shader: add S_BITCMP0_B64 and S_BITCMP1_B64 (#688)

* shader: add S_BITCMP0_B64 and S_BITCMP1_B64

* test: move S_BITCMP_B64 coverage to compute suite

* test: use the existing compute CTest entry for bit comparisons

* test: remove the dedicated S_BITCMP_B64 command-line selector

---------

Co-authored-by: nmzik <Nmzik@mail.ru>
This commit is contained in:
Elija
2026-09-20 17:36:12 +02:00
committed by GitHub
parent c3cc8454b9
commit 549495575c
6 changed files with 132 additions and 1 deletions
@@ -74,7 +74,8 @@ constexpr OpcodeMap SOPC_OPCODE_LIST[] = {
{0x03u, Opcode::S_CMP_GE_I32}, {0x04u, Opcode::S_CMP_LT_I32}, {0x05u, Opcode::S_CMP_LE_I32},
{0x06u, Opcode::S_CMP_EQ_U32}, {0x07u, Opcode::S_CMP_LG_U32}, {0x08u, Opcode::S_CMP_GT_U32},
{0x09u, Opcode::S_CMP_GE_U32}, {0x0au, Opcode::S_CMP_LT_U32}, {0x0bu, Opcode::S_CMP_LE_U32},
{0x0cu, Opcode::S_BITCMP0_B32}, {0x0du, Opcode::S_BITCMP1_B32}, {0x12u, Opcode::S_CMP_EQ_U64},
{0x0cu, Opcode::S_BITCMP0_B32}, {0x0du, Opcode::S_BITCMP1_B32},
{0x0eu, Opcode::S_BITCMP0_B64}, {0x0fu, Opcode::S_BITCMP1_B64}, {0x12u, Opcode::S_CMP_EQ_U64},
{0x13u, Opcode::S_CMP_LG_U64},
};
@@ -73,6 +73,8 @@ enum class Opcode {
S_SUB_I32,
S_BITCMP0_B32,
S_BITCMP1_B32,
S_BITCMP0_B64,
S_BITCMP1_B64,
S_BITSET0_B32,
S_BITSET1_B32,
S_BITSET0_B64,
@@ -507,6 +507,18 @@ bool Translator::S_BITCMP_B32(const Decoder::Instruction& inst, bool expected) {
return true;
}
bool Translator::S_BITCMP_B64(const Decoder::Instruction& inst, bool expected) {
const auto value = ReadU32Pair(inst.src0);
const auto offset = ir.BitwiseAnd(ReadU32(inst.src1), IR::U32(IR::Value(63u)));
const auto word_bit = ir.BitwiseAnd(offset, IR::U32(IR::Value(31u)));
const auto word =
ir.Select(ir.ULessThan(offset, IR::U32(IR::Value(32u))), value[0], value[1]);
const auto bit =
IR::U32(ir.Emit(IR::ValueOpcode::BitFieldUExtract, {word, word_bit, IR::Value(1u)}));
WriteCompareResult(inst.dst, ir.IEqual(bit, IR::U32(IR::Value(expected ? 1u : 0u))));
return true;
}
bool Translator::V_ALIGNBIT_B32(const Decoder::Instruction& inst) {
const auto hi = ReadU32(inst.src0);
const auto lo = ReadU32(inst.src1);
@@ -208,6 +208,8 @@ bool Translator::EmitScalar(const Decoder::Instruction& inst) {
case O::S_BFE_U64: return S_BFE_U64(inst);
case O::S_BITCMP0_B32: return S_BITCMP_B32(inst, false);
case O::S_BITCMP1_B32: return S_BITCMP_B32(inst, true);
case O::S_BITCMP0_B64: return S_BITCMP_B64(inst, false);
case O::S_BITCMP1_B64: return S_BITCMP_B64(inst, true);
case O::S_PACK_LL_B32_B16: return PackB16(inst, false, false);
case O::S_PACK_LH_B32_B16: return PackB16(inst, false, true);
case O::S_PACK_HH_B32_B16: return PackB16(inst, true, true);
@@ -205,6 +205,7 @@ private:
bool V_BFE_U32(const Decoder::Instruction& inst, bool sign);
bool V_BFI_B32(const Decoder::Instruction& inst);
bool S_BITCMP_B32(const Decoder::Instruction& inst, bool expected);
bool S_BITCMP_B64(const Decoder::Instruction& inst, bool expected);
bool V_ALIGNBIT_B32(const Decoder::Instruction& inst);
bool V_ALIGNBYTE_B32(const Decoder::Instruction& inst);
bool V_LSHL_ADD_U32(const Decoder::Instruction& inst);