[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAN4B415NhNXv3T8c5tkK4fu4HNkPJ09G5Y9m8Eo7yybFvO+V+g@mail.gmail.com>
Date: Sat, 1 Aug 2015 18:32:36 +0200
From: Luis Rocha <luiscrocha@...il.com>
To: john-users@...ts.openwall.com
Subject: Re: Nvidia CUDA drivers
Hello,
My bad, did not knew that Nvida supported OpenCL .... JtR Its now running
with OpenCL.
Anyway, just saw these graphic cards are almost useless but wanted to
report that some of the OpenCL formats fail to run in case its relevant for
someone - please see below.
Thx
Luis
lrocha@...e8:~/JohnTheRipper/run$ ./john
John the Ripper 1.8.0.6-jumbo-1-565-gc0560a4 [linux-gnu 64-bit AVX-ac]
lrocha@...e8:~/JohnTheRipper/run$ ./john --test --format=oldoffice-opencl
Device 0: NVS 300
Benchmarking: oldoffice-opencl, MS Office <= 2003 [MD5/SHA1 RC4 OpenCL]...
Options used: -I ./kernels -cl-mad-enable -cl-nv-verbose -D__GPU__
-DDEVICE_INFO=18 -DDEV_VER_MAJOR=331 -DDEV_VER_MINOR=113 -D_OPENCL_COMPILER
-DASCII -DPLAINTEXT_LENGTH=19
Build log: ptxas error : Entry function 'oldoffice_md5' uses too much
shared data (0x401c bytes, 0x4000 max)
ptxas error : Entry function 'oldoffice_sha1' uses too much shared data
(0x401c bytes, 0x4000 max)
ptxas info : 0 bytes gmem, 356 bytes cmem[0]
ptxas info : Compiling entry function 'oldoffice_md5' for 'sm_12'
ptxas info : Used 54 registers, 16412 bytes smem, 44 bytes cmem[1], 64
bytes lmem
ptxas info : Compiling entry function 'oldoffice_sha1' for 'sm_12'
ptxas info : Used 31 registers, 16412 bytes smem, 76 bytes cmem[1], 64
bytes lmem
ptxas info : Compiling entry function 'oldoffice_utf16' for 'sm_12'
ptxas info : Used 6 registers, 28 bytes smem, 4 bytes cmem[1]
Error -42 building kernel $JOHN/kernels/oldoffice_kernel.cl. DEVICE_INFO=18
OpenCL error (CL_INVALID_BINARY) in file (common-opencl.c) at line (977) -
(clBuildProgram failed.)
lrocha@...e8:~/JohnTheRipper/run$ ./john --test --format=sha256crypt-opencl
Device 0: NVS 300
lrocha@...e8:~/JohnTheRipper/run$ ./john --test --format=sha512crypt-opencl
Device 0: NVS 300
Benchmarking: sha512crypt-opencl, crypt(3) $6$ (rounds=5000) [SHA512
OpenCL]... FAILED (cmp_all(1))
lrocha@...e8:~/JohnTheRipper/run$ ./john --test --format=Raw-MD5-opencl
Device 0: NVS 300
Benchmarking: Raw-MD5-opencl [MD5 OpenCL]... Options used: -I ./kernels
-cl-mad-enable -cl-nv-verbose -D__GPU__ -DDEVICE_INFO=18
-DDEV_VER_MAJOR=331 -DDEV_VER_MINOR=113 -D_OPENCL_COMPILER -D
OFFSET_TABLE_SIZE=11 -D HASH_TABLE_SIZE=5 -D SHIFT64_OT_SZ=5 -D
SHIFT64_HT_SZ=1 -D NUM_LOADED_HASHES=5 -D NUM_INT_KEYS=1 -D
SELECT_CMP_STEPS=4 -D BITMAP_SIZE_BITS_LESS_ONE=32767 -D
USE_LOCAL_BITMAPS=1 -D IS_STATIC_GPU_MASK=0 -D CONST_CACHE_SIZE=65536 -D
LOC_0=-1 -D LOC_1=-1 -D LOC_2=-1 -D LOC_3=-1
Build log: ptxas error : Entry function 'md5' uses too much shared data
(0x4038 bytes, 0x4000 max)
ptxas info : 0 bytes gmem
ptxas info : Compiling entry function 'md5' for 'sm_12'
ptxas info : Used 23 registers, 16440 bytes smem, 48 bytes cmem[1], 64
bytes lmem
Error -42 building kernel $JOHN/kernels/md5_kernel.cl. DEVICE_INFO=18
OpenCL error (CL_INVALID_BINARY) in file (common-opencl.c) at line (977) -
(clBuildProgram failed.)
lrocha@...e8:~/JohnTheRipper/run$ ./john --test --format=Raw-SHA1-opencl
Device 0: NVS 300
Benchmarking: Raw-SHA1-opencl [SHA1 OpenCL]... Options used: -I ./kernels
-cl-mad-enable -cl-nv-verbose -D__GPU__ -DDEVICE_INFO=18
-DDEV_VER_MAJOR=331 -DDEV_VER_MINOR=113 -D_OPENCL_COMPILER -D
OFFSET_TABLE_SIZE=11 -D HASH_TABLE_SIZE=5 -D SHIFT64_OT_SZ=5 -D
SHIFT64_HT_SZ=1 -D SHIFT128_OT_SZ=3 -D SHIFT128_HT_SZ=1 -D
NUM_LOADED_HASHES=4 -D NUM_INT_KEYS=1 -D SELECT_CMP_STEPS=4 -D
BITMAP_SIZE_BITS_LESS_ONE=32767 -D USE_LOCAL_BITMAPS=1 -D
IS_STATIC_GPU_MASK=0 -D CONST_CACHE_SIZE=65536 -D LOC_0=-1 -D LOC_1=-1 -D
LOC_2=-1 -D LOC_3=-1
Build log: ptxas error : Entry function 'sha1' uses too much shared data
(0x4038 bytes, 0x4000 max)
ptxas info : 0 bytes gmem
ptxas info : Compiling entry function 'sha1' for 'sm_12'
ptxas info : Used 26 registers, 16440 bytes smem, 60 bytes cmem[1], 64
bytes lmem
Error -42 building kernel $JOHN/kernels/sha1_kernel.cl. DEVICE_INFO=18
OpenCL error (CL_INVALID_BINARY) in file (common-opencl.c) at line (977) -
(clBuildProgram failed.)
On Fri, Jul 31, 2015 at 11:37 AM, Luis Rocha <luiscrocha@...il.com> wrote:
> Hello All,
>
> I am trying to test some small Nvidia GPU on Ubuntu with CUDA drivers and
> JtR but for some reason its failing. Thought to reach out to you guys to
> see if anyone has any idea on what could be the problem
>
>
>
> lrocha@...e8:~/JohnTheRipper/run$ lspci -v | grep NVI
> 01:00.0 VGA compatible controller: NVIDIA Corporation GT218 [NVS 300] (rev
> a2) (prog-if 00 [VGA controller])
> 02:00.0 VGA compatible controller: NVIDIA Corporation GT218 [NVS 300] (rev
> a2) (prog-if 00 [VGA controller])
>
>
> I installed the drivers using:
> apt-get install libcuda1-331 nvidia-cuda-dev nvidia-cuda-toolkit
>
>
> lrocha@...e8:~/JohnTheRipper/run$ nvopencc -v
> NVIDIA (R) CUDA Open64 Compiler
> Cuda compilation tools, release 5.5, V5.5.0
> Built on 2013-07-17
> Open64 Compiler Suite: Version 4.1
> Built on: 2013-07-17
> Thread model: posix
> GNU gcc version 3.4.5 (Open64 4.2 driver)
>
>
> and compiled the latest JtR community version with the following settings:
>
> Configured for building John the Ripper jumbo:
>
> Target CPU .................................. x86_64 AVX, 64-bit LE
> AES-NI support .............................. depends on OpenSSL
> Target OS ................................... linux-gnu
> Cross compiling ............................. no
> Legacy arch header .......................... x86-64.h
>
> Optional libraries/features found:
> Fuzzing test ................................ no
> Experimental code ........................... no
> OpenMPI support (default disabled) .......... no
> Fork support ................................ yes
> OpenMP support .............................. yes (not for fast formats)
> OpenCL support .............................. yes
> CUDA support ................................ yes
> Generic crypt(3) format ..................... yes
> Rexgen (extra cracking mode) ................ no
> GMP (PRINCE mode and faster SRP formats) .... no
> PCAP (vncpcap2john and SIPdump) ............. no
> Z (pkzip format, gpg2john) .................. yes
> BZ2 (gpg2john extra decompression logic) .... yes
> 128-bit integer (faster PRINCE mode) ........ yes
> Memory map (share/page large files) ......... yes
>
> Development options (these may hurt performance when enabled):
> Memdbg memory debugging settings ............ disabled
> AddressSanitizer ("ASan") ................... disabled
> UndefinedBehaviorSanitizer ("UbSan") ........ disabled
>
>
> Now, when I try to see the Cuda devices I always an abnormal number of
> CUDA devices found and more surprising it changes every time .... and the
> Device name is always gibberish
>
> lrocha@...e8:~/JohnTheRipper/run$ ./john --list=cuda-devices | more
> 35436117 CUDA devices found:
> CUDA Device #0
> Name: U�
>
> lrocha@...e8:~/JohnTheRipper/run$ ./john --list=cuda-devices | more
> 46810709 CUDA devices found:
> CUDA Device #0
> Name: UF�
>
>
>
> Any idea what could be?
>
> Thanks
> Luis
>
>
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
