R9 290x benchmark crashing at scrypt
#1
hi!

I have my usual R9 290x (watercooled) and I was running benchmarks with version 3.30 and 3.40 

Using latest AMD drivers: 17.2.1, Windows 10 

The benchmarks on both the 3.30 and 3.40 crash exactly when scrypt is being benchmarked. The error is "Hash64.exe has stopped working". 

Any ideas what could it be?
#2
Also, the 3.30 seems faster than the 3.40 on my card:

3.30
Md5: 13043 Mh/sec
SHA1: 4704 Mh/sec

3.40:
Md5: 11208 Mh/sec
SHA1: 4044 Mh/sec
#3
No, it's more like 11208 sounds to be correct, see here: https://docs.google.com/spreadsheets/d/1...1439721324
#4
atom: any feedback regarding why the benchmark is crashing when it is reaching "scrypt"? This is happening for both 3.30 and 3.40

btw, the graph you provided proves that earlier versions of hashcat (3.20 and 3.30) are actually faster in MD5 than version 3.40... So those numbers for version 3.30 above would kind of make sense. Kind of.
#5
How can I report a potential bug or malfunction if my benchmark crashes as soon as it reaches scrypt for version 3.30 and 3.40?

Anything I can try to find out why, like older versions? Please help.
#6
Bug reports go on Github
#7
I have seen the same error with AMDGPU-pro 16.40. This is a driver issue, not a hashcat issue.
#8
(03-08-2017, 09:54 AM)atom Wrote: I have seen the same error with AMDGPU-pro 16.40. This is a driver issue, not a hashcat issue.

I believe you, but why scrypt crashes, where as all the other algorithms are fine..?
#9
How can we know, the segfault occurs inside the AMD library
#10
Weird. I ran WPA and MD5 cracking all day without issues, so it seems that it is only limited to scrypt specifically.