Hello,

I'm runing MW wu's cuda only, using a Nvidia GeForce GTX 275. My computer is a CPU i7 - 860 and Windows 7 64 bits.

Everything was crunching correctly until the server upload new WU called 0.50 (cuda_opend) de_separation_16 (or 14, or 17...) _3s_fix_1...

Every WU stopped immediatly after the begining, saying there is a calcul error.

I am using a regular Boinc client 6.10.43.

If someone has any clue, it will help.

In advance, thenk you very much.

Lionel.

I just helped Ron on the same problem, your driver is just a little too old for the openCL app.

http://www.nvidia.com/object/win7-winvista-64bit-260.99-whql-driver.html

---

Well I downloaded the latest driver from Nvidia Web site before posting my thread...

The site does not show you running it though.

It should be 260.99.

---

i am having the same problem. i have updated my drivers and restarted my system. help please

Hello,

The problem is fixed. The first installation of the driver didn't work for any reason. So I upgrade Boinc Manager to the 6.10.58 version and reinstall the latest nvidia driver 260.99 (Windows indicates this is the version : 8.17.12.6099, date : 16.10.2010).
Then I reboot the computer and WU's are running correctly.

sorry to bother you about my problem. I hope the solution will help.

Bests regards.

Lionel.

I'm already running the current version of BOINC, but did load the newer nVidia driver today. I'm still getting massive computation errors. Oddly, my older 8600 GT is working fine. My newer GTX 260 is not.

---

Correction, GPU-Z indicates that the video engine load for both cards is 0%, but one of them (device 0) is crunching. I don't know, maybe that's the 260, maybe not. I've got a cold and I am not seeing clearly.

---

I'm already running the current version of BOINC, but did load the newer nVidia driver today. I'm still getting massive computation errors. Oddly, my older 8600 GT is working fine. My newer GTX 260 is not.

It looks like BOINC is trying to run it on the 8600 which doesn't have doubles so it won't work, which is a problem.

I don't suppose there is a fix for this, is there? I've got 3 monitors hooked up so I can't just yank the card out, and I'm not going to buy a new card to get around this, so maybe I should be looking for a new project? I'm sure it's going to clog things on your end with all of my computation errors being returned.

---

I don't suppose there is a fix for this, is there? I've got 3 monitors hooked up so I can't just yank the card out, and I'm not going to buy a new card to get around this, so maybe I should be looking for a new project? I'm sure it's going to clog things on your end with all of my computation errors being returned.

If you do not want to crunch on the 8600, there is a way to exclude it via a cc_config.xml file

---

I changed the value from 1 to 0 and that didn't help on "use all gpu's" in the cc_config.xml file. Is there a different way to word this so that the 8600 doesn't grab more WU's?

---

I get these errors (WU):

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
 - exit code -1073741819 (0xc0000005)
</message>
<stderr_txt>
<search_application> milkywayathome separation 0.50 Windows x86 double OpenCL </search_application>
Found 1 platforms
Platform 0 information:
  Platform name:       NVIDIA CUDA
  Platform version:    OpenCL 1.0 CUDA 3.2.1
  Platform vendor:     
  Platform profile:    
  Platform extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll 
Using device 1 on platform 0
Found 3 CL devices
Device GeForce GTX 295 (NVIDIA Corporation:0x10de)
Type:                CL_DEVICE_TYPE_GPU
Driver version:      265.90
Version:             OpenCL 1.0 CUDA
Compute capability:  1.3
Little endian:       CL_TRUE
Error correction:    CL_FALSE
Image support:       CL_TRUE
Address bits:        32
Max compute units:   30
Clock frequency:     1075 Mhz
Global mem size:     911605760
Max mem alloc:       227901440
Global mem cache:    0
Cacheline size:      0
Local mem type:      CL_LOCAL
Local mem size:      16384
Max const args:      9
Max const buf size:  65536
Max parameter size:  4352
Max work group size: 512
Max work item dim:   3
Max work item sizes: { 512, 512, 64 }
Mem base addr align: 2048
Min type align size: 128
Timer resolution:    1000 ns
Double extension:    MW_CL_KHR_FP64
Extensions:          cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll  cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 
Found a compute capability 1.3 device. Using -cl-nv-maxrregcount=32 

Compiler flags:
-cl-mad-enable -cl-no-signed-zeros -cl-strict-aliasing -cl-finite-math-only -DUSE_CL_MATH_TYPES=0 -DUSE_MAD=1 -DUSE_FMA=0 -cl-nv-verbose  -cl-nv-maxrregcount=32  -DDOUBLEPREC=1 -DMILKYWAY_MATH_COMPILATION -DNSTREAM=3 -DFAST_H_PROB=1 -DAUX_BG_PROFILE=0 -DUSE_IMAGES=1 -DI_DONT_KNOW_WHY_THIS_DOESNT_WORK_HERE=0  

Build status: CL_BUILD_ERROR


Unhandled Exception Detected...

- Unhandled Exception Record -
Reason: Access Violation (0xc0000005) at address 0x02B40A20 read attempt to address 0x00000000

Windows 7 64bit, one GTX295 + one 9600GSO

How do I use only one of my multiple GPUs?
You can tell BOINC which GPU to ignore using the <ignore_cuda_dev>0|1|2|3</ignore_cuda_dev> and <ignore_ati_dev>0|1|2|3</ignore_ati_dev> flags in the <options> section of cc_config.xml

GPUs are counted per brand and from the first in your system to the last, while the first GPU of the same brand is designated GPU 0, the second GPU 1 etc.

So for example, you have a 3 GPU system, two Nvidia and one ATI. You only want BOINC to use the first Nvidia and the ATI, not the second Nvidia.

<cc_config>
    <options>
     <ignore_cuda_dev>1</ignore_cuda_dev>
    </options>
</cc_config>

http://boincfaq.mundayweb.com/index.php?language=1&view=471

---

OK, when I saw the new (.50) workunits, I checked and saw I needed updated drivers, and installed 260.99 on both my systems. The remaining (.24) units ran through, creating output files which are still waiting for upload. The (.50) units also ran to completion in typical times for my GTX260 cards, but instead of creating files for upload, they went straight to "ready to report", even though the project was, and is, "down" for all intents and purposes.

Is there still a problem with the new version? I can't think of anything on my end that could be wrong.

OK, when I saw the new (.50) workunits, I checked and saw I needed updated drivers, and installed 260.99 on both my systems. The remaining (.24) units ran through, creating output files which are still waiting for upload. The (.50) units also ran to completion in typical times for my GTX260 cards, but instead of creating files for upload, they went straight to "ready to report", even though the project was, and is, "down" for all intents and purposes.

Is there still a problem with the new version? I can't think of anything on my end that could be wrong.

That doesn't sound like anything is wrong at all and is perfectly normal. There isn't a special output file anymore.

I get these errors ([url=http://milkyway.cs.rpi.edu/milkyway/result.php?

- Unhandled Exception Record -
Reason: Access Violation (0xc0000005) at address 0x02B40A20 read attempt to address 0x00000000
[/code]

Windows 7 64bit, one GTX295 + one 9600GSO

That's weird, but I might have fixed where this probably happens for the next release.