Added alternate reduce-add/reduce-mul intrinsics.

The new reduce_add/reduce_mul procs perform the corresponding arithmetic
reduction in different orders than sequential order. These alternative
orders can often offer better SIMD hardware utilization.

Two different orders are added: pair-wise (operating on pairs of
adjacent elements) or bisection-wise (operating element-wise on the
first and last N/2 elements of the vector).
This commit is contained in:
Barinzaya
2025-05-05 16:38:45 -04:00
parent 9814370659
commit dd5b7852ce
5 changed files with 274 additions and 2 deletions
+8
View File
@@ -170,8 +170,12 @@ BuiltinProc__simd_begin,
BuiltinProc_simd_extract,
BuiltinProc_simd_replace,
BuiltinProc_simd_reduce_add_bisect,
BuiltinProc_simd_reduce_mul_bisect,
BuiltinProc_simd_reduce_add_ordered,
BuiltinProc_simd_reduce_mul_ordered,
BuiltinProc_simd_reduce_add_pairs,
BuiltinProc_simd_reduce_mul_pairs,
BuiltinProc_simd_reduce_min,
BuiltinProc_simd_reduce_max,
BuiltinProc_simd_reduce_and,
@@ -518,8 +522,12 @@ gb_global BuiltinProc builtin_procs[BuiltinProc_COUNT] = {
{STR_LIT("simd_extract"), 2, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_replace"), 3, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_add_bisect"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_mul_bisect"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_add_ordered"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_mul_ordered"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_add_pairs"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_mul_pairs"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_min"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_max"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},
{STR_LIT("simd_reduce_and"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics},