summaryrefslogtreecommitdiffstats
path: root/src/amd/common/ac_llvm_build.c
diff options
context:
space:
mode:
authorMarek Olšák <[email protected]>2017-05-19 15:02:34 +0200
committerMarek Olšák <[email protected]>2017-05-29 01:52:16 +0200
commite019ea8f4b6f4e76a6d52e9ff8d6e11b19d97a94 (patch)
treec3908268c25ac519f61c5f2c60168faabd9fe998 /src/amd/common/ac_llvm_build.c
parente1942c970ff0f5602cfc42b9954eadfc2562c952 (diff)
radeonsi: move building llvm.SI.load.const into ac_build_buffer_load
Reviewed-by: Nicolai Hähnle <[email protected]>
Diffstat (limited to 'src/amd/common/ac_llvm_build.c')
-rw-r--r--src/amd/common/ac_llvm_build.c46
1 files changed, 34 insertions, 12 deletions
diff --git a/src/amd/common/ac_llvm_build.c b/src/amd/common/ac_llvm_build.c
index 3df9f53ed7a..237e9291d41 100644
--- a/src/amd/common/ac_llvm_build.c
+++ b/src/amd/common/ac_llvm_build.c
@@ -633,14 +633,46 @@ ac_build_buffer_load(struct ac_llvm_context *ctx,
unsigned inst_offset,
unsigned glc,
unsigned slc,
- bool can_speculate)
+ bool can_speculate,
+ bool allow_smem)
{
+ LLVMValueRef offset = LLVMConstInt(ctx->i32, inst_offset, 0);
+ if (voffset)
+ offset = LLVMBuildAdd(ctx->builder, offset, voffset, "");
+ if (soffset)
+ offset = LLVMBuildAdd(ctx->builder, offset, soffset, "");
+
+ /* TODO: VI and later generations can use SMEM with GLC=1.*/
+ if (allow_smem && !glc && !slc) {
+ assert(vindex == NULL);
+
+ LLVMValueRef result[4];
+
+ for (int i = 0; i < num_channels; i++) {
+ if (i) {
+ offset = LLVMBuildAdd(ctx->builder, offset,
+ LLVMConstInt(ctx->i32, 4, 0), "");
+ }
+ LLVMValueRef args[2] = {rsrc, offset};
+ result[i] = ac_build_intrinsic(ctx, "llvm.SI.load.const.v4i32",
+ ctx->f32, args, 2,
+ AC_FUNC_ATTR_READNONE |
+ AC_FUNC_ATTR_LEGACY);
+ }
+ if (num_channels == 1)
+ return result[0];
+
+ if (num_channels == 3)
+ result[num_channels++] = LLVMGetUndef(ctx->f32);
+ return ac_build_gather_values(ctx, result, num_channels);
+ }
+
unsigned func = CLAMP(num_channels, 1, 3) - 1;
LLVMValueRef args[] = {
LLVMBuildBitCast(ctx->builder, rsrc, ctx->v4i32, ""),
vindex ? vindex : LLVMConstInt(ctx->i32, 0, 0),
- LLVMConstInt(ctx->i32, inst_offset, 0),
+ offset,
LLVMConstInt(ctx->i1, glc, 0),
LLVMConstInt(ctx->i1, slc, 0)
};
@@ -650,16 +682,6 @@ ac_build_buffer_load(struct ac_llvm_context *ctx,
const char *type_names[] = {"f32", "v2f32", "v4f32"};
char name[256];
- if (voffset) {
- args[2] = LLVMBuildAdd(ctx->builder, args[2], voffset,
- "");
- }
-
- if (soffset) {
- args[2] = LLVMBuildAdd(ctx->builder, args[2], soffset,
- "");
- }
-
snprintf(name, sizeof(name), "llvm.amdgcn.buffer.load.%s",
type_names[func]);