Hello!
I am running oclHashcat 1.01 on Ubuntu 13.10_x64 with the latest Catalyst 13.12-driver. oclhashcat works, but the 2nd GPU ist far too slow.
1st GPU is a 5850 HD, 2nd a 7790 HD.
Code:
Session.Name...: oclHashcat
Status.........: Running
Input.Mode.....: Mask (?a?a?a?a?a?a?a) [7]
Hash.Target....: ***
Hash.Type......: NTLM
Time.Started...: Thu Feb 27 21:01:06 2014 (12 mins, 40 secs)
Time.Estimated.: Thu Feb 27 23:15:39 2014 (2 hours, 1 min)
Speed.GPU.#1...: 7530.0 MH/s
Speed.GPU.#2...: 1176.5 MH/s
Speed.GPU.#*...: 8706.4 MH/s
Recovered......: 0/1 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 6588427501568/69833729609375 (9.43%)
Rejected.......: 0/6588427501568 (0.00%)
HWMon.GPU.#1...: 95% Util, 67c Temp, 56% Fan
HWMon.GPU.#2...: 95% Util, 67c Temp, 56% Fan
The same hardware does the job well running Windows 7 Prof. x64, thereby the 2nd GPU nearly reaches the speed of the first one. I tried to crossfire both cards and to connect a monitor (via HDMI) to the card with GPU#2. Unfortunatelly no changes :-(
Any ideas?
Hey,
did you try to run the same attack with the command line parameter -d 1 , afterwards -d 2 .
Furthermore you should try to run the benchmark oclHashcat64.bin -b -m 1000, also oclHashcat64.bin -b -m 1000 -d 1 and oclHashcat64.bin -b -m 1000 -d 2
New 1.02 version (not yet released
https://twitter.com/hashcat/status/437598775687659521 ) may change this situation / fix this problem somehow, because several things changed since last release. Dunno.
But it could also be a hw / driver issue.
Btw, dunno if it helps or changes sth, but did you try to swap the 2 cards?
It would be also important that you post the device list (with mem, clocks etc) which you will always see (if --quiet was not used) at the beginning of the run.
same here.. wasn't like that before v1
Code:
andreas@Daedalus:~/Downloads/oclHashcat-1.01$ ./oclHashcat64.bin -b -m 1000
oclHashcat v1.01 starting in benchmark-mode...
Device #1: Cypress, 512MB, 760Mhz, 18MCU
Device #2: Bonaire, 922MB, 1050Mhz, 14MCU
Hashtype: NTLM
Workload: 1024 loops, 256 accel
Speed.GPU.#1.: 8047.2 MH/s
Speed.GPU.#2.: 4697.9 MH/s
Speed.GPU.#*.: 12745.0 MH/s
andreas@Daedalus:~/Downloads/oclHashcat-1.01$ ./oclHashcat64.bin -b -m 1000 -d 1
oclHashcat v1.01 starting in benchmark-mode...
Device #1: Cypress, 512MB, 760Mhz, 18MCU
Device #2: skipped by user
Hashtype: NTLM
Workload: 1024 loops, 256 accel
Speed.GPU.#1.: 8050.4 MH/s
andreas@Daedalus:~/Downloads/oclHashcat-1.01$ ./oclHashcat64.bin -b -m 1000 -d 2
oclHashcat v1.01 starting in benchmark-mode...
Device #1: skipped by user
Device #2: Bonaire, 922MB, 1050Mhz, 14MCU
Hashtype: NTLM
Workload: 1024 loops, 256 accel
Speed.GPU.#1.: 4697.9 MH/s
Swapping the cards is the last try I think of, because it's a headless server running in a basement room of my house ;-). And what I also noticed is that GPU1 runs at peak core clock, while GPU2 does by far not. And amdconfig always shows a usage of 0%:
Code:
andreas@Daedalus:~$ !141
amdconfig --odgc --adapter=all
Adapter 0 - ATI Radeon HD 5800 Series
Core (MHz) Memory (MHz)
Current Clocks : 760 1050
Current Peak : 760 1050
Configurable Peak Range : [550-875] [900-1225]
GPU load : 95%
Adapter 1 - AMD Radeon HD 7700 Series
Core (MHz) Memory (MHz)
Current Clocks : 527 1500
Current Peak : 1050 1500
Configurable Peak Range : [300-1200] [150-1600]
GPU load : 0%
O.k., I did further investigations. The problem is NOT that it´s the 2nd card, but obviously the card itself. I put the 7790 in another computer, made a clean new setup of both, Windows 7 and Ubuntu 13.10 with the actual drivers.
Same result, card gives only about 1700 MH/s on Linux, but 6000 on Windows.
I may be wrong but I think Bonaire is an OD6 card? Perhaps it has the weird Powertune that throttles in favor of low noise just like R9 290X. Upcoming oHC v1.02 might behave a lot better but until then I suggest you try
https://github.com/epixoip/od6config and see if it helps.
(03-02-2014, 10:58 PM)magnum Wrote: [ -> ]I may be wrong but I think Bonaire is an OD6 card? Perhaps it has the weird Powertune that throttles in favor of low noise just like R9 290X. Upcoming oHC v1.02 might behave a lot better but until then I suggest you try https://github.com/epixoip/od6config and see if it helps.
od6config --autofix increases the core and memory clock to their max (and the fan speed also
). This increases cracking speed from 1700 to 2300 MH/s, which is a linear increase according to the core clock. But it´s still much slower than running under Win7 (about 7000 MH/s)...
maybe the windows driver is magic
found the problem and solution. I have to make crossfire
and that made 2nd gpu slow. released crossfire cable and both came the same speed
(03-05-2014, 06:58 PM)tbm Wrote: [ -> ]found the problem and solution. I have to make crossfire and that made 2nd gpu slow. released crossfire cable and both came the same speed
Not in my case
. No matter if I use crossfire cable or not, or even if i use the 7790 as the only graphics card, I get poor results under Linux but respectable ones under Win7.