fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-24 21:50:12 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

intel/elk: Fix optimize_extract_to_float for i2f of unsigned extract

Browse source 
Fixes fs-uint-to-float-of-extract-int8.shader_test and
fs-uint-to-float-of-extract-int16.shader_test added by piglit!883.

v2: Expand the comment explaining the potential problem. Suggested by
Caio.

Fixes: e6022281f2 ("intel/elk: Rename files to use elk prefix")
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27891>
This commit is contained in:
Ian Romanick 2024-02-29 09:21:20 -08:00
parent fc2360167c
commit 0fa17962d6
1 changed files with 38 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

43
src/intel/compiler/elk/elk_fs_nir.cpp
View file

@ -478,6 +478,9 @@ optimize_extract_to_float(nir_to_elk_state &ntb, nir_alu_instr *instr,
const intel_device_info *devinfo = ntb.devinfo;
const fs_builder &bld = ntb.bld;
/* No fast path for f16 or f64. */
assert(instr->op == nir_op_i2f32 || instr->op == nir_op_u2f32);
if (!instr->src[0].src.ssa->parent_instr)
return false;
@ -487,16 +490,46 @@ optimize_extract_to_float(nir_to_elk_state &ntb, nir_alu_instr *instr,
nir_alu_instr *src0 =
nir_instr_as_alu(instr->src[0].src.ssa->parent_instr);
if (src0->op != nir_op_extract_u8 && src0->op != nir_op_extract_u16 &&
src0->op != nir_op_extract_i8 && src0->op != nir_op_extract_i16)
unsigned bytes;
bool is_signed;
switch (src0->op) {
case nir_op_extract_u8:
case nir_op_extract_u16:
bytes = src0->op == nir_op_extract_u8 ? 1 : 2;
/* i2f(extract_u8(a, b)) and u2f(extract_u8(a, b)) produce the same
* result. Ditto for extract_u16.
*/
is_signed = false;
break;
case nir_op_extract_i8:
case nir_op_extract_i16:
bytes = src0->op == nir_op_extract_i8 ? 1 : 2;
/* The fast path can't handle u2f(extract_i8(a, b)) because the implicit
* sign extension of the extract_i8 is lost. For example,
* u2f(extract_i8(0x0000ff00, 1)) should produce 4294967295.0, but a
* fast path could either give 255.0 (by implementing the fast path as
* u2f(extract_u8(x))) or -1.0 (by implementing the fast path as
* i2f(extract_i8(x))). At one point in time, we incorrectly implemented
* the former.
*/
if (instr->op != nir_op_i2f32)
return false;
is_signed = true;
break;
default:
return false;
}
unsigned element = nir_src_as_uint(src0->src[1].src);
/* Element type to extract.*/
const elk_reg_type type = elk_int_type(
src0->op == nir_op_extract_u16 || src0->op == nir_op_extract_i16 ? 2 : 1,
src0->op == nir_op_extract_i16 || src0->op == nir_op_extract_i8);
const elk_reg_type type = elk_int_type(bytes, is_signed);
elk_fs_reg op0 = get_nir_src(ntb, src0->src[0].src);
op0.type = elk_type_for_nir_type(devinfo,