[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAKGDhHXnLE9KWG7BXQ1nZfE7TY9AsT1EKpoRsiD+SE-jtPN_Hw@mail.gmail.com>
Date: Sat, 15 Aug 2015 16:40:46 +0200
From: Agnieszka Bielec <bielecagnieszka8@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: PHC: Argon2 on GPU
I wanted to manually measure a time in one crypt_all() but somehow it
doesn't work when GWS=x is not set (I tested 2 methods to measure
time)
[a@...er run]$ ./john --test --format=argon2i-opencl --v=4
Benchmarking: argon2i-opencl [Blake2 OpenCL]...
memory per hash : 1.46 MB
Device 0: Tahiti [AMD Radeon HD 7900 Series]
Options used: -I ./kernels -cl-mad-enable -D__GPU__ -DDEVICE_INFO=138
-DDEV_VER_MAJOR=1800 -DDEV_VER_MINOR=5 -D_OPENCL_COMPILER
-DBINARY_SIZE=256 -DSALT_SIZE=64 -DPLAINTEXT_LENGTH=32
Calculating best global worksize (GWS); max. 1s single kernel invocation.
crypt all 256 256 64
Elapsed time: 0.030995 seconds
gws: 256 387 c/s 387 rounds/s 660.142ms per crypt_all()!
crypt all 512 512 64
Elapsed time: 0.000000 seconds
gws: 512 718 c/s 718 rounds/s 712.825ms per crypt_all()+
crypt all 1024 1024 64
Elapsed time: 0.000000 seconds
gws: 1024 1299 c/s 1299 rounds/s 788.154ms per crypt_all()+
Local worksize (LWS) 64, global worksize (GWS) 1024
crypt all 1 64 64
Elapsed time: 0.270959 seconds
crypt all 2 64 64
Elapsed time: 0.000000 seconds
crypt all 3 64 64
Elapsed time: 0.000000 seconds
crypt all 4 64 64
Elapsed time: 0.000000 seconds
crypt all 5 64 64
Elapsed time: 0.000000 seconds
crypt all 7 64 64
Elapsed time: 0.000000 seconds
crypt all 10 64 64
Elapsed time: 0.000000 seconds
crypt all 14 64 64
Elapsed time: 0.000000 seconds
crypt all 1024 1024 64
Elapsed time: 0.000000 seconds
using different password for benchmarking
crypt all 1024 1024 64
Elapsed time: 0.000000 seconds
qqqqqqqqqqqqqqqqqqqqqqqqq
crypt all 1024 1024 64
Elapsed time: 0.000000 seconds
qqqqqqqqqqqqqqqqqqqqqqqqq
DONE
Speed for cost 1 (t) of 3, cost 2 (m) of 1500, cost 3 (l) of 1
ten int 1024
clock :f500000000000000
aaa Many salts: 417 c/s real, 102400 c/s virtual
zzzzz Only one salt: 419 c/s real, 102400 c/s virtual
___
[a@...er run]$ GWS=1024 ./john --test --format=argon2i-opencl --v=4
Benchmarking: argon2i-opencl [Blake2 OpenCL]...
memory per hash : 1.46 MB
Device 0: Tahiti [AMD Radeon HD 7900 Series]
Local worksize (LWS) 64, global worksize (GWS) 1024
crypt all 1 64 64
Elapsed time: 0.139979 seconds
crypt all 2 64 64
Elapsed time: 0.002000 seconds
crypt all 3 64 64
Elapsed time: 0.002000 seconds
crypt all 4 64 64
Elapsed time: 0.002000 seconds
crypt all 5 64 64
Elapsed time: 0.001000 seconds
crypt all 7 64 64
Elapsed time: 0.000999 seconds
crypt all 10 64 64
Elapsed time: 0.001000 seconds
crypt all 14 64 64
Elapsed time: 0.000999 seconds
crypt all 1024 1024 64
Elapsed time: 0.002000 seconds
using different password for benchmarking
crypt all 1024 1024 64
Elapsed time: 0.002000 seconds
crypt all 1024 1024 64
Elapsed time: 0.003000 seconds
qqqqqqqqqqqqqqqqqqqqqqqqq
crypt all 1024 1024 64
Elapsed time: 0.002999 seconds
crypt all 1024 1024 64
Elapsed time: 0.003000 seconds
qqqqqqqqqqqqqqqqqqqqqqqqq
DONE
Speed for cost 1 (t) of 3, cost 2 (m) of 1500, cost 3 (l) of 1
ten int 2048
clock :9d00000000000000
aaa Many salts: 1304 c/s real, 204800 c/s virtual
zzzzz Only one salt: 1304 c/s real, 204800 c/s virtual
______________________
I wanted to test bcrypt-opencl but --v=4 doesn't work here. why?
[a@...er run]$ ./john --test --format=bcrypt-opencl --v=4
Device 0: Tahiti [AMD Radeon HD 7900 Series]
Local worksize (LWS) 4, Global worksize (GWS) 2048
Benchmarking: bcrypt-opencl ("$2a$05", 32 iterations) [Blowfish
OpenCL]... using different password for benchmarking
qqqqqqqqqqqqqqqqqqqqqqqqq
DONE
Speed for cost 1 (iteration count) of 32
ten int 6144
clock :9000000000000000
aaa Raw: 4266 c/s real, 307200 c/s virtual
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.
