform and -platformId are two separate concepts.
The reason why you're getting the warning "Warning: Using OpenCL platform but no platformId specified, setting platformId=0" is because proteneer added the feature in v0.3 to enable us to benchmark heterogeneous systems. Previously you could not set whether you wanted to use the AMD implementation or the NVidia one. So the hard default was to only use platformId '0', which in both of our cases was the AMD implementation.
I am not entirely sure why the AMD drivers allow OpenCL on the CPU and Nvidia doesn't, I just know as far back as when F@H released memtestCL it has. It might be because they have been committed to OpenCL longer than NVidia (who have been committed to CUDA.)
So, to recap:
-platform tells the benchmark app that you want to use either CUDA or OpenCL (depending on which you choose)
If you chose OpenCL then you need -platformId because you have more than one implementation of OpenCL (AMD and NVidia) and the app needs to know whether to use platformId 0 and platformId 1, as per:
Code: Select all
[2] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 AMD-APP (831.4)
NAME = AMD Accelerated Parallel Processing
VENDOR = Advanced Micro Devices, Inc.
-- 1 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
Once you have selected platform 0 (AMD) or platform 1 (NVidia) then you need to select which device you want to use, based on:
Code: Select all
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Juniper
DEVICE_VENDOR = Advanced Micro Devices, Inc.
DEVICE_VERSION = OpenCL 1.1 AMD-APP (831.4)
-- 1 --
DEVICE_NAME = Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz
DEVICE_VENDOR = GenuineIntel
DEVICE_VERSION = OpenCL 1.1 AMD-APP (831.4)
(1) device(s) found on platform 1:
-- 0 --
DEVICE_NAME = GeForce GTX 570
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
So, using
OpenCL:
On platformId '0'
-platformId 0 -deviceId 0 is your AMD GPU,
-platformId 0 -deviceId 1 is your CPU,
On platformId '1':
-platformId 1 -deviceId 0 is your NVidia GPU
ADD: You don't need platformId when using CUDA because it doesn't support it. I could be wrong but it should never be an issue because only NVidia does CUDA, so you shouldn't have more than one set of drivers?
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 2:24 am
by PinHead
Makes sense k1wi,
Seems I have a new set of minimum required switches.
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 2:39 am
by Zagen30
I figured out why FAHBench was crashing on the OpenCL tests. My setup is that I use the Intel HD4000 in my 3770k to drive my monitor, while my pair of GTX 460s are completely dedicated to folding, and apparently this throws the current iteration (0.3) of FAHBench for a loop. When I had one of my 460s drive the monitor, the OpenCL tests ran fine.
I did try using -platformId when the HD4000 was the primary GPU, but that didn't solve the problem. In this setup, the HD4000 is detected on platform 0 while the 460s are on platform 1. The OpenCL test fails immediately on the HD4000. When I specified platformId 1 and either of the 460s, the test would get as far as "Checking for accuracy..." on the explicit test before crashing.
Interestingly, the CUDA tests do not encounter any issues with the Intel chip driving the display. I also do not see the CPU itself as a valid device in any setup.
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 2:49 am
by k1wi
Zagen30 wrote:I figured out why FAHBench was crashing on the OpenCL tests. My setup is that I use the Intel HD4000 in my 3770k to drive my monitor, while my pair of GTX 460s are completely dedicated to folding, and apparently this throws the current iteration (0.3) of FAHBench for a loop. When I had one of my 460s drive the monitor, the OpenCL tests ran fine.
I did try using -platformId when the HD4000 was the primary GPU, but that didn't solve the problem. In this setup, the HD4000 is detected on platform 0 while the 460s are on platform 1. The OpenCL test fails immediately on the HD4000. When I specified platformId 1 and either of the 460s, the test would get as far as "Checking for accuracy..." on the explicit test before crashing.
Interestingly, the CUDA tests do not encounter any issues with the Intel chip driving the display. I also do not see the CPU itself as a valid device in any setup.
what does 'FAHBench.exe --display-devices' show?
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 3:50 am
by Zagen30
With the HD4000 as display driver:
Code: Select all
[2] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1
NAME = Intel(R) OpenCL
VENDOR = Intel(R) Corporation
-- 1 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(1) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Intel(R) HD Graphics 4000
DEVICE_VENDOR = Intel(R) Corporation
DEVICE_VERSION = OpenCL 1.1
(2) device(s) found on platform 1:
-- 0 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
-- 1 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
Invalid Platform (please use either OpenCL or CUDA)
Without:
Code: Select all
[1] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
-- 1 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
Invalid Platform (please use either OpenCL or CUDA)
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 4:21 am
by proteneer
Zagen30 wrote:With the HD4000 as display driver:
Code: Select all
[2] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1
NAME = Intel(R) OpenCL
VENDOR = Intel(R) Corporation
-- 1 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(1) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Intel(R) HD Graphics 4000
DEVICE_VENDOR = Intel(R) Corporation
DEVICE_VERSION = OpenCL 1.1
(2) device(s) found on platform 1:
-- 0 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
-- 1 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
Invalid Platform (please use either OpenCL or CUDA)
Without:
Code: Select all
[1] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
-- 1 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
Invalid Platform (please use either OpenCL or CUDA)
Hi Zagen - can you please try the 2013 Intel OpenCL Beta Drivers? We're actually quite curious as to how the HD4000 series stack up. The 2012 drivers, AFAIK, is somewhat buggy still.
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 8:02 am
by Zagen30
I installed the beta drivers, 9.17.10.2932, and the OpenCL 2013 beta SDK.
The main page for it suggested driver version 9.17.10.2792, but I think that's because it was last updated in August or so. If the newest drivers don't work, I can look for ones that do.
The net result is that display-devices now lists the OpenCL platform as being on version 1.2, and the CPU is now visible:
Code: Select all
[2] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.2
NAME = Intel(R) OpenCL
VENDOR = Intel(R) Corporation
-- 1 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz
DEVICE_VENDOR = Intel(R) Corporation
DEVICE_VERSION = OpenCL 1.2 (Build 42537)
-- 1 --
DEVICE_NAME = Intel(R) HD Graphics 4000
DEVICE_VENDOR = Intel(R) Corporation
DEVICE_VERSION = OpenCL 1.1
(2) device(s) found on platform 1:
-- 0 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
-- 1 --
DEVICE_NAME = GeForce GTX 460
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
I still can't get it to run, however. Running the bench on the CPU nets me the following, followed by a standard Windows report about FAHBench.exe crashing:
Code: Select all
Explicit:
Checking for accuracy...0x66B09BC8 (0x02797080 0x0279FE80 0x00000000 0x00000000)
, clUnloadCompiler() + 0x10C48 bytes(s)
0x66B00376 (0x02797080 0x0279FE80 0x197F8E88 0x00000001), clUnloadCompiler() + 0
x73F6 bytes(s)
0x66B19D07 (0x0279FE80 0x197F8E88 0x00000001 0x00000000), clEnqueueWriteBuffer()
+ 0x67 bytes(s)
0x6C362BCB (0x05BB3F10 0x00000001 0x2CEA61FB 0x00000014), ?upload@OpenCLArray@Op
enMM@@QAEXPBX_N@Z() + 0x6B bytes(s)
0x6C36C7EC (0x00000002 0x2CEA7E1F 0x00000000 0x0000000F), ?tagAtomsInMolecule@Op
enCLContext@OpenMM@@CAXHHAAV?$vector@HV?$allocator@H@std@@@std@@AAV?$vector@V?$v
ector@HV?$allocator@H@std@@@std@@V?$allocator@V?$vector@HV?$allocator@H@std@@@st
d@@@2@@4@@Z() + 0x40C bytes(s)
0x779BE38C (0x0199AE48 0x00000000 0x00000001 0x00D3F414), RtlInitUnicodeString()
+ 0x164 bytes(s)
0x6C3EA1DE (0x00000015 0x00000000 0x00000080 0x00000020), ??0PlatformData@OpenCL
Platform@OpenMM@@QAE@ABVSystem@2@ABV?$basic_string@DU?$char_traits@D@std@@V?$all
ocator@D@2@@std@@11@Z() + 0x30E bytes(s)
0x6ECC63B5 (0x779C3CEE 0x00000068 0x00000070 0x05F4E10A), ?assign@?$basic_string
@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@QAEAAV12@ABV12@II@Z() + 0x8B byte
s(s)
0x779C3CC3 (0x00000068 0x00000070 0x05F4E10A 0x05F4E108), RtlImageNtHeader() + 0
xB3F bytes(s)
0x779C3CEE (0x00D3F418 0x0000000F 0x037C33DC 0x00000006), RtlImageNtHeader() + 0
xB6A bytes(s)
0x6ECC4F2A (0x00D3F418 0x0000000F 0x037C33DC 0x00000006), ?_Copy_s@?$char_traits
@D@std@@SAPADPADIPBDI@Z() + 0x17 bytes(s)
0x6ECC5242 (0x00D3F418 0x0000000F 0x037C33DC 0x00000003), ?deallocate@?$allocato
r@_W@std@@QAEXPA_WI@Z() + 0x1C1 bytes(s)
0x6ECC59D5 (0x00EB0000 0x00000000 0x00000068 0x6FC3CCC9), ?allocate@?$allocator@
D@std@@QAEPADI@Z() + 0xAC bytes(s)
0x6FC33DB8 (0x0199AE48 0x037C35B8 0x037C3568 0x00D3F414), malloc() + 0x79 bytes(
s)
0x6C3EA6AA (0x00EB9A14 0x07BFC0E4 0x6C4CD96A 0x02CC7A50), ?contextCreated@OpenCL
Platform@OpenMM@@UBEXAAVContextImpl@2@ABV?$map@V?$basic_string@DU?$char_traits@D
@std@@V?$allocator@D@2@@std@@V12@U?$less@V?$basic_string@DU?$char_traits@D@std@@
V?$allocator@D@2@@std@@@2@V?$allocator@U?$pair@$$CBV?$basic_string@DU?$char_trai
ts@D@std@@V?$allocator@D@2@@std@@V12@@std@@@2@@std@@@Z() + 0x22A bytes(s)
0x6ECD5984 (0x02CC7A50 0x037C2690 0x02CC7A50 0x037C2690), ??$?MDU?$char_traits@D
@std@@V?$allocator@D@1@@std@@YA_NABV?$basic_string@DU?$char_traits@D@std@@V?$all
ocator@D@2@@0@0@Z() + 0x10 bytes(s)
0x6C4CD96A (0x00000001 0x6C496DC8 0x02FE0F50 0x00D3F69C), ?getNextToken@Parser@L
epton@@CA?AVParseToken@2@ABV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D
@2@@std@@H@Z() + 0x12CA bytes(s)
0x6C3F7E54 (0x08B48450 0x0199AE48 0x09CC88E0 0x04E603E0), ?contextDestroyed@Open
CLPlatform@OpenMM@@UBEXAAVContextImpl@2@@Z() + 0xD754 bytes(s)
0x6C49417D (0x00000001 0x010B4E4E 0x0199AE48 0x09CC88E0), ??0Context@OpenMM@@QAE
@AAVSystem@1@AAVIntegrator@1@AAVPlatform@1@ABV?$map@V?$basic_string@DU?$char_tra
its@D@std@@V?$allocator@D@2@@std@@V12@U?$less@V?$basic_string@DU?$char_traits@D@
std@@V?$allocator@D@2@@std@@@2@V?$allocator@U?$pair@$$CBV?$basic_string@DU?$char
_traits@D@std@@V?$allocator@D@2@@std@@V12@@std@@@2@@std@@@Z() + 0x6D bytes(s)
0x6C51BD36 (0x00D3F728 0x00D3F690 0x6E65704F 0x2C004C43), ?compareWithReferenceP
latform@ValidateOpenMMForces@OpenMM@@QAEHAAVContext@2@PAV?$basic_string@DU?$char
_traits@D@std@@V?$allocator@D@2@@std@@@Z() + 0x2186 bytes(s)
0x010B51B1 (0x00D3F648 0xFFFFFFFF 0x010C12E8 0x00D3F728)
0x010B3691 (0x010BC05C 0x010BC080 0x010BBF88 0x24C2A031)
0x010B3E97 (0x00000001 0x555C3A43 0x73726573 0x6C794B5C)
0x010B9A7B (0x7EFDE000 0x00D3FDFC 0x779C9EF2 0x7EFDE000)
0x76D833AA (0x010B90C8 0x7EFDE000 0x00000000 0x00000000), BaseThreadInitThunk()
+ 0x12 bytes(s)
0x779C9EC5 (0x7EFDE000 0x00000000 0x00000000 0x00000000), RtlInitializeException
Chain() + 0x36 bytes(s)
0x010B90C8 (0x00000000 0x00000000 0x00000000 0x00000000)
0x7EFDE000 (0x146AFFFF 0x0BDDC868 0x01CEE801 0x65830000) <unknown module>
0xFD9DE900 (0x0BDDC868 0x01CEE801 0x65830000 0x4DFF00FC) <unknown module>
0x146AFFFF (0x01CEE801 0x65830000 0x4DFF00FC 0x8B3A7810) <unknown module>
0x0BDDC868 (0x65830000 0x4DFF00FC 0x8B3A7810 0x4D2B084D) <unknown module>
0x01CEE801 (0x4DFF00FC 0x8B3A7810 0x4D2B084D 0x084D890C) <unknown module>
0x65830000 (0x8B3A7810 0x4D2B084D 0x084D890C 0xEB1455FF), CreateFrontEndInstance
() + 0x1A8380 bytes(s)
0x4DFF00FC (0x4D2B084D 0x084D890C 0xEB1455FF 0xEC458BED) <unknown module>
0x8B3A7810 (0x084D890C 0xEB1455FF 0xEC458BED 0x8BE44589) <unknown module>
0x4D2B084D (0xEB1455FF 0xEC458BED 0x8BE44589 0x008BE445) <unknown module>
0x084D890C (0xEC458BED 0x8BE44589 0x008BE445 0x8BE04589) <unknown module>
0xEB1455FF (0x8BE44589 0x008BE445 0x8BE04589 0x3881E045) <unknown module>
0xEC458BED (0x008BE445 0x8BE04589 0x3881E045 0xE06D7363) <unknown module>
0x8BE44589 (0x8BE04589 0x3881E045 0xE06D7363 0x45C70B74) <unknown module>
0x008BE445 (0x3881E045 0xE06D7363 0x45C70B74 0x000000DC) <unknown module>
0x8BE04589 (0xE06D7363 0x45C70B74 0x000000DC 0xDC458B00) <unknown module>
0x3881E045 (0x45C70B74 0x000000DC 0xDC458B00 0x04E0E8C3) <unknown module>
0xE06D7363 (0x000000DC 0xDC458B00 0x04E0E8C3 0x658B0000) <unknown module>
0x45C70B74 (0xDC458B00 0x04E0E8C3 0x658B0000 0xFC45C7E8) <unknown module>
Running it on the GPU still results in the crash window after about 4 seconds, with nothing output to the command line.
I did notice
in the release notes some tips that I thought might have been causing the problem. It states that Visual C++ 2010 is required, so I downloaded the proper Visual Studio 2010 express package. It also mentions certain registry keys that I believe weren't correct before, but should now be what the notes indicate. Neither of those changes seemed to make a difference.
Re: Folding@Home Benchmark Beta Testing
Posted: Wed Jan 16, 2013 8:32 am
by proteneer
It looks like intel isn't quite standard compliant yet. Thanks Zagen, this is very useful for us to know.
Re: Folding@Home Benchmark Beta Testing
Posted: Thu Jan 17, 2013 12:28 am
by bruce
PinHead wrote:(not sure why the CPU was detected ) ?
Think of it this way: OpenCL is designed to let a programmer find and use any device that can do his job. It's not just for GPUs.
A CPU might have N cores that can process single/double precision which makes it generically equivalent to a GPU that also has N cores that can process single/double precision. If the programmer can obtain a list of all the possible devices that have the desired characteristics, the choice of which one to assign the work to can be based on any other information that might be available to compare them -- including performance results, how busy it is, how much latency is involved, etc. etc.
The Intel OpenCL code should detect an Intel GPU and an Intel CPU if you happen to have them, whether or not NVidia or AMD's OpenCL code supports those features. Neither one is ready to be put on the V7 Whitelist yet but that doesn't mean it will never happen -- ("not soon").
Re: Folding@Home Benchmark Beta Testing
Posted: Thu Jan 17, 2013 11:40 pm
by PinHead
Win 7 64 bit
AMD/ATI and Nvidia GPU's installed
FAHBench.exe v 0.3
I can't seem to get OpenCL to run, "Explicit: clSetKernelArg" .
I think I am forming the command line correctly!???
Code: Select all
E:\FAH_GPU_Bench\FAHBench_0_3>FAHBench.exe --display-devices --disable-splash
[2] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 AMD-APP (831.4)
NAME = AMD Accelerated Parallel Processing
VENDOR = Advanced Micro Devices, Inc.
-- 1 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 CUDA 4.2.1
NAME = NVIDIA CUDA
VENDOR = NVIDIA Corporation
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Juniper
DEVICE_VENDOR = Advanced Micro Devices, Inc.
DEVICE_VERSION = OpenCL 1.1 AMD-APP (831.4)
-- 1 --
DEVICE_NAME = Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz
DEVICE_VENDOR = GenuineIntel
DEVICE_VERSION = OpenCL 1.1 AMD-APP (831.4)
(1) device(s) found on platform 1:
-- 0 --
DEVICE_NAME = GeForce GTX 570
DEVICE_VENDOR = NVIDIA Corporation
DEVICE_VERSION = OpenCL 1.1 CUDA
Invalid Platform (please use either OpenCL or CUDA)
E:\FAH_GPU_Bench\FAHBench_0_3>FAHBench.exe -platformId 0 -deviceId 0 -platform OpenCL -precision single
O O
P R O T E N E E R C--N \ \ N
| C C=O / \-C
C / | N-C \
.C-C C/ C C | C
/ \ O | | / N |
C C | | O C C /-C
\_N_/ \ N _C_ C | / O / C
C-/ \_C/ \N-/ \ N /-C-\ C | | O /
| | C-/ \C/ N-/ \_ N\ /C\ -C N | |
O | | | \C/ C/ N/ \_C__/ \ C-\ C
C O | | | | C-/ N/ \-C
\_C C O | O | |
\ \-O C C O
| \ \
C N Folding@Home C--N C
\ | Benchmark (Beta) | |
N--C O |
\ Yutong Zhao C=O
N [email protected] /
O
for official stats, please visit www.fahbench.com
Explicit:
clSetKernelArg
E:\FAH_GPU_Bench\FAHBench_0_3>FAHBench.exe -deviceId 0 -platform OpenCL -precision single
O O
P R O T E N E E R C--N \ \ N
| C C=O / \-C
C / | N-C \
.C-C C/ C C | C
/ \ O | | / N |
C C | | O C C /-C
\_N_/ \ N _C_ C | / O / C
C-/ \_C/ \N-/ \ N /-C-\ C | | O /
| | C-/ \C/ N-/ \_ N\ /C\ -C N | |
O | | | \C/ C/ N/ \_C__/ \ C-\ C
C O | | | | C-/ N/ \-C
\_C C O | O | |
\ \-O C C O
| \ \
C N Folding@Home C--N C
\ | Benchmark (Beta) | |
N--C O |
\ Yutong Zhao C=O
N [email protected] /
O
for official stats, please visit www.fahbench.com
Warning: Using OpenCL platform but no platformId specified, setting platformId=0
Explicit:
clSetKernelArg
E:\FAH_GPU_Bench\FAHBench_0_3>FAHBench.exe -platformId 1 -deviceId 0 -platform OpenCL -precision single
O O
P R O T E N E E R C--N \ \ N
| C C=O / \-C
C / | N-C \
.C-C C/ C C | C
/ \ O | | / N |
C C | | O C C /-C
\_N_/ \ N _C_ C | / O / C
C-/ \_C/ \N-/ \ N /-C-\ C | | O /
| | C-/ \C/ N-/ \_ N\ /C\ -C N | |
O | | | \C/ C/ N/ \_C__/ \ C-\ C
C O | | | | C-/ N/ \-C
\_C C O | O | |
\ \-O C C O
| \ \
C N Folding@Home C--N C
\ | Benchmark (Beta) | |
N--C O |
\ Yutong Zhao C=O
N [email protected] /
O
for official stats, please visit www.fahbench.com
Explicit:
Checking for accuracy...clSetKernelArg
Re: Folding@Home Benchmark Beta Testing
Posted: Thu Jan 17, 2013 11:53 pm
by k1wi
Your line arguments result in a successful outcome for me.
clSetKernelArg isn't an error message either, so proteneer will have to get back to you as to why its happening
Re: Folding@Home Benchmark Beta Testing
Posted: Fri Jan 18, 2013 3:53 am
by Mad_Max
proteneer
Can you post size of protein used in test simulation in current FAHbench? (like number of atoms or name if it a common known protein)
And it is a same protein for Implicit and Explicit simulations?
Or where is not any real proteins and it is a pure syntetic/virtual test?
Re: Folding@Home Benchmark Beta Testing
Posted: Fri Jan 18, 2013 4:05 am
by Mad_Max
I got same error as PinHead: clSetKernelArg
And i use same ATI Card - HD 5750 (Juniper chip) - OpenCL device #0
Even i try to run test on CPU (OpenCL device #1) - same error
Code: Select all
FAHBench_0_3\FAHBench.exe --display-devices --disable-splash
[1] compatible platform(s):
-- 0 --
PROFILE = FULL_PROFILE
VERSION = OpenCL 1.1 AMD-APP-SDK-v2.4 (595.10)
NAME = AMD Accelerated Parallel Processing
VENDOR = Advanced Micro Devices, Inc.
(2) device(s) found on platform 0:
-- 0 --
DEVICE_NAME = Juniper
DEVICE_VENDOR = Advanced Micro Devices, Inc.
DEVICE_VERSION = OpenCL 1.1 AMD-APP-SDK-v2.4 (595.10)
-- 1 --
DEVICE_NAME = AMD Phenom(tm) II X6 1055T Processor
DEVICE_VENDOR = AuthenticAMD
DEVICE_VERSION = OpenCL 1.1 AMD-APP-SDK-v2.4 (595.10)
Invalid Platform (please use either OpenCL or CUDA)
Re: Folding@Home Benchmark Beta Testing
Posted: Fri Jan 18, 2013 7:35 pm
by proteneer
Do you guys have the latest drivers?
The Protein has 23558 atoms (DHFR)
Re: Folding@Home Benchmark Beta Testing
Posted: Fri Jan 18, 2013 10:15 pm
by codysluder
cuda will never run on anything except NVIDIA. Someday there will be opencl code that will run on NV,ATI,Intel,CPUs,etc. but now Intel, ATI, NV only test their opencl code on their own hardware.