Closed Vectorized closed 1 year ago
Basically, spams macros in the OpenCL Kernel before building the program, such that we can avoid transferring the entire message each time.
Around 20% to 25% faster.
For hunting leading zeros, use 255 as the total zeros threshold for better performance.
255
Basically, spams macros in the OpenCL Kernel before building the program, such that we can avoid transferring the entire message each time.
Around 20% to 25% faster.
For hunting leading zeros, use
255as the total zeros threshold for better performance.