v0.10 / Nvidia CUDA / descrypt / ERROR: cuLaunchKernel() 701
#1
Question 
Code:
$ ./cudaHashcat-lite64.bin -m 1500 MYDESCRYPTHASH
cudaHashcat-lite v0.10 by atom starting...

Password lengths range: 1 - 55
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: GeForce 8600 GTS, 511MB, 1550Mhz, 4MCU

[s]tatus [p]ause [r]esume [q]uit => ERROR: cuLaunchKernel() 701

Nvidia drivers version: 304.48;
./cudaExample.sh works OK.

Any ideas?
#2
(12-08-2012, 04:22 AM)moujik Wrote:
Code:
$ ./cudaHashcat-lite64.bin -m 1500 MYDESCRYPTHASH
cudaHashcat-lite v0.10 by atom starting...

Password lengths range: 1 - 55
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: GeForce 8600 GTS, 511MB, 1550Mhz, 4MCU

[s]tatus [p]ause [r]esume [q]uit => ERROR: cuLaunchKernel() 701

Nvidia drivers version: 304.48;
./cudaExample.sh works OK.

Any ideas?

CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES
This indicates that a launch did not occur because it did not have appropriate resources.

This error usually indicates that the user has attempted to pass too many arguments to the device kernel, or the kernel launch specifies too many threads for the kernel's register count. Passing arguments of the wrong size (i.e. a 64-bit pointer when a 32-bit int is expected) is equivalent to passing too many arguments and can also result in this error.
#3
DEScrypt requires more resources than MD5. The GeForce 8600 GTS is a bit to cheap.
#4
(12-09-2012, 11:18 AM)atom Wrote: DEScrypt requires more resources than MD5. The GeForce 8600 GTS is a bit to cheap.

A year ago the same hardware was indeed slow, but at least worked: http://hashcat.net/forum/thread-533.html
#5
newer versions require more ram. time to upgrade.
#6
(12-14-2012, 06:29 AM)epixoip Wrote: newer versions require more ram. time to upgrade.

Can you please clarify what is the minimum RAM requirement for v0.10? Smile
#7
He had in mind GPU RAM. It's possible that 256mb RAM is just too small for its 'enormous' 32 SP count for hashcat kernel to load.
#8
(12-16-2012, 08:09 PM)KT819GM Wrote: He had in mind GPU RAM. It's possible that 256mb RAM is just too small for its 'enormous' 32 SP count for hashcat kernel to load.

I meant Graphics RAM too, and my card has 512 MB of it (as shown in the 1st post).
What do you mean by "32 SP", is it a count of "shader processors" or smth. else?
#9
SP - Stream processors. Sorry didn't look at log, answered from my memory as even newer saw 8600gt/gts/gtx with 512Mb. Anyway, still thinking that for descrypt this card is too slow. Like HD4xxx series support was dropped for about same reason.