fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-04-28 10:50:39 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

ac/llvm: implement 16-bit packed VS outputs and FS inputs

Browse source 
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9051>
This commit is contained in:
Marek Olšák 2021-02-09 22:06:09 -05:00
parent 80b67a3b44
commit 1dff495057
3 changed files with 29 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
src/amd/llvm/ac_llvm_build.c
View file

@ -954,7 +954,7 @@ LLVMValueRef ac_build_fs_interp(struct ac_llvm_context *ctx, LLVMValueRef llvm_c
LLVMValueRef ac_build_fs_interp_f16(struct ac_llvm_context *ctx, LLVMValueRef llvm_chan,
LLVMValueRef attr_number, LLVMValueRef params, LLVMValueRef i,
LLVMValueRef j)
LLVMValueRef j, bool high_16bits)
{
LLVMValueRef args[6];
LLVMValueRef p1;
@ -962,7 +962,7 @@ LLVMValueRef ac_build_fs_interp_f16(struct ac_llvm_context *ctx, LLVMValueRef ll
args[0] = i;
args[1] = llvm_chan;
args[2] = attr_number;
args[3] = ctx->i1false;
args[3] = high_16bits ? ctx->i1true : ctx->i1false;
args[4] = params;
p1 = ac_build_intrinsic(ctx, "llvm.amdgcn.interp.p1.f16", ctx->f32, args, 5,
@ -972,7 +972,7 @@ LLVMValueRef ac_build_fs_interp_f16(struct ac_llvm_context *ctx, LLVMValueRef ll
args[1] = j;
args[2] = llvm_chan;
args[3] = attr_number;
args[4] = ctx->i1false;
args[4] = high_16bits ? ctx->i1true : ctx->i1false;
args[5] = params;
return ac_build_intrinsic(ctx, "llvm.amdgcn.interp.p2.f16", ctx->f16, args, 6,

2
src/amd/llvm/ac_llvm_build.h
View file

@ -229,7 +229,7 @@ LLVMValueRef ac_build_fs_interp(struct ac_llvm_context *ctx, LLVMValueRef llvm_c
LLVMValueRef ac_build_fs_interp_f16(struct ac_llvm_context *ctx, LLVMValueRef llvm_chan,
LLVMValueRef attr_number, LLVMValueRef params, LLVMValueRef i,
LLVMValueRef j);
LLVMValueRef j, bool high_16bits);
LLVMValueRef ac_build_fs_interp_mov(struct ac_llvm_context *ctx, LLVMValueRef parameter,
LLVMValueRef llvm_chan, LLVMValueRef attr_number,

29
src/amd/llvm/ac_nir_to_llvm.c
View file

@ -2240,7 +2240,22 @@ static void visit_store_output(struct ac_nir_context *ctx, nir_intrinsic_instr *
continue;
LLVMValueRef value = ac_llvm_extract_elem(&ctx->ac, src, chan - component);
LLVMBuildStore(ctx->ac.builder, value, ctx->abi->outputs[base * 4 + chan]);
LLVMValueRef output_addr = ctx->abi->outputs[base * 4 + chan];
if (LLVMGetElementType(LLVMTypeOf(output_addr)) == ctx->ac.f32 &&
LLVMTypeOf(value) == ctx->ac.f16) {
LLVMValueRef output, index;
/* Insert the 16-bit value into the low or high bits of the 32-bit output
* using read-modify-write.
*/
index = LLVMConstInt(ctx->ac.i32, nir_intrinsic_io_semantics(instr).high_16bits, 0);
output = LLVMBuildLoad(ctx->ac.builder, output_addr, "");
output = LLVMBuildBitCast(ctx->ac.builder, output, ctx->ac.v2f16, "");
output = LLVMBuildInsertElement(ctx->ac.builder, output, value, index, "");
value = LLVMBuildBitCast(ctx->ac.builder, output, ctx->ac.f32, "");
}
LLVMBuildStore(ctx->ac.builder, value, output_addr);
}
if (ctx->ac.postponed_kill)
@ -3202,7 +3217,8 @@ static LLVMValueRef barycentric_model(struct ac_nir_context *ctx)
static LLVMValueRef load_interpolated_input(struct ac_nir_context *ctx, LLVMValueRef interp_param,
unsigned index, unsigned comp_start,
unsigned num_components, unsigned bitsize)
unsigned num_components, unsigned bitsize,
bool high_16bits)
{
LLVMValueRef attr_number = LLVMConstInt(ctx->ac.i32, index, false);
LLVMValueRef interp_param_f;
@ -3228,7 +3244,8 @@ static LLVMValueRef load_interpolated_input(struct ac_nir_context *ctx, LLVMValu
LLVMValueRef llvm_chan = LLVMConstInt(ctx->ac.i32, comp_start + comp, false);
if (bitsize == 16) {
values[comp] = ac_build_fs_interp_f16(&ctx->ac, llvm_chan, attr_number,
ac_get_arg(&ctx->ac, ctx->args->prim_mask), i, j);
ac_get_arg(&ctx->ac, ctx->args->prim_mask), i, j,
high_16bits);
} else {
values[comp] = ac_build_fs_interp(&ctx->ac, llvm_chan, attr_number,
ac_get_arg(&ctx->ac, ctx->args->prim_mask), i, j);
@ -3352,6 +3369,9 @@ static LLVMValueRef visit_load(struct ac_nir_context *ctx, nir_intrinsic_instr *
ac_build_fs_interp_mov(&ctx->ac, LLVMConstInt(ctx->ac.i32, vertex_id, false), llvm_chan,
attr_number, ac_get_arg(&ctx->ac, ctx->args->prim_mask));
values[chan] = LLVMBuildBitCast(ctx->ac.builder, values[chan], ctx->ac.i32, "");
if (instr->dest.ssa.bit_size == 16 &&
nir_intrinsic_io_semantics(instr).high_16bits)
values[chan] = LLVMBuildLShr(ctx->ac.builder, values[chan], LLVMConstInt(ctx->ac.i32, 16, 0), "");
values[chan] =
LLVMBuildTruncOrBitCast(ctx->ac.builder, values[chan],
instr->dest.ssa.bit_size == 16 ? ctx->ac.i16 : ctx->ac.i32, "");
@ -3799,7 +3819,8 @@ static void visit_intrinsic(struct ac_nir_context *ctx, nir_intrinsic_instr *ins
unsigned index = nir_intrinsic_base(instr);
unsigned component = nir_intrinsic_component(instr);
result = load_interpolated_input(ctx, interp_param, index, component,
instr->dest.ssa.num_components, instr->dest.ssa.bit_size);
instr->dest.ssa.num_components, instr->dest.ssa.bit_size,
nir_intrinsic_io_semantics(instr).high_16bits);
break;
}
case nir_intrinsic_emit_vertex: