Mesa (master): aco: use v_bfe_u32 for unsigned reductions sign-extension on GFX6-GFX7
GitLab Mirror
gitlab-mirror at kemper.freedesktop.org
Fri Jun 5 14:30:23 UTC 2020
Module: Mesa
Branch: master
Commit: ee4bc13de2aacb7bab24a3e55e44e7e50434df94
URL: http://cgit.freedesktop.org/mesa/mesa/commit/?id=ee4bc13de2aacb7bab24a3e55e44e7e50434df94
Author: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Date: Fri Jun 5 08:54:52 2020 +0200
aco: use v_bfe_u32 for unsigned reductions sign-extension on GFX6-GFX7
Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Reviewed-by: Daniel Schürmann <daniel at schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5327>
---
src/amd/compiler/aco_lower_to_hw_instr.cpp | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/src/amd/compiler/aco_lower_to_hw_instr.cpp b/src/amd/compiler/aco_lower_to_hw_instr.cpp
index 40d466904ef..a278b66ce82 100644
--- a/src/amd/compiler/aco_lower_to_hw_instr.cpp
+++ b/src/amd/compiler/aco_lower_to_hw_instr.cpp
@@ -591,7 +591,14 @@ void emit_reduction(lower_context *ctx, aco_opcode op, ReduceOp reduce_op, unsig
sdwa->dst_sel = sdwa_udword;
bld.insert(std::move(sdwa));
} else if (ctx->program->chip_class == GFX6 || ctx->program->chip_class == GFX7) {
- bld.vop3(aco_opcode::v_bfe_i32, Definition(PhysReg{tmp}, v1),
+ aco_opcode opcode;
+
+ if (reduce_op == imin16 || reduce_op == imax16 || reduce_op == iadd16)
+ opcode = aco_opcode::v_bfe_i32;
+ else
+ opcode = aco_opcode::v_bfe_u32;
+
+ bld.vop3(opcode, Definition(PhysReg{tmp}, v1),
Operand(PhysReg{tmp}, v1), Operand(0u), Operand(16u));
}
}
More information about the mesa-commit
mailing list