phc-discussions - Re: [PHC] yescrypt on GPU

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20150507180241.GA14575@openwall.com>
Date: Thu, 7 May 2015 21:02:41 +0300
From: Solar Designer <solar@...nwall.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] yescrypt on GPU

On Sat, May 02, 2015 at 05:34:27AM +0300, Solar Designer wrote:
> The yescrypt cryptocoin stuff is starting to pay off.  djm34 has just
> implemented support for BSTY mining on GPU, in both OpenCL and CUDA
> (with tiny bits of inline PTX assembly, even - for things such as the
> pwxform MULs):
> 
> https://bitcointalk.org/index.php?topic=775289.msg11252741#msg11252741
> 
> The code is still very dirty.  I expect it won't build or work for most
> people as-is, yet.  However, it looks reasonably well optimized, and
> specialized to the yescrypt settings that BSTY uses (e.g., loop counts
> are precomputed and hard-coded, etc.)

Further in that thread, djm34 mentions getting 1.5 kh/s on GTX 980.  The
best speed another person reported so far is 980 h/s on GTX 750 Ti.

> For comparison, a (much cheaper) quad-core CPU does ~3400 h/s.

To be fair, GTX 750 Ti is cheap, so its speed is impressive.  This
suggests a higher-end GPU (perhaps Titan X) should reach CPU speed.
I am puzzled why higher speeds aren't being reported yet, if 750 Ti in
fact reaches 980 h/s.  Maybe it's just too few people trying this out.

OTOH, there are bigger CPUs, too.

> bsty     27511 99.7  0.1 242304 216548 pts/1   R+   04:47 100:23 cicc -arch compute_35 -m64 -ftz=0 -prec_div=1 -prec_sqrt=1 -fmad=1 -maxreg 128 -nvvmir-library /usr/local/cuda/bin/../nvvm/libdevice/libdevice.compute_35.10.bc --orig_src_file_name yescrypt/cuda_yescrypt.cu /tmp/tmpxft_00006af6_00000000-9_cuda_yescrypt.cpp3.i -o /tmp/tmpxft_00006af6_00000000-5_cuda_yescrypt.ptx

I restarted this build in "screen", but it was still still building a
day later, so I stopped it:

bsty     14657 99.7 14.9 19924888 19761748 pts/12 R+ May02 1795:55 cicc -arch compute_35 -m64 -ftz=0 -prec_div=1 -prec_sqrt=1 -fmad=1 -maxreg 128 -nvvmir-library /usr/local/cuda/bin/../nvvm/libdevice/libdevice.compute_35.10.bc --orig_src_file_name yescrypt/cuda_yescrypt.cu /tmp/tmpxft_00003887_00000000-9_cuda_yescrypt.cpp3.i -o /tmp/tmpxft_00003887_00000000-5_cuda_yescrypt.ptx

Yesterday, I also tried out djm34's newer sgminer commit.  This time,
the kernel built for both AMD and NVIDIA, but it was giving HW errors on
HD 7990 (at a reported speed of ~730 h/s for both GPUs combined) and a
poor speed on GTX TITAN (as low as 35 h/s).

I guess I could have better luck with different Catalyst and CUDA SDK
versions, but I didn't bother trying yet.

Alexander