On 8/3/23 13:37, Mariam Harutyunyan via Gcc-patches wrote:
This patch adds CRC support for the RISC-V architecture. It adds internal
functions and built-ins specifically designed to handle CRC computations
efficiently.
If the target is ZBC, the clmul instruction is used for the CRC code
generation; otherwise, table-based CRC is generated. A table with 256
elements is used to store precomputed CRCs.
These CRC calculation algorithms have higher performance than the naive CRC
calculation algorithm.
So just a little bit of additional information for others that might be
looking at this thread.
The end goal here is to actually detect bitwise CRC implementations in
the gimple optimizers and turn them into table lookups or carryless
multiplies in RTL.
Mariam has that working end-to-end and has proposed a talk for the
Cauldron on the topic.
The idea here is to carve out the RTL side which we think provides
potential value to end users (the ability to use the builtin to get an
performant CRC implementation) and to get community feedback on the
implementation.
So this isn't the end of the line for this work, just the first carved
out chunk.
Jeff
