Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 18 Aug 2012 07:04:45 -0400
From: Erik Winkler <ewinkler@...ls.com>
To: john-dev@...ts.openwall.com
Subject: Re: OSX 10.8 (GT 650M) build fail

I get the same error on my macbook with a Nvidia 9400M GPU on MacOS 10.8:
./run/john -te=1 -fo=raw-md5-opencl -dev=1
OpenCL platform 0: Apple, 2 device(s).
Using device 1: GeForce 9400M
Compilation log: 
Error building kernel. Returned build code: -11. DEVICE_INFO=130
OpenCL error (CL_BUILD_PROGRAM_FAILURE) in file (common-opencl.c) at line (136) - (clBuildProgram failed.)

With john-1.7.9-jumbo6, I get this error:

../run/john -te=1 -fo=raw-md5-opencl -dev=1
OpenCL platform 0: Apple, 2 device(s).
Using device 1: GeForce 9400M
Compilation log: 
OpenCL error (CL_INVALID_PROGRAM_EXECUTABLE) in file (opencl_rawmd5_fmt.c) at line (218) - (Error creating kernel. Double-check kernel name?)

You may want to get the 4.2 CUDA Toolkit and try CUDA (http://www.nvidia.com/content/cuda/cuda-downloads.html).  Here is the make target I used that you will have to add to your Makefile:

macosx-x86-64-cuda:
        $(LN) x86-64.h arch.h
        @echo "#define JOHN_BLD" '"'$@'"' > john_build_rule.h
        $(MAKE) $(PROJ) \
                JOHN_OBJS="$(JOHN_OBJS) $(CUDA_OBJS) x86-64.o sse-intrinsics.o" \
                CFLAGS="$(CFLAGS) -DBSD -DHAVE_CUDA -m64 -Wno-deprecated-declarations" \
                ASFLAGS="$(ASFLAGS) -m64 -DUNDERSCORES -DBSD -DALIGN_LOG" \
                LDFLAGS="$(LDFLAGS) -L$(CUDA64PATH) -L$(NVIDIA_CUDA)/lib64 -lcudart -m64"

You also need to change your NVCC_FLAGS to support 64-bit (32-bit is so last decade!):
NVCC_FLAGS = -c -Xptxas -v -arch sm_10 -m64

And here are some of the test results I was able to get comparing GPU to CPU with latest magnum jumbo:

1) phpass
.../run/john -test -format:phpass-cuda -dev=0
Benchmarking: phpass MD5 ($P$9 lengths 1 to 15) [CUDA]... DONE
Raw:	14796 c/s real, 14864 c/s virtual

../run/john -test -format:phpass
Benchmarking: phpass MD5 ($P$9) [128/128 SSE2 intrinsics 4x4x1]... DONE
Raw:	5280 c/s real, 5280 c/s virtual

2) mscash2
./run/john -test -format:mscash2-cuda -dev=0
Benchmarking: M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 [CUDA]... DONE
Raw:	626 c/s real, 624 c/s virtual

../run/john -test -format:mscash2
Benchmarking: M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 [128/128 SSE2 intrinsics 4x]... DONE
Raw:	308 c/s real, 308 c/s virtual

Some of the CUDA hashes fail though:

../run/john -test -format:mscash-cuda -dev=0
Benchmarking: M$ Cache Hash MD4 len(pass)=8, len(salt)=13 [CUDA]... FAILED (get_hash[0](0))


Erik

> Now on GPU:
> 
> Myrices-RMBP:run myrice$ ../run/john -te=1 -fo=raw-md5-opencl -dev=1
> OpenCL platform 0: Apple, 2 device(s).
> Using device 1: GeForce GT 650M
> Compilation log:
> Error building kernel. Returned build code: -11. DEVICE_INFO=130
> OpenCL error (CL_BUILD_PROGRAM_FAILURE) in file (common-opencl.c) at
> line (136) - (clBuildProgram failed.)
> 
> I still have no idea about it. Any ideas?
> 
> Thanks
> myrice

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.