Simon Josefsson <si...@josefsson.org> writes: > However, the Salsa20 stream cipher seems to use the core function with > uint32_t inputs, and scrypt does too. So perhaps we should expose a > uint32_t->uint32_t interface as well that does no endian conversion. It > seems scrypt doesn't little endian convert inputs but expects the output > uint32_t to be little endian converted though... I'll bring this up > with Colin.
Nope, I was mistaken. Still it seems that an uint32_t->uint32_t interface without endian conversion may be beneficial for speed. Then scrypt won't have to convert from uint32_t to uint8_t just to have salsa convert from uint8_t to uint32_t back again, do the operation and convert it from uint32_t to uint8_t and have scrypt convert it from uint8_t to uint32_t. I haven't measured how much overhead endian conversion really consumes though. Do you think a uint32_t->uint32_t interface would be acceptable? Externally the uint32 interface could be seen as a convenience function, although internally it would be the core and the uint8_t interface would just do endian conversion. /Simon _______________________________________________ nettle-bugs mailing list nettle-bugs@lists.lysator.liu.se http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs