llrCUDA version 3.8.3 is released!

[Jean Penné]

Originally Posted by Jeff Gilchrist

Just did another test using our latest PRP with beta3:

llrCUDA - GPU Version 3.8.3b3 ; linked with CUDA Version 8.0.44
2^13380298-27 is a Fermat Probable prime! (4027872 decimal digits) Time : 136632.210 sec.

By comparison using 4 threads on an Intel Core i7-6700K took (for base 3-Fermat PRP test): 12456.743 sec.

So the GPU version is running about 11 times slower. Is that what you would expect for speed with a fermat number like this? Are there specific types of numbers that should be a lot faster with the GPU code?

You are Fermat testing a base two and k == 1 number ; so, unfortunately, it is optimal for this present version of the program, but this speed depends on the GPU card you are using...
Indeed, your multi-core CPU is much faster!

Regards,
Jean

[henryzz]

Originally Posted by Jean Penné

You are Fermat testing a base two and k == 1 number ; so, unfortunately, it is optimal for this present version of the program, but this speed depends on the GPU card you are using...
Indeed, your multi-core CPU is much faster!

Regards,
Jean

If you look at post 9, he is using a P100 which is quite a good gpu.

[Jean Penné]

Originally Posted by Jean Penné

You are Fermat testing a base two and k == 1 number ; so, unfortunately, it is optimal for this present version of the program, but this speed depends on the GPU card you are using...
Indeed, your multi-core CPU is much faster!

Regards,
Jean

I was wrong in this post :
The test is all but optimal, because it is done in generic modular reduction mode!
That is to say : this version of llrCUDA is not presently suited to test 2^n+c numbers when abs(c) != 1 because generic mode is forced.

sorry for this drawback...

Jean