09-17-2011, 12:18 AM
Excellent piece of work!
I haven't gotten a chance to try all the new goodies, but it seems pretty stable (one "ERROR: cuCtxSynchronize() 702")
And faster, though not quite the speed improvement you show (GTX560).
But then, I still haven't found an effective method for tuning the gpu-accel and gpu-loops parameters
(GPU load goes to 99% regardless, but higher values tend to cause instability, i.e. CUDA errors)
I see you found a way to incorporate some of the "output which wordlist" feature request (http://hashcat.net/forum/showthread.php?tid=483). Works well.
With the pushing of features into oclHashcat-plus, I'm also wondering if this will be (sooner or later) adopted as the sole ocl framework. As a non-open source but non-commercial application, support for four different frameworks (on multiple platforms) seems like an unnecessary burden.
I haven't gotten a chance to try all the new goodies, but it seems pretty stable (one "ERROR: cuCtxSynchronize() 702")
And faster, though not quite the speed improvement you show (GTX560).
But then, I still haven't found an effective method for tuning the gpu-accel and gpu-loops parameters
(GPU load goes to 99% regardless, but higher values tend to cause instability, i.e. CUDA errors)
I see you found a way to incorporate some of the "output which wordlist" feature request (http://hashcat.net/forum/showthread.php?tid=483). Works well.
With the pushing of features into oclHashcat-plus, I'm also wondering if this will be (sooner or later) adopted as the sole ocl framework. As a non-open source but non-commercial application, support for four different frameworks (on multiple platforms) seems like an unnecessary burden.