Closed frank-dittrich closed 2 months ago
For some formats, the speed isn't bad (compared to CPU format or POCL), but for others it is much worse.
[fd@localhost run]$ ./john --test --device=1 --format=md5crypt-opencl
Device 1: Intel(R) UHD Graphics 620 [0x5917]
Benchmarking: md5crypt-opencl, crypt(3) $1$ [MD5 OpenCL]... LWS=128 GWS=24576 (192 blocks) DONE
Warning: "Many salts" test limited: 17/256
Many salts: 204800 c/s real, 244322 c/s virtual
Only one salt: 201831 c/s real, 242902 c/s virtual
[fd@localhost run]$ ./john --test --device=2 --format=md5crypt-opencl
Device 2: pthread-Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz
Benchmarking: md5crypt-opencl, crypt(3) $1$ [MD5 OpenCL]... LWS=8 GWS=128 (16 blocks) DONE
Many salts: 43136 c/s real, 6357 c/s virtual
Only one salt: 42496 c/s real, 6221 c/s virtual
[fd@localhost run]$ ./john --test --format=md5crypt
Will run 8 OpenMP threads
Benchmarking: md5crypt, crypt(3) $1$ (and variants) [MD5 256/256 AVX2 8x3]... (8xOMP) DONE
Many salts: 311424 c/s real, 39099 c/s virtual
Only one salt: 298752 c/s real, 37484 c/s virtual
[fd@localhost run]$ ./john --test --device=1 --format=bcrypt-opencl
Device 1: Intel(R) UHD Graphics 620 [0x5917]
LWS=8 GWS=4096 Benchmarking: bcrypt-opencl ("$2a$05", 32 iterations) [Blowfish OpenCL]... DONE
Speed for cost 1 (iteration count) of 32
Raw: 416 c/s real, 74472 c/s virtual
[fd@localhost run]$ ./john --test --device=2 --format=bcrypt-opencl
Device 2: pthread-Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz
LWS=16 GWS=256 Benchmarking: bcrypt-opencl ("$2a$05", 32 iterations) [Blowfish OpenCL]... DONE
Speed for cost 1 (iteration count) of 32
Raw: 4035 c/s real, 537 c/s virtual
fd@localhost run]$ ./john --test --format=bcrypt
Will run 8 OpenMP threads
Benchmarking: bcrypt ("$2a$05", 32 iterations) [Blowfish 32/64 X3]... (8xOMP) DONE
Speed for cost 1 (iteration count) of 32
Raw: 5516 c/s real, 693 c/s virtual
Just in case this is useful:
$ ./john --list=opencl-devices --devices=1
Platform #0 name: Intel(R) OpenCL HD Graphics, version: OpenCL 3.0
Device #0 (1) name: Intel(R) UHD Graphics 620 [0x5917]
Device vendor: Intel(R) Corporation
Device type: GPU (LE)
Device version: OpenCL 3.0 NEO
Driver version: 21.15.19533
Native vector widths: char 16, short 8, int 4, long 1
Preferred vector width: char 16, short 8, int 4, long 1
Global Memory: 25482 MiB
Global Memory Cache: 512 KiB
Local Memory: 64 KiB (Local)
Constant Buffer size: 4095 MiB
Max memory alloc. size: 4095 MiB
Max clock (MHz): 1150
Profiling timer res.: 83 ns
Max Work Group Size: 256
Parallel compute cores: 24
Stream processors: 192 (24 x 8)
Speed index: 220800
It looks much better than Beignet, only axcrypt2-opencl and mscash2-opencl failed.
That's great to know and have recorded in here. No need to track this as an issue, so closing.
Since Beignet seems to be deprecated, I tested NEO OpenCL
https://github.com/intel/compute-runtime/blob/master/FAQ.md https://github.com/intel/compute-runtime/blob/master/opencl/doc/DISTRIBUTIONS.md
It looks much better than Beignet, only axcrypt2-opencl and mscash2-opencl failed.