Are my bcrypt results still realistic?

[teka]

Hello everyone. I need some bcrypt speed results as a reference. So I ran oclHashcat in bcrypt mode.

But I had to set the following parameters: -n 2 -u 4
Any parameter above that gave me "ERROR: cuStreamSynchronize() 700".

Now I want to know if my results are still reasonable for being used as a reference against another hash function implemented on the same GPU: I want to say: look, hash function xyz has this speed, while bcrypt has that speed although the CPU runtimes are the same.
Can I do that or should I assume that bcrypt could normaly be attacked more efficient?


P.S.: GeForce GTX480

[atom]

Yes, bcrypt is one of those rare hashes that runs on CPU as fast as on GPU

[teka]

Yes, bcrypt is one of those rare hashes that runs on CPU as fast as on GPU

So, you think I can reasonably present those hash rates in a scientific publication as the hash rates achievable with my card, and I don't have to multiply by the factor I manually downgraded?

[atom]

Here are some numbers:

CPU AMD FXâ„¢-8120, stock clocked, hashcat v0.48 with XOP instruction support, 8 threads: 5030 H/s
GPU AMD hd7970, stock clocked, oclHashcat v1.20 in benchmark mode: 4217 H/s

Both the CPU and the GPU are not the best models but good ones, means they are in the top league.
Both used the same hash: $2a$05$....

[teka]

Here are some numbers:

CPU AMD FXâ„¢-8120, stock clocked, hashcat v0.48 with XOP instruction support, 8 threads: 5030 H/s
GPU AMD hd7970, stock clocked, oclHashcat v1.20 in benchmark mode: 4217 H/s

Both the CPU and the GPU are not the best models but good ones, means they are in the top league.
Both used the same hash: $2a$05$....

OK. The hash rate I achieved for cost=5 was 1279 H/s. I think that sounds reasonable.

Thanks for the help.