Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[Arm64] Vector Load/Store structure instructions (dotnet#33461)
This adds support in the JIT emitter for Vector Load/Store structure instructions (C3.2.10 - Arm Architecture Reference Manual): - LD1 (1-4 registers) - LD2 - LD3 - LD4 - LD1R - LD2R - LD3R - LD4R - ST1 (1-4 registers) - ST2 - ST3 - ST4 in the following addressing modes: - Base register only - Post-indexed by a 64-bit register - Post-indexed by an immediate, equal to the number of bytes transferred Also adds support in JitDump for printing of * A SIMD vector register list. For example, ld1 {v5.16b, v6.16b, v7.16b, v8.16b}, [x9] * A SIMD vector element list. For example, st1 {v0.b}[3], [x1],#1
- Loading branch information