Closed GoogleCodeExporter closed 8 years ago
please update your gpu drivers
Original comment by lukas.l...@gmail.com
on 3 Jan 2011 at 7:08
AMD doesn't provide adequate Linux drivers on their site. I was forced to use
the drivers on the CD that came with the graphics card. Anything else makes the
system unstable.
Original comment by jared.pe...@gmail.com
on 3 Jan 2011 at 7:13
Try downgrading the SDK then. The situation with AMD GPU-drivers for Linux is
*very* bad and is not going to change any time soon.
Original comment by lukas.l...@gmail.com
on 3 Jan 2011 at 7:17
That didn't work, unfortunately.
Original comment by jared.pe...@gmail.com
on 3 Jan 2011 at 9:31
there is little we can do then except waiting for new drivers
Original comment by lukas.l...@gmail.com
on 9 Jan 2011 at 5:16
Catalyst 11.1 seems to be working, can you try that?
https://groups.google.com/group/pyrit/browse_thread/thread/22164ab0fc47541b?hl=e
n
Original comment by lukas.l...@gmail.com
on 28 Jan 2011 at 7:32
@jared
please, if new Catalyst 11.1 are able to manage your 6870, please provide some
real(*) test of PMK/s (i.e. report how many secont to test 50 million
passwords) and report results: this will allow to solve the question of issue
173 comment 63 and 68. Thanks you.
(*)the benchmark and benchmark_long does not provide trustable PMK/s.
Original comment by pyrit.lo...@gmail.com
on 28 Jan 2011 at 10:15
Thank you for the update, I'm going to install *nix on my machine again and I
will try the newer Catalyst and report my results as soon as I get them.
Original comment by jared.pe...@gmail.com
on 28 Jan 2011 at 3:45
Apologies for the delayed response.
I am still having issues and they still seem to be driver-related. I have
installed Catalyst 11.1 several times on clean installations of Ubuntu. If I
start up CCC, it claims that it can't find a supported video card and the
resolution behaves as such. I plan on trying Arch Linux or another distribution
to see if it helps (or produces different issues).
Original comment by jared.pe...@gmail.com
on 3 Feb 2011 at 4:50
After debian 5.0, ubuntu and bactrack4 gave me headaches, I moved to debian 6.0
and problems gone. Maybe you can try the same way
Original comment by pyrit.lo...@gmail.com
on 3 Feb 2011 at 10:17
To confirm, are you saying that by moving over to Debian 6 with the Catalyst
11.1 drivers, you were able to get Pyrit to detect the 6870, is that correct?
Did you use the SVN repo for your build or a pre-built version?
Are you on x86 or x86-64?
Are you using cpyrit-opencl or cpyrit-calpp?
Did you install python-pyopencl?
Are you running X while running pyrit?
Original comment by absolute...@gmail.com
on 4 Feb 2011 at 7:26
it's probably the kernel version <-> ati drivers
Original comment by lukas.l...@gmail.com
on 4 Feb 2011 at 7:46
no, sorry I was not clear. I want only to report that ubuntu and backtrack was
not good for me and debian 5 too old. Debian 6 works fine, but with catalyst
10.6/7 and radeon HD5870 (I have not HD6xxx).
I used svn, X86_64, calpp and twm.
btw, X MUST run otherwise ati driver does not run.
Original comment by pyrit.lo...@gmail.com
on 4 Feb 2011 at 8:17
any update here?
Original comment by lukas.l...@gmail.com
on 15 Feb 2011 at 4:11
I'm afraid not. Still waiting on more driver updates from AMD.
Original comment by jared.pe...@gmail.com
on 15 Feb 2011 at 5:00
benchmark results for the 6970 were posted on the mailing list at
https://groups.google.com/group/pyrit/browse_thread/thread/22164ab0fc47541b
Original comment by lukas.l...@gmail.com
on 15 Feb 2011 at 5:03
It may be worth noting that AMD has this page
(http://support.amd.com/us/kbarticles/Pages/AMDCatalyst111bHotfix.aspx) on
their website, and it seems rather relevant. However the link doesn't work and
I can't seem to find the executable anywhere else on the net.
Am I mistaken in assuming that the 6900 series is newer than the 6800 series?
If this is the case, I can't understand why it is that the 6800 cards don't
seem to work well, if at all.
Original comment by jared.pe...@gmail.com
on 15 Feb 2011 at 5:11
AMD drivers are *very* fragile when it comes to kernel- and X11-versions. They
are also used to lag behind a few months.
The result posted on the mailing list was done with BT4 which happens to run
from a LiveCD. You may want to download BT4 yourself, resemble the conditions
as posted on the mailing list and try your GPU there...
Original comment by lukas.l...@gmail.com
on 15 Feb 2011 at 5:14
My computer doesn't play nice with BackTrack for some reason, although I
suspect that the CD/DVD drive could be the problem. I'll put it on a LiveUSB
and post my results with BT4 R2 if I can get it running.
Original comment by jared.pe...@gmail.com
on 15 Feb 2011 at 5:17
Couldn't run BT4 for whatever reason, I kept getting a kernel panic with no
helpful info on the web. I'm instead trying to match the kernel version (which
is 2.6.35-25 at this point in time) and hope that Catalyst works when it's
finished.
Original comment by jared.pe...@gmail.com
on 15 Feb 2011 at 6:35
The problem isn't with Backtrack, it seems. My hardware simply doesn't like the
kernel. I've read that it can be fixed by using a 64-bit OS, so I'll try that
when I have the time to build the kernel again.
Original comment by jared.pe...@gmail.com
on 16 Feb 2011 at 12:28
I'm very happy to report that it's finally working. I checked AMD's site on a
hunch, and they released Catalyst 11.2 since I last checked today. It fixed my
ugly graphics problem, but more importantly, pyrit works. However it seems that
I made a mistake while assigning variables for the Stream SDK as well (the
included PDF wasn't entirely clear to me), so that could just as easily have
been the problem.
Kernel: 2.6.35-25-generic (Ubuntu 10.4 x86)
ATI Stream SDK: 2.3
Catalyst 11.2
I'd be more than happy to provide any other potentially useful information.
Here's my benchmark for the 6870:
Computed 32625.92 PMKs/s total.
#1: 'OpenCL-Device 'Barts'': 31661.9 PMKs/s (RTT 2.8)
#2: 'CPU-Core (SSE2)': 673.8 PMKs/s (RTT 3.0)
#3: 'CPU-Core (SSE2)': 669.3 PMKs/s (RTT 3.0)
#4: 'CPU-Core (SSE2)': 671.9 PMKs/s (RTT 3.0)
#5: 'CPU-Core (SSE2)': 672.2 PMKs/s (RTT 3.0)
#6: 'CPU-Core (SSE2)': 696.1 PMKs/s (RTT 3.0)
I notice the PMKs/s don't quite add up, I believe this was mentioned elsewhere.
Not that it concerns me much, with the performance I'm getting anyway.
Original comment by jared.pe...@gmail.com
on 16 Feb 2011 at 1:48
Original comment by lukas.l...@gmail.com
on 16 Feb 2011 at 7:25
Original issue reported on code.google.com by
jared.pe...@gmail.com
on 3 Jan 2011 at 7:02