summaryrefslogtreecommitdiff
path: root/src/wasm
diff options
context:
space:
mode:
Diffstat (limited to 'src/wasm')
-rw-r--r--src/wasm/literal.cpp51
-rw-r--r--src/wasm/wasm-binary.cpp32
-rw-r--r--src/wasm/wasm-stack.cpp24
-rw-r--r--src/wasm/wasm-validator.cpp8
4 files changed, 110 insertions, 5 deletions
diff --git a/src/wasm/literal.cpp b/src/wasm/literal.cpp
index d60e2f8a9..65c2b4e62 100644
--- a/src/wasm/literal.cpp
+++ b/src/wasm/literal.cpp
@@ -856,6 +856,10 @@ Literal Literal::convertUIToF64() const {
WASM_UNREACHABLE("invalid type");
}
+Literal Literal::convertF32ToF16() const {
+ return Literal(fp16_ieee_from_fp32_value(getf32()));
+}
+
template<typename F> struct AsInt { using type = void; };
template<> struct AsInt<float> { using type = int32_t; };
template<> struct AsInt<double> { using type = int64_t; };
@@ -1791,8 +1795,7 @@ Literal Literal::splatI16x8() const { return splat<Type::i32, 8>(*this); }
Literal Literal::splatI32x4() const { return splat<Type::i32, 4>(*this); }
Literal Literal::splatI64x2() const { return splat<Type::i64, 2>(*this); }
Literal Literal::splatF16x8() const {
- uint16_t f16 = fp16_ieee_from_fp32_value(getf32());
- return splat<Type::i32, 8>(Literal(f16));
+ return splat<Type::i32, 8>(convertF32ToF16());
}
Literal Literal::splatF32x4() const { return splat<Type::f32, 4>(*this); }
Literal Literal::splatF64x2() const { return splat<Type::f64, 2>(*this); }
@@ -1848,7 +1851,7 @@ Literal Literal::replaceLaneI64x2(const Literal& other, uint8_t index) const {
}
Literal Literal::replaceLaneF16x8(const Literal& other, uint8_t index) const {
return replace<8, &Literal::getLanesF16x8>(
- *this, Literal(fp16_ieee_from_fp32_value(other.getf32())), index);
+ *this, other.convertF32ToF16(), index);
}
Literal Literal::replaceLaneF32x4(const Literal& other, uint8_t index) const {
return replace<4, &Literal::getLanesF32x4>(*this, other, index);
@@ -2286,14 +2289,20 @@ Literal Literal::geF64x2(const Literal& other) const {
other);
}
+static Literal passThrough(const Literal& literal) { return literal; }
+static Literal toFP16(const Literal& literal) {
+ return literal.convertF32ToF16();
+}
+
template<int Lanes,
LaneArray<Lanes> (Literal::*IntoLanes)() const,
- Literal (Literal::*BinaryOp)(const Literal&) const>
+ Literal (Literal::*BinaryOp)(const Literal&) const,
+ Literal (*Convert)(const Literal&) = passThrough>
static Literal binary(const Literal& val, const Literal& other) {
LaneArray<Lanes> lanes = (val.*IntoLanes)();
LaneArray<Lanes> other_lanes = (other.*IntoLanes)();
for (size_t i = 0; i < Lanes; ++i) {
- lanes[i] = (lanes[i].*BinaryOp)(other_lanes[i]);
+ lanes[i] = Convert((lanes[i].*BinaryOp)(other_lanes[i]));
}
return Literal(lanes);
}
@@ -2418,6 +2427,38 @@ Literal Literal::subI64x2(const Literal& other) const {
Literal Literal::mulI64x2(const Literal& other) const {
return binary<2, &Literal::getLanesI64x2, &Literal::mul>(*this, other);
}
+Literal Literal::addF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::add, &toFP16>(*this,
+ other);
+}
+Literal Literal::subF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::sub, &toFP16>(*this,
+ other);
+}
+Literal Literal::mulF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::mul, &toFP16>(*this,
+ other);
+}
+Literal Literal::divF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::div, &toFP16>(*this,
+ other);
+}
+Literal Literal::minF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::min, &toFP16>(*this,
+ other);
+}
+Literal Literal::maxF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::max, &toFP16>(*this,
+ other);
+}
+Literal Literal::pminF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::pmin, &toFP16>(*this,
+ other);
+}
+Literal Literal::pmaxF16x8(const Literal& other) const {
+ return binary<8, &Literal::getLanesF16x8, &Literal::pmax, &toFP16>(*this,
+ other);
+}
Literal Literal::addF32x4(const Literal& other) const {
return binary<4, &Literal::getLanesF32x4, &Literal::add>(*this, other);
}
diff --git a/src/wasm/wasm-binary.cpp b/src/wasm/wasm-binary.cpp
index 865ca39ca..e84639801 100644
--- a/src/wasm/wasm-binary.cpp
+++ b/src/wasm/wasm-binary.cpp
@@ -6174,6 +6174,38 @@ bool WasmBinaryReader::maybeVisitSIMDBinary(Expression*& out, uint32_t code) {
curr = allocator.alloc<Binary>();
curr->op = ExtMulHighUVecI64x2;
break;
+ case BinaryConsts::F16x8Add:
+ curr = allocator.alloc<Binary>();
+ curr->op = AddVecF16x8;
+ break;
+ case BinaryConsts::F16x8Sub:
+ curr = allocator.alloc<Binary>();
+ curr->op = SubVecF16x8;
+ break;
+ case BinaryConsts::F16x8Mul:
+ curr = allocator.alloc<Binary>();
+ curr->op = MulVecF16x8;
+ break;
+ case BinaryConsts::F16x8Div:
+ curr = allocator.alloc<Binary>();
+ curr->op = DivVecF16x8;
+ break;
+ case BinaryConsts::F16x8Min:
+ curr = allocator.alloc<Binary>();
+ curr->op = MinVecF16x8;
+ break;
+ case BinaryConsts::F16x8Max:
+ curr = allocator.alloc<Binary>();
+ curr->op = MaxVecF16x8;
+ break;
+ case BinaryConsts::F16x8Pmin:
+ curr = allocator.alloc<Binary>();
+ curr->op = PMinVecF16x8;
+ break;
+ case BinaryConsts::F16x8Pmax:
+ curr = allocator.alloc<Binary>();
+ curr->op = PMaxVecF16x8;
+ break;
case BinaryConsts::F32x4Add:
curr = allocator.alloc<Binary>();
curr->op = AddVecF32x4;
diff --git a/src/wasm/wasm-stack.cpp b/src/wasm/wasm-stack.cpp
index 1c2c2c42b..b7bfea617 100644
--- a/src/wasm/wasm-stack.cpp
+++ b/src/wasm/wasm-stack.cpp
@@ -1873,6 +1873,30 @@ void BinaryInstWriter::visitBinary(Binary* curr) {
<< U32LEB(BinaryConsts::I64x2ExtmulHighI32x4U);
break;
+ case AddVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Add);
+ break;
+ case SubVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Sub);
+ break;
+ case MulVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Mul);
+ break;
+ case DivVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Div);
+ break;
+ case MinVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Min);
+ break;
+ case MaxVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Max);
+ break;
+ case PMinVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Pmin);
+ break;
+ case PMaxVecF16x8:
+ o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F16x8Pmax);
+ break;
case AddVecF32x4:
o << int8_t(BinaryConsts::SIMDPrefix) << U32LEB(BinaryConsts::F32x4Add);
break;
diff --git a/src/wasm/wasm-validator.cpp b/src/wasm/wasm-validator.cpp
index 0bdc18658..24f5379fb 100644
--- a/src/wasm/wasm-validator.cpp
+++ b/src/wasm/wasm-validator.cpp
@@ -1813,6 +1813,14 @@ void FunctionValidator::visitBinary(Binary* curr) {
case ExtMulHighSVecI64x2:
case ExtMulLowUVecI64x2:
case ExtMulHighUVecI64x2:
+ case AddVecF16x8:
+ case SubVecF16x8:
+ case MulVecF16x8:
+ case DivVecF16x8:
+ case MinVecF16x8:
+ case MaxVecF16x8:
+ case PMinVecF16x8:
+ case PMaxVecF16x8:
case AddVecF32x4:
case SubVecF32x4:
case MulVecF32x4:
generated by cgit v1.2.3 (git 2.39.1) at 2025-08-17 01:49:51 +0000