+- hashcat Forum (https://hashcat.net/forum)
+-- Forum: Deprecated; Previous versions (https://hashcat.net/forum/forum-29.html)
+--- Forum: Old oclHashcat Support (https://hashcat.net/forum/forum-38.html)
+--- Thread: ERROR: cuModuleLoad() 301 (/thread-3967.html)
Pages:
1
2
ERROR: cuModuleLoad() 301 - qq403723786 - 01-09-2015
os: linux 2.6.32
version:cudaHashcat-1.31
gpu:nvida k80
ERROR: cuModuleLoad() 301
RE: ERROR: cuModuleLoad() 301 - atom - 01-09-2015
Typical error when user used "7z e" to unpack the archive. Please use "7z x"
RE: ERROR: cuModuleLoad() 301 - rincebrain - 01-12-2015
(01-09-2015, 04:55 PM)atom Wrote: Typical error when user used "7z e" to unpack the archive. Please use "7z x"
I used "7z x" to unpack the archive, and it still gives me:
Code:
PS D:\cudaHashcat-1.31> .\cudaHashcat64.exe -b
cudaHashcat v1.31 starting in benchmark-mode...
WARN: NvAPI_EnumPhysicalGPUs() -6 NVAPI_NVIDIA_DEVICE_NOT_FOUND
Device #1: Tesla K80, 11519MB, 823Mhz, 13MCU
Device #2: Tesla K80, 11519MB, 823Mhz, 13MCU
ERROR: cuModuleLoad() 301K80 on Windows Server 2012 R2.
RE: ERROR: cuModuleLoad() 301 - undeath - 01-12-2015
is your forceware driver properly installed?
RE: ERROR: cuModuleLoad() 301 - epixoip - 01-13-2015
Tesla uses an older driver, so that's why you're encountering difficulties. Current Tesla driver is 340.65 while current GTX driver is 343.36. cudaHashcat 1.31 requires 331.67 or later.
RE: ERROR: cuModuleLoad() 301 - rincebrain - 01-13-2015
I'm using the driver 341.21, which is later than 331.67 unless I'm quite bad at math.
But then, that issue should crop up for all Tesla-series cards, not just the K80, I believe?
RE: ERROR: cuModuleLoad() 301 - epixoip - 01-13-2015
Where did you find that driver? Latest version is 340.65, released 2014.12.12.
http://www.nvidia.com/download/driverResults.aspx/80874/en-us
If you're using the Geforce driver that would probably explain the "NvAPI_EnumPhysicalGPUs() -6 NVAPI_NVIDIA_DEVICE_NOT_FOUND" error, as the Geforce drivers do not support Tesla.
RE: ERROR: cuModuleLoad() 301 - rincebrain - 01-13-2015
(01-13-2015, 12:08 PM)epixoip Wrote: Where did you find that driver? Latest version is 340.65, released 2014.12.12.
http://www.nvidia.com/download/driverResults.aspx/80874/en-us
If you're using the Geforce driver that would probably explain the "NvAPI_EnumPhysicalGPUs() -6 NVAPI_NVIDIA_DEVICE_NOT_FOUND" error, as the Geforce drivers do not support Tesla.
To clarify, I'm another user getting cuModuleLoad() 301 with a Tesla K80, but on Windows Server 2012 R2, not Linux x64. I can make another thread, if that's preferable, but it seemed to be the same problem.
http://www.nvidia.com/download/driverResults.aspx/81098/en-us
RE: ERROR: cuModuleLoad() 301 - epixoip - 01-13-2015
Sorry, didn't realize you aren't the OP.
The release notes for 341.21 say it is the first driver to support the K80:
Quote:New in Release 341.21
- Added support for Tesla K80
So it is possible that hashcat is triggering some bug in this new driver, or vice versa. You may need to hop on IRC and work with atom directly on this.
The other thing that comes to mind is that I don't know how well synced the Tesla driver versions and Geforce driver versions are. "331.67 or later" in Geforce terms may not mean the same thing in Tesla terms. I recognize that this is the latest driver, but it is possible that this driver doesn't have something that hashcat requires.
There aren't very many Tesla users around here since they have extremely low Perf/$ for hash cracking, so it's entirely possible that this issue may affect all Tesla users on 1.31 and we just don't know about it. Or, since the K80 is a new card, it's also entirely possible that this is an isolated issue affecting only K80 users (all two of you.) Either way I think the next step is probably to involve atom in this discussion.
RE: ERROR: cuModuleLoad() 301 - rincebrain - 01-14-2015
Quick addendum - Tesla K80 is an sm_37 chip, so 301 comes up, indeed, from a FILE_NOT_FOUND - because there's no sm_37 kernels in -1.31.