search

data61 / cuda-fixnum

Extended-precision modular arithmetic library that targets CUDA.
Other
41 stars 28 forks source link

Consider storing precomputed values in shared memory #28

Open unzvfu opened 6 years ago

unzvfu commented 6 years ago

It will relieve register pressure. Allows easily sharing the data between warps. Any other benefits?

Could cause bank conflicts when sharing a single word between multiple threads; to investigate.

unzvfu commented 4 years ago

Follow up at https://github.com/unzvfu/cuda-fixnum/issues/13.