hashcat crash on benchmark
#1
I've been trying to get my RX580 rig to work with hashcat for days now but I can't get it to work. I currently have 6 RX580's mounted on a AsRock H110 motherboard. I'm running Windows 10 x64.

Somehow when I benchmark MD5 it works but after that it crashes. Also when I try to benchmark WPA it just shows 0H/s

I've tried installing the newest and some of the older AMD drivers but none of the ones I tried worked like it should. Is there any way to troubleshoot or debug this?

Full benchmark (attempt)
Code:
$ ./hashcat.exe -b
hashcat (v5.1.0-1495-g53254b45) starting in benchmark mode...

Benchmarking uses hand-optimized kernel code by default.
You can use it in your cracking session by setting the -O option.
Note: Using optimized kernel code limits the maximum supported password length.
To disable the optimized kernel code in benchmark mode, use the -w option.

OpenCL API (OpenCL 2.1 AMD-APP (2906.10)) - Platform #1 [Advanced Micro Devices, Inc.]
======================================================================================
* Device #1: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #2: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #3: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #4: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #5: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #6: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU

OpenCL API (OpenCL 2.1 ) - Platform #2 [Intel(R) Corporation]
=============================================================
* Device #7: Intel(R) HD Graphics 630, 6359/6423 MB (3211 MB allocatable), 24MCU
* Device #8: Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz, skipped

Benchmark relevant options:
===========================
* --optimized-kernel-enable

Hashmode: 0 - MD5

Speed.#1.........: 13054.5 MH/s (91.80ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#2.........: 13019.8 MH/s (92.06ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#3.........: 13085.6 MH/s (91.46ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#4.........: 12753.3 MH/s (93.96ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#5.........: 13022.1 MH/s (92.10ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#6.........: 12984.0 MH/s (92.33ms) @ Accel:256 Loops:512 Thr:256 Vec:1
Speed.#7.........:  555.9 MH/s (89.34ms) @ Accel:64 Loops:128 Thr:256 Vec:4
Speed.#*.........: 78475.2 MH/s

Hashmode: 100 - SHA1

Started: Sun Dec  8 12:13:15 2019
Stopped: Sun Dec  8 12:13:44 2019


WPA benchmark attempt:
Code:
$ ./hashcat.exe -b -m 2500
hashcat (v5.1.0-1495-g53254b45) starting in benchmark mode...

Benchmarking uses hand-optimized kernel code by default.
You can use it in your cracking session by setting the -O option.
Note: Using optimized kernel code limits the maximum supported password length.
To disable the optimized kernel code in benchmark mode, use the -w option.

/cygdrive/c/Users/rig_0/Documents/hashcat/OpenCL/m02500-optimized.cl: Optimized kernel requested but not needed - falling back to pure kernel
OpenCL API (OpenCL 2.1 AMD-APP (2906.10)) - Platform #1 [Advanced Micro Devices, Inc.]
======================================================================================
* Device #1: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #2: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #3: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #4: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #5: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU
* Device #6: Ellesmere, 3968/4096 MB (3264 MB allocatable), 36MCU

OpenCL API (OpenCL 2.1 ) - Platform #2 [Intel(R) Corporation]
=============================================================
* Device #7: Intel(R) HD Graphics 630, 6359/6423 MB (3211 MB allocatable), 24MCU
* Device #8: Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz, skipped

Benchmark relevant options:
===========================
* --optimized-kernel-enable

Hashmode: 2500 - WPA-EAPOL-PBKDF2 (Iterations: 4095)

Speed.#1.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#2.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#3.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#4.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#5.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#6.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#7.........:        0 H/s (0.00ms) @ Accel:0 Loops:0 Thr:256 Vec:1
Speed.#*.........:        0 H/s

Started: Sun Dec  8 12:16:16 2019
Stopped: Sun Dec  8 12:16:38 2019
Reply
#2
try testing only 1 device, -d 1

are you using cmd on windows ? that ./hashcat.exe syntax is somehow strange or is this powershell ?

are you using the beta version from https://hashcat.net/beta ?


I would say this is a driver problem... maybe you can fix it by disabling the "Intel HD Graphics 630", at least for the testing purpose
Reply
#3
(12-08-2019, 01:31 PM)philsmd Wrote: try testing only 1 device, -d 1

are you using cmd on windows ? that ./hashcat.exe syntax is somehow strange or is this powershell ?

are you using the beta version from https://hashcat.net/beta ?


I would say this is a driver problem... maybe you can fix it by disabling the "Intel HD Graphics 630", at least for the testing purpose

I was using cygwin since that is required to install the windows github version. I'm now using the beta version and when selecting single or multiple devices it seems to be working. Running a benchmark on one of the 6 GPU's works fine but as soon as i start the benchmark on all 6 of them it crashes my GPU's after the 4th algorithm or so, any idea why this is happening?
Reply
#4
you could also test with multiple devices, like this:

hashcat -b -m 100 -d 1,2,3,4


hashcat -b -m 100 -d 1,2,3,4,5



hashcat -b -m 100 -d 1,2,3,4,5,6

and see how many devices work and when exactly it starts failing....


in theory it could be anything, from power issues to driver issues etc... the most unlikely thing is probably a hashcat bug in this particular situation Wink (but also not impossible)

I would tend to a hardware or driver problem...
Reply
#5
Looks like some kind of hardware error. Are you using risers?

Possibly a heat problem, try running
hashcat -a3 00000000000000000000000000000000 ?a?a?a?a?a?a?a?a?a

for a few minutes and see what happens
Reply
#6
It appears to be cracking like it should now. I've been running a bruteforce for a few minutes to see if it would keep up on 10 GPU's (added 4) and it looks just fine. Will do a benchmark later on to see if everything is working now.

It looks like it was the internal Intel graphics that caused it to act all weird. skipping it with the -d option fixed that, thanks!
Reply
#7
good to hear that. of course always needing to add -d 1,2,3,4,5,6,7,8,9,10 is actually quite annoying... so maybe you can somehow completely disable the non-dedicated GPU (in bios/uefi for instance or via operating system configurations) or use something like aliases etc (to not forget the -d parameter).

for instance something like this under linux:
alias hashcat="hashcat -d 1,2,3,4,5,6,7,8,9,10"
Reply