I'd like to extract the most performance possible from oclvanitygen. Can anyone tell me what the best practices are with this tool?

Examples may include:

  • Use operating system X, or video card Y
  • Ensure that the file specified with the -F option has more than 7 characters per row
  • Ensure that there are no more than Z quantity of lines in the file
  • Set a custom grid size with the -g option (what is this?)
  • How many work items per thread ( -w option)
  • What custom options should I use (-d option)
  • The -d option selects which device to use. 0 is normally CPU and 1 is normally GPU.
    – lurf jurv
    Jan 10 '13 at 15:12

1) Make sure the Vanitygen you use has the AMD patch (if you have an AMD GPU, Vanitygen Plus has it)

2) Update your GPU drivers (can't be done on Mac)

3) Play with each of these constants. Their effect is different among the GPUs and drivers, and you may never know if they can have a (relatively) significant positive effect performance.

-w <worksize> Set work items per thread in a work unit
-t <threads>  Set target thread count per multiprocessor
-g <x>x<y>    Set grid size
-b <invsize>  Set modular inverse ops per thread

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.