/
LR_02_050: Instrinsify math/bits on riscv64
LR_02_050: Instrinsify math/bits on riscv64
About
Many of the functions in math.bits translate directly to a small number of CPU instruction. On other architectures the GO compiler inserts these code sequences directly into the calling function and there's no need to call a generic function that performs the computation using a sequence of bit manipulations. This instrinsification of math/bits is not currently performed on riscv64, even though RVA22U64 gives us the instructions we need to do this. In short we need to add riscv64 instrinsics to the math/bits package where appropriate.
, multiple selections available,
Related content
LR_02_0020: Implement GORISCV64 for compilation baseline
LR_02_0020: Implement GORISCV64 for compilation baseline
More like this
LR_02_038: Add bitmanip intrinsics to the Go assembler
LR_02_038: Add bitmanip intrinsics to the Go assembler
More like this
LR_02_040: Document the Go assembler
LR_02_040: Document the Go assembler
More like this
LR_02_024: Use Bitmanip instruction in compiler where optimal
LR_02_024: Use Bitmanip instruction in compiler where optimal
More like this
LR_02_041: Improve error detection and error reporting on the Go riscv64 assembler
LR_02_041: Improve error detection and error reporting on the Go riscv64 assembler
More like this
LR_02_039: Implement go tool objdump for riscv64
LR_02_039: Implement go tool objdump for riscv64
More like this