nir: Rename Boolean-related opcodes to include 32 in the name

This is a squash of a bunch of individual changes:

    nir/builder: Generate 32-bit bool opcodes transparently

    nir/algebraic: Remap Boolean opcodes to the 32-bit variant

    Use 32-bit opcodes in the NIR producers and optimizations

        Generated with a little hand-editing and the following sed commands:

        sed -i 's/nir_op_ball_fequal/nir_op_b32all_fequal/g' **/*.c
        sed -i 's/nir_op_bany_fnequal/nir_op_b32any_fnequal/g' **/*.c
        sed -i 's/nir_op_ball_iequal/nir_op_b32all_iequal/g' **/*.c
        sed -i 's/nir_op_bany_inequal/nir_op_b32any_inequal/g' **/*.c
        sed -i 's/nir_op_\([fiu]lt\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]ge\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]ne\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]eq\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fi]\)ne32g/nir_op_\1neg/g' **/*.c
        sed -i 's/nir_op_bcsel/nir_op_b32csel/g' **/*.c

     Use 32-bit opcodes in the NIR back-ends

        Generated with a little hand-editing and the following sed commands:

        sed -i 's/nir_op_ball_fequal/nir_op_b32all_fequal/g' **/*.c
        sed -i 's/nir_op_bany_fnequal/nir_op_b32any_fnequal/g' **/*.c
        sed -i 's/nir_op_ball_iequal/nir_op_b32all_iequal/g' **/*.c
        sed -i 's/nir_op_bany_inequal/nir_op_b32any_inequal/g' **/*.c
        sed -i 's/nir_op_\([fiu]lt\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]ge\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]ne\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fiu]eq\)/nir_op_\132/g' **/*.c
        sed -i 's/nir_op_\([fi]\)ne32g/nir_op_\1neg/g' **/*.c
        sed -i 's/nir_op_bcsel/nir_op_b32csel/g' **/*.c

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
This commit is contained in:
Jason Ekstrand
2018-10-18 11:44:38 -05:00
committed by Jason Ekstrand
parent b569093566
commit 80e8dfe9de
18 changed files with 323 additions and 256 deletions
+33 -33
View File
@@ -1056,10 +1056,10 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
break;
}
case nir_op_flt:
case nir_op_fge:
case nir_op_feq:
case nir_op_fne: {
case nir_op_flt32:
case nir_op_fge32:
case nir_op_feq32:
case nir_op_fne32: {
fs_reg dest = result;
const uint32_t bit_size = nir_src_bit_size(instr->src[0].src);
@@ -1068,16 +1068,16 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
brw_conditional_mod cond;
switch (instr->op) {
case nir_op_flt:
case nir_op_flt32:
cond = BRW_CONDITIONAL_L;
break;
case nir_op_fge:
case nir_op_fge32:
cond = BRW_CONDITIONAL_GE;
break;
case nir_op_feq:
case nir_op_feq32:
cond = BRW_CONDITIONAL_Z;
break;
case nir_op_fne:
case nir_op_fne32:
cond = BRW_CONDITIONAL_NZ;
break;
default:
@@ -1100,12 +1100,12 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
break;
}
case nir_op_ilt:
case nir_op_ult:
case nir_op_ige:
case nir_op_uge:
case nir_op_ieq:
case nir_op_ine: {
case nir_op_ilt32:
case nir_op_ult32:
case nir_op_ige32:
case nir_op_uge32:
case nir_op_ieq32:
case nir_op_ine32: {
fs_reg dest = result;
const uint32_t bit_size = nir_src_bit_size(instr->src[0].src);
@@ -1114,18 +1114,18 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
brw_conditional_mod cond;
switch (instr->op) {
case nir_op_ilt:
case nir_op_ult:
case nir_op_ilt32:
case nir_op_ult32:
cond = BRW_CONDITIONAL_L;
break;
case nir_op_ige:
case nir_op_uge:
case nir_op_ige32:
case nir_op_uge32:
cond = BRW_CONDITIONAL_GE;
break;
case nir_op_ieq:
case nir_op_ieq32:
cond = BRW_CONDITIONAL_Z;
break;
case nir_op_ine:
case nir_op_ine32:
cond = BRW_CONDITIONAL_NZ;
break;
default:
@@ -1178,18 +1178,18 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
case nir_op_fdot2:
case nir_op_fdot3:
case nir_op_fdot4:
case nir_op_ball_fequal2:
case nir_op_ball_iequal2:
case nir_op_ball_fequal3:
case nir_op_ball_iequal3:
case nir_op_ball_fequal4:
case nir_op_ball_iequal4:
case nir_op_bany_fnequal2:
case nir_op_bany_inequal2:
case nir_op_bany_fnequal3:
case nir_op_bany_inequal3:
case nir_op_bany_fnequal4:
case nir_op_bany_inequal4:
case nir_op_b32all_fequal2:
case nir_op_b32all_iequal2:
case nir_op_b32all_fequal3:
case nir_op_b32all_iequal3:
case nir_op_b32all_fequal4:
case nir_op_b32all_iequal4:
case nir_op_b32any_fnequal2:
case nir_op_b32any_inequal2:
case nir_op_b32any_fnequal3:
case nir_op_b32any_inequal3:
case nir_op_b32any_fnequal4:
case nir_op_b32any_inequal4:
unreachable("Lowered by nir_lower_alu_reductions");
case nir_op_fnoise1_1:
@@ -1508,7 +1508,7 @@ fs_visitor::nir_emit_alu(const fs_builder &bld, nir_alu_instr *instr)
inst->saturate = instr->dest.saturate;
break;
case nir_op_bcsel:
case nir_op_b32csel:
if (optimize_frontfacing_ternary(instr, result))
return;
@@ -109,18 +109,18 @@ analyze_boolean_resolves_block(nir_block *block)
uint8_t resolve_status;
nir_alu_instr *alu = nir_instr_as_alu(instr);
switch (alu->op) {
case nir_op_ball_fequal2:
case nir_op_ball_iequal2:
case nir_op_ball_fequal3:
case nir_op_ball_iequal3:
case nir_op_ball_fequal4:
case nir_op_ball_iequal4:
case nir_op_bany_fnequal2:
case nir_op_bany_inequal2:
case nir_op_bany_fnequal3:
case nir_op_bany_inequal3:
case nir_op_bany_fnequal4:
case nir_op_bany_inequal4:
case nir_op_b32all_fequal2:
case nir_op_b32all_iequal2:
case nir_op_b32all_fequal3:
case nir_op_b32all_iequal3:
case nir_op_b32all_fequal4:
case nir_op_b32all_iequal4:
case nir_op_b32any_fnequal2:
case nir_op_b32any_inequal2:
case nir_op_b32any_fnequal3:
case nir_op_b32any_inequal3:
case nir_op_b32any_fnequal4:
case nir_op_b32any_inequal4:
/* These are only implemented by the vec4 backend and its
* implementation emits resolved booleans. At some point in the
* future, this may change and we'll have to remove some of the
+57 -57
View File
@@ -833,34 +833,34 @@ static enum brw_conditional_mod
brw_conditional_for_nir_comparison(nir_op op)
{
switch (op) {
case nir_op_flt:
case nir_op_ilt:
case nir_op_ult:
case nir_op_flt32:
case nir_op_ilt32:
case nir_op_ult32:
return BRW_CONDITIONAL_L;
case nir_op_fge:
case nir_op_ige:
case nir_op_uge:
case nir_op_fge32:
case nir_op_ige32:
case nir_op_uge32:
return BRW_CONDITIONAL_GE;
case nir_op_feq:
case nir_op_ieq:
case nir_op_ball_fequal2:
case nir_op_ball_iequal2:
case nir_op_ball_fequal3:
case nir_op_ball_iequal3:
case nir_op_ball_fequal4:
case nir_op_ball_iequal4:
case nir_op_feq32:
case nir_op_ieq32:
case nir_op_b32all_fequal2:
case nir_op_b32all_iequal2:
case nir_op_b32all_fequal3:
case nir_op_b32all_iequal3:
case nir_op_b32all_fequal4:
case nir_op_b32all_iequal4:
return BRW_CONDITIONAL_Z;
case nir_op_fne:
case nir_op_ine:
case nir_op_bany_fnequal2:
case nir_op_bany_inequal2:
case nir_op_bany_fnequal3:
case nir_op_bany_inequal3:
case nir_op_bany_fnequal4:
case nir_op_bany_inequal4:
case nir_op_fne32:
case nir_op_ine32:
case nir_op_b32any_fnequal2:
case nir_op_b32any_inequal2:
case nir_op_b32any_fnequal3:
case nir_op_b32any_inequal3:
case nir_op_b32any_fnequal4:
case nir_op_b32any_inequal4:
return BRW_CONDITIONAL_NZ;
default:
@@ -880,20 +880,20 @@ vec4_visitor::optimize_predicate(nir_alu_instr *instr,
nir_instr_as_alu(instr->src[0].src.ssa->parent_instr);
switch (cmp_instr->op) {
case nir_op_bany_fnequal2:
case nir_op_bany_inequal2:
case nir_op_bany_fnequal3:
case nir_op_bany_inequal3:
case nir_op_bany_fnequal4:
case nir_op_bany_inequal4:
case nir_op_b32any_fnequal2:
case nir_op_b32any_inequal2:
case nir_op_b32any_fnequal3:
case nir_op_b32any_inequal3:
case nir_op_b32any_fnequal4:
case nir_op_b32any_inequal4:
*predicate = BRW_PREDICATE_ALIGN16_ANY4H;
break;
case nir_op_ball_fequal2:
case nir_op_ball_iequal2:
case nir_op_ball_fequal3:
case nir_op_ball_iequal3:
case nir_op_ball_fequal4:
case nir_op_ball_iequal4:
case nir_op_b32all_fequal2:
case nir_op_b32all_iequal2:
case nir_op_b32all_fequal3:
case nir_op_b32all_iequal3:
case nir_op_b32all_fequal4:
case nir_op_b32all_iequal4:
*predicate = BRW_PREDICATE_ALIGN16_ALL4H;
break;
default:
@@ -1340,18 +1340,18 @@ vec4_visitor::nir_emit_alu(nir_alu_instr *instr)
case nir_op_fddy_fine:
unreachable("derivatives are not valid in vertex shaders");
case nir_op_ilt:
case nir_op_ult:
case nir_op_ige:
case nir_op_uge:
case nir_op_ieq:
case nir_op_ine:
case nir_op_ilt32:
case nir_op_ult32:
case nir_op_ige32:
case nir_op_uge32:
case nir_op_ieq32:
case nir_op_ine32:
assert(nir_dest_bit_size(instr->dest.dest) < 64);
/* Fallthrough */
case nir_op_flt:
case nir_op_fge:
case nir_op_feq:
case nir_op_fne: {
case nir_op_flt32:
case nir_op_fge32:
case nir_op_feq32:
case nir_op_fne32: {
enum brw_conditional_mod conditional_mod =
brw_conditional_for_nir_comparison(instr->op);
@@ -1372,14 +1372,14 @@ vec4_visitor::nir_emit_alu(nir_alu_instr *instr)
break;
}
case nir_op_ball_iequal2:
case nir_op_ball_iequal3:
case nir_op_ball_iequal4:
case nir_op_b32all_iequal2:
case nir_op_b32all_iequal3:
case nir_op_b32all_iequal4:
assert(nir_dest_bit_size(instr->dest.dest) < 64);
/* Fallthrough */
case nir_op_ball_fequal2:
case nir_op_ball_fequal3:
case nir_op_ball_fequal4: {
case nir_op_b32all_fequal2:
case nir_op_b32all_fequal3:
case nir_op_b32all_fequal4: {
unsigned swiz =
brw_swizzle_for_size(nir_op_infos[instr->op].input_sizes[0]);
@@ -1391,14 +1391,14 @@ vec4_visitor::nir_emit_alu(nir_alu_instr *instr)
break;
}
case nir_op_bany_inequal2:
case nir_op_bany_inequal3:
case nir_op_bany_inequal4:
case nir_op_b32any_inequal2:
case nir_op_b32any_inequal3:
case nir_op_b32any_inequal4:
assert(nir_dest_bit_size(instr->dest.dest) < 64);
/* Fallthrough */
case nir_op_bany_fnequal2:
case nir_op_bany_fnequal3:
case nir_op_bany_fnequal4: {
case nir_op_b32any_fnequal2:
case nir_op_b32any_fnequal3:
case nir_op_b32any_fnequal4: {
unsigned swiz =
brw_swizzle_for_size(nir_op_infos[instr->op].input_sizes[0]);
@@ -1798,7 +1798,7 @@ vec4_visitor::nir_emit_alu(nir_alu_instr *instr)
inst->saturate = instr->dest.saturate;
break;
case nir_op_bcsel:
case nir_op_b32csel:
enum brw_predicate predicate;
if (!optimize_predicate(instr, &predicate)) {
emit(CMP(dst_null_d(), op[0], brw_imm_d(0), BRW_CONDITIONAL_NZ));