Andrew Brown 54b45d28a3 x64: lower fcvt_from_uint to VCVTUDQ2PS when possible ...

When AVX512VL and AVX512F are available, use a single instruction
(`VCVTUDQ2PS`) instead of a length 9-instruction sequence. This
optimization is a port from the legacy x86 backend.

2021-05-19 12:20:11 -07:00

..

unwind

Allow unwind support to work without a frame pointer

2021-04-14 15:32:31 +02:00

args.rs

x64: lower fcvt_from_uint to VCVTUDQ2PS when possible

2021-05-19 12:20:11 -07:00

emit_tests.rs

x64: lower fcvt_from_uint to VCVTUDQ2PS when possible

2021-05-19 12:20:11 -07:00

emit.rs

x64: lower fcvt_from_uint to VCVTUDQ2PS when possible

2021-05-19 12:20:11 -07:00

mod.rs

x64: implement vselect with variable blend instructions

2021-05-17 11:23:33 -07:00

regs.rs

x86-64 Windows fastcall ABI support.

2021-03-03 19:53:18 -08:00

unwind.rs

Rework/simplify unwind infrastructure and implement Windows unwind.

2021-03-11 20:03:52 -08:00