10-16-2013, 12:35 AM
(10-15-2013, 04:34 PM)atom Wrote: remember that the .restore file is written only at specific times. it might not be an issue, you maybe stopped to early and it did not had a chance to write the file. especially when you increase the -n value and in combination with a slow hash, the runtime for a single complete kernel run can take a lot of time.
the background therefore is the gpu kernels that are running. the restore progress can only be written when a full kernel run is finished. if it wouldn't do it that way parts of the keyspace that wasn't tested would be lost (the parts that did not have been calculated). so it has to wait for kernel run finish, then update progress count, then it can write .restore file.
While this is quite possible, I ran his command line on linux and got a proper session file even after quitting after just 4 minutes. Per yesterdays conversation with someone else using windows with cudaHashcat-plus, the might be a bug in the windows version. He had the same problem with the restore file being empty.