Skip to content

[clang] Fix size and alignment of packed sub-byte integer vectors #161796

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
steffenlarsen wants to merge 8 commits into
base: main
Choose a base branch
from
Open

[clang] Fix size and alignment of packed sub-byte integer vectors #161796

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
545a9df
[clang] Fix size and alignment of packed sub-byte integer vectors
steffenlarsen Oct 3, 2025
4c2d4e4
Split packed size selection into a separate function and fix bit-casts
steffenlarsen Oct 6, 2025
e13eff3
Remove unused variable and fix formatting
steffenlarsen Oct 13, 2025
3176e9e
Remove null-type check and add tests for padded type bitcasts
steffenlarsen Oct 14, 2025
d4d29f8
Merge branch 'main' into sub_byte_packing
steffenlarsen Oct 27, 2025
dc452c8
Merge branch 'main' into sub_byte_packing
steffenlarsen Nov 10, 2025
644a544
Merge branch 'main' into sub_byte_packing
steffenlarsen Nov 21, 2025
ecc35c9
Merge branch 'main' into sub_byte_packing
steffenlarsen Nov 25, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions clang/include/clang/AST/ASTContext.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2677,6 +2677,16 @@ class ASTContext : public RefCountedBase<ASTContext> {
return getTypeSizeInCharsIfKnown(QualType(Ty, 0));
}

/// Return the size of an element inside a given vector type.
uint64_t getVectorElementSize(const VectorType *VTy) const {
QualType EltTy = VTy->getElementType();
if (VTy->isPackedVectorBoolType(*this))
Copy link
Collaborator

@efriedma-quic efriedma-quic Nov 25, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we might need an equivalent of isPackedVectorBoolType for bitints. Not that I expect anyone to be using bitint vectors in HLSL anytime soon, but better to be consistent.

Copy link
Contributor Author

@steffenlarsen steffenlarsen Nov 27, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The patch adds isBitIntVectorType which could be used below. Is that what you had in mind, or do you mean that isBitIntVectorType should check for HLSL as well?

Copy link
Collaborator

@efriedma-quic efriedma-quic Dec 2, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe rename isBitIntVectorType() to isPackedBitIntVectorType(), make it check for HLSL, and use it here instead of checking EltTy->isBitIntType().

return 1;
if (EltTy.getTypePtrOrNull() && EltTy->isBitIntType())
Copy link
Collaborator

@efriedma-quic efriedma-quic Oct 13, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think a vector type can have a null element type?

Copy link
Contributor Author

@steffenlarsen steffenlarsen Oct 14, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wasn't sure, but removing it doesn't seem to trigger any issues.

return EltTy->castAs<BitIntType>()->getNumBits();
return getTypeSize(EltTy);
}

/// Return the ABI-specified alignment of a (complete) type \p T, in
/// bits.
unsigned getTypeAlign(QualType T) const { return getTypeInfo(T).Align; }
Expand Down
7 changes: 7 additions & 0 deletions clang/include/clang/AST/TypeBase.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2640,6 +2640,8 @@ class alignas(TypeAlignment) Type : public ExtQualsTypeCommonBase {
// Extended vector type with bool element that is packed. HLSL doesn't pack
// its bool vectors.
bool isPackedVectorBoolType(const ASTContext &ctx) const;
// Vector type with packed _BitInt elements.
bool isBitIntVectorType() const;
bool isSubscriptableVectorType() const;
bool isMatrixType() const; // Matrix type.
bool isConstantMatrixType() const; // Constant matrix type.
Expand Down Expand Up @@ -8681,6 +8683,11 @@ inline bool Type::isExtVectorBoolType() const {
return cast<ExtVectorType>(CanonicalType)->getElementType()->isBooleanType();
}

inline bool Type::isBitIntVectorType() const {
return isVectorType() &&
cast<VectorType>(CanonicalType)->getElementType()->isBitIntType();
}

inline bool Type::isSubscriptableVectorType() const {
return isVectorType() || isSveVLSBuiltinType();
}
Expand Down
5 changes: 1 addition & 4 deletions clang/lib/AST/ASTContext.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2093,10 +2093,7 @@ TypeInfo ASTContext::getTypeInfoImpl(const Type *T) const {
case Type::ExtVector:
case Type::Vector: {
const auto *VT = cast<VectorType>(T);
TypeInfo EltInfo = getTypeInfo(VT->getElementType());
Width = VT->isPackedVectorBoolType(*this)
? VT->getNumElements()
: EltInfo.Width * VT->getNumElements();
Width = getVectorElementSize(VT) * VT->getNumElements();
// Enforce at least byte size and alignment.
Width = std::max<unsigned>(8, Width);
Align = std::max<unsigned>(8, Width);
Expand Down
48 changes: 25 additions & 23 deletions clang/lib/AST/ExprConstant.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7589,8 +7589,8 @@ class APValueToBufferConverter {
QualType EltTy = VTy->getElementType();
unsigned NElts = VTy->getNumElements();

if (VTy->isPackedVectorBoolType(Info.Ctx)) {
// Special handling for OpenCL bool vectors:
if (VTy->isPackedVectorBoolType(Info.Ctx) || VTy->isBitIntVectorType()) {
// Special handling for OpenCL bool and sub-byte vectors:
// Since these vectors are stored as packed bits, but we can't write
// individual bits to the BitCastBuffer, we'll buffer all of the elements
// together into an appropriately sized APInt and write them all out at
Expand All @@ -7599,18 +7599,21 @@ class APValueToBufferConverter {
// have to worry about writing data which should have been left
// uninitialized.
bool BigEndian = Info.Ctx.getTargetInfo().isBigEndian();
uint64_t EltSize = Info.Ctx.getVectorElementSize(VTy);

llvm::APInt Res = llvm::APInt::getZero(NElts);
llvm::APInt Res = llvm::APInt::getZero(NElts * EltSize);
for (unsigned I = 0; I < NElts; ++I) {
const llvm::APSInt &EltAsInt = Val.getVectorElt(I).getInt();
assert(EltAsInt.isUnsigned() && EltAsInt.getBitWidth() == 1 &&
"bool vector element must be 1-bit unsigned integer!");

Res.insertBits(EltAsInt, BigEndian ? (NElts - I - 1) : I);
assert(!VTy->isPackedVectorBoolType(Info.Ctx) ||
(EltAsInt.isUnsigned() && EltAsInt.getBitWidth()) == 1 &&
"bool vector element must be 1-bit unsigned integer!");
uint64_t BitOffset = EltSize * (BigEndian ? (NElts - I - 1) : I);
Res.insertBits(EltAsInt, BitOffset);
}

SmallVector<uint8_t, 8> Bytes(NElts / 8);
llvm::StoreIntToMemory(Res, &*Bytes.begin(), NElts / 8);
uint64_t NumBytes = NElts * EltSize / 8;
Copy link
Collaborator

@efriedma-quic efriedma-quic Oct 13, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this divide guaranteed to be exact? Rounding down is suspicious.

Copy link
Contributor Author

@steffenlarsen steffenlarsen Oct 14, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Because of the check that prevents a bitcast to/from padded sub-byte vectors, this should always be an exact division. If you prefer, I can change it to do a round-up division, but it wouldn't add much value in the current implementation.

SmallVector<uint8_t, 8> Bytes(NumBytes);
llvm::StoreIntToMemory(Res, &*Bytes.begin(), NumBytes);
Buffer.writeObject(Offset, Bytes);
} else {
// Iterate over each of the elements and write them out to the buffer at
Expand Down Expand Up @@ -7852,13 +7855,11 @@ class BufferToAPValueConverter {
std::optional<APValue> visit(const VectorType *VTy, CharUnits Offset) {
QualType EltTy = VTy->getElementType();
unsigned NElts = VTy->getNumElements();
unsigned EltSize =
VTy->isPackedVectorBoolType(Info.Ctx) ? 1 : Info.Ctx.getTypeSize(EltTy);

SmallVector<APValue, 4> Elts;
Elts.reserve(NElts);
if (VTy->isPackedVectorBoolType(Info.Ctx)) {
// Special handling for OpenCL bool vectors:
if (VTy->isPackedVectorBoolType(Info.Ctx) || VTy->isBitIntVectorType()) {
// Special handling for OpenCL bool and sub-byte vectors:
// Since these vectors are stored as packed bits, but we can't read
// individual bits from the BitCastBuffer, we'll buffer all of the
// elements together into an appropriately sized APInt and write them all
Expand All @@ -7867,20 +7868,22 @@ class BufferToAPValueConverter {
// we don't have to worry about reading any padding data which didn't
// actually need to be accessed.
bool BigEndian = Info.Ctx.getTargetInfo().isBigEndian();
uint64_t EltSize = Info.Ctx.getVectorElementSize(VTy);
bool IsSigned = EltTy->isSignedIntegerType();

uint64_t NumBytes = NElts * EltSize / 8;
SmallVector<uint8_t, 8> Bytes;
Bytes.reserve(NElts / 8);
if (!Buffer.readObject(Offset, CharUnits::fromQuantity(NElts / 8), Bytes))
Bytes.reserve(NumBytes);
if (!Buffer.readObject(Offset, CharUnits::fromQuantity(NumBytes), Bytes))
return std::nullopt;

APSInt SValInt(NElts, true);
llvm::LoadIntFromMemory(SValInt, &*Bytes.begin(), Bytes.size());
APSInt SValInt(NElts * EltSize);
llvm::LoadIntFromMemory(SValInt, Bytes.data(), Bytes.size());

for (unsigned I = 0; I < NElts; ++I) {
llvm::APInt Elt =
SValInt.extractBits(1, (BigEndian ? NElts - I - 1 : I) * EltSize);
Elts.emplace_back(
APSInt(std::move(Elt), !EltTy->isSignedIntegerType()));
uint64_t BitOffset = EltSize * (BigEndian ? (NElts - I - 1) : I);
llvm::APInt Elt = SValInt.extractBits(EltSize, BitOffset);
Elts.emplace_back(APSInt(std::move(Elt), !IsSigned));
}
} else {
// Iterate over each of the elements and read them from the buffer at
Expand Down Expand Up @@ -7986,8 +7989,7 @@ static bool checkBitCastConstexprEligibilityType(SourceLocation Loc,
if (const auto *VTy = Ty->getAs<VectorType>()) {
QualType EltTy = VTy->getElementType();
unsigned NElts = VTy->getNumElements();
unsigned EltSize =
VTy->isPackedVectorBoolType(Ctx) ? 1 : Ctx.getTypeSize(EltTy);
unsigned EltSize = Ctx.getVectorElementSize(VTy);

if ((NElts * EltSize) % Ctx.getCharWidth() != 0) {
// The vector's size in bits is not a multiple of the target's byte size,
Expand Down
28 changes: 15 additions & 13 deletions clang/test/CodeGenCXX/ext-int.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -573,7 +573,7 @@ void VectorTest(uint16_t4 first, uint16_t4 second) {

typedef unsigned _BitInt(4) uint4_t4 __attribute__((ext_vector_type(4)));
void VectorTest(uint4_t4 first, uint4_t4 second) {
// LIN64: define{{.*}} void @_Z10VectorTestDv4_DU4_S0_(i32 %{{.+}}, i32 %{{.+}})
// LIN64: define{{.*}} void @_Z10VectorTestDv4_DU4_S0_(i16 %{{.+}}, i16 %{{.+}})
// LIN32: define{{.*}} void @_Z10VectorTestDv4_DU4_S0_(<4 x i4> %{{.+}}, <4 x i4> %{{.+}})
// WIN64: define dso_local void @"?VectorTest@@YAXT?$__vector@U?$_UBitInt@$03@__clang@@$03@__clang@@0@Z"(<4 x i4> %{{.+}}, <4 x i4> %{{.+}})
// WIN32: define dso_local void @"?VectorTest@@YAXT?$__vector@U?$_UBitInt@$03@__clang@@$03@__clang@@0@Z"(<4 x i4> inreg %{{.+}}, <4 x i4> inreg %{{.+}})
Expand All @@ -585,23 +585,25 @@ void VectorTest(uint4_t4 first, uint4_t4 second) {

typedef unsigned _BitInt(2) uint2_t2 __attribute__((ext_vector_type(2)));
uint2_t2 TestBitIntVector2x2Alloca(uint2_t2 v1, uint2_t2 v2) {
// LIN64: define dso_local i16 @_Z25TestBitIntVector2x2AllocaDv2_DU2_S0_(i16 %[[V1Coerce:.+]], i16 %[[V2Coerce:.+]])
// LIN64: %[[RetVal:.+]] = alloca <2 x i2>, align 2
// LIN64: %[[V1Addr:.+]] = alloca <2 x i2>, align 2
// LIN64: %[[V2Addr:.+]] = alloca <2 x i2>, align 2
// LIN64: %[[RetValCoerce:.+]] = alloca i16, align 2
// LIN64: call void @llvm.memcpy.p0.p0.i64(ptr align 2 %[[RetValCoerce]], ptr align 2 %[[RetVal]], i64 1, i1 false)
// LIN64: %[[Ret:.+]] = load i16, ptr %[[RetValCoerce]], align 2
// LIN64: ret i16 %[[Ret]]
// LIN64: define dso_local i8 @_Z25TestBitIntVector2x2AllocaDv2_DU2_S0_(i8 %[[V1Coerce:.+]], i8 %[[V2Coerce:.+]])
// LIN64: %[[RetVal:.+]] = alloca <2 x i2>, align 1
// LIN64: %[[V1Addr:.+]] = alloca <2 x i2>, align 1
// LIN64: %[[V2Addr:.+]] = alloca <2 x i2>, align 1
// LIN64: %[[V1Val:.+]] = load <2 x i2>, ptr %[[V1Addr]], align 1
// LIN64: %[[V2Val:.+]] = load <2 x i2>, ptr %[[V2Addr]], align 1
// LIN64: %[[AddVal:.+]] = add <2 x i2> %0, %1
// LIN64: store <2 x i2> %[[AddVal]], ptr %[[RetVal]], align 1
// LIN64: %[[Ret:.+]] = load i8, ptr %[[RetVal]], align 1
// LIN64: ret i8 %[[Ret]]

// LIN32: define dso_local <2 x i2> @_Z25TestBitIntVector2x2AllocaDv2_DU2_S0_(<2 x i2> %{{.+}}, <2 x i2> %{{.+}})
// LIN32: %[[V1Addr:.+]] = alloca <2 x i2>, align 2
// LIN32: %[[V2Addr:.+]] = alloca <2 x i2>, align 2
// LIN32: %[[V1Addr:.+]] = alloca <2 x i2>, align 1
// LIN32: %[[V2Addr:.+]] = alloca <2 x i2>, align 1
// LIN32: ret <2 x i2> %[[Ret:.+]]

// WIN: define dso_local <2 x i2> @"?TestBitIntVector2x2Alloca@@YAT?$__vector@U?$_UBitInt@$01@__clang@@$01@__clang@@T12@0@Z"(<2 x i2>{{.*}}, <2 x i2>{{.*}})
// WIN: %[[V1:.+]] = alloca <2 x i2>, align 2
// WIN: %[[V2:.+]] = alloca <2 x i2>, align 2
// WIN: %[[V1:.+]] = alloca <2 x i2>, align 1
// WIN: %[[V2:.+]] = alloca <2 x i2>, align 1
// WIN: ret <2 x i2> %[[Ret:.+]]
return v1 + v2;
}
Expand Down
28 changes: 13 additions & 15 deletions clang/test/CodeGenCXX/matrix-vector-bit-int.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,27 +70,25 @@ i512x3 v3(i512x3 a) {
return a + a;
}

// CHECK-LABEL: define dso_local i32 @_Z2v4Dv3_DB4_(
// CHECK-SAME: i32 [[A_COERCE:%.*]]) #[[ATTR0]] {
// CHECK-LABEL: define dso_local i16 @_Z2v4Dv3_DB4_(
// CHECK-SAME: i16 [[A_COERCE:%.*]]) #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <3 x i4>, align 4
// CHECK-NEXT: [[A:%.*]] = alloca <3 x i4>, align 4
// CHECK-NEXT: [[A_ADDR:%.*]] = alloca <3 x i4>, align 4
// CHECK-NEXT: [[RETVAL_COERCE:%.*]] = alloca i32, align 4
// CHECK-NEXT: store i32 [[A_COERCE]], ptr [[A]], align 4
// CHECK-NEXT: [[LOADVECN:%.*]] = load <4 x i4>, ptr [[A]], align 4
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <3 x i4>, align 2
// CHECK-NEXT: [[A:%.*]] = alloca <3 x i4>, align 2
// CHECK-NEXT: [[A_ADDR:%.*]] = alloca <3 x i4>, align 2
// CHECK-NEXT: store i16 [[A_COERCE]], ptr [[A]], align 2
// CHECK-NEXT: [[LOADVECN:%.*]] = load <4 x i4>, ptr [[A]], align 2
// CHECK-NEXT: [[A1:%.*]] = shufflevector <4 x i4> [[LOADVECN]], <4 x i4> poison, <3 x i32> <i32 0, i32 1, i32 2>
// CHECK-NEXT: [[EXTRACTVEC:%.*]] = shufflevector <3 x i4> [[A1]], <3 x i4> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison>
// CHECK-NEXT: store <4 x i4> [[EXTRACTVEC]], ptr [[A_ADDR]], align 4
// CHECK-NEXT: [[LOADVECN2:%.*]] = load <4 x i4>, ptr [[A_ADDR]], align 4
// CHECK-NEXT: store <4 x i4> [[EXTRACTVEC]], ptr [[A_ADDR]], align 2
// CHECK-NEXT: [[LOADVECN2:%.*]] = load <4 x i4>, ptr [[A_ADDR]], align 2
// CHECK-NEXT: [[EXTRACTVEC3:%.*]] = shufflevector <4 x i4> [[LOADVECN2]], <4 x i4> poison, <3 x i32> <i32 0, i32 1, i32 2>
// CHECK-NEXT: [[LOADVECN4:%.*]] = load <4 x i4>, ptr [[A_ADDR]], align 4
// CHECK-NEXT: [[LOADVECN4:%.*]] = load <4 x i4>, ptr [[A_ADDR]], align 2
// CHECK-NEXT: [[EXTRACTVEC5:%.*]] = shufflevector <4 x i4> [[LOADVECN4]], <4 x i4> poison, <3 x i32> <i32 0, i32 1, i32 2>
// CHECK-NEXT: [[ADD:%.*]] = add <3 x i4> [[EXTRACTVEC3]], [[EXTRACTVEC5]]
// CHECK-NEXT: store <3 x i4> [[ADD]], ptr [[RETVAL]], align 4
// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[RETVAL_COERCE]], ptr align 4 [[RETVAL]], i64 2, i1 false)
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[RETVAL_COERCE]], align 4
// CHECK-NEXT: ret i32 [[TMP0]]
// CHECK-NEXT: store <3 x i4> [[ADD]], ptr [[RETVAL]], align 2
// CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[RETVAL]], align 2
// CHECK-NEXT: ret i16 [[TMP0]]
//
i4x3 v4(i4x3 a) {
return a + a;
Expand Down
22 changes: 22 additions & 0 deletions clang/test/SemaCXX/constexpr-builtin-bit-cast.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -510,6 +510,28 @@ constexpr bool9 bad_short_to_bool9 = __builtin_bit_cast(bool9, static_cast<unsig
// expected-note@+1 {{bit_cast involving type 'bool __attribute__((ext_vector_type(17)))' (vector of 17 'bool' values) is not allowed in a constant expression; element size 1 * element count 17 is not a multiple of the byte size 8}}
constexpr bool17 bad_int_to_bool17 = __builtin_bit_cast(bool17, 0x0001CAFEU);

template <int Bits, int N> using packed_vec_t = _BitInt(Bits) __attribute__((ext_vector_type(N)));

static_assert(round_trip<packed_vec_t<2, 4>>(static_cast<unsigned char>(0)), "");
static_assert(round_trip<packed_vec_t<2, 4>>(static_cast<unsigned char>(1)), "");
static_assert(round_trip<packed_vec_t<2, 4>>(static_cast<unsigned char>(0x55)), "");
static_assert(round_trip<packed_vec_t<2, 8>>(static_cast<short>(0)), "");
static_assert(round_trip<packed_vec_t<2, 8>>(static_cast<short>(-1)), "");
static_assert(round_trip<packed_vec_t<2, 8>>(static_cast<short>(0x5555)), "");

static_assert(bit_cast<unsigned char>(packed_vec_t<2, 4>{1, -2, 0, -1}) == (LITTLE_END ? 0xC9 : 0x63), "");
static_assert(bit_cast<unsigned short>(packed_vec_t<2, 8>{1, -2, 0, -1, -2, -1, 1, 0}) == (LITTLE_END ? 0x1EC9 : 0x63B4), "");

static_assert(round_trip<packed_vec_t<4, 2>>(static_cast<unsigned char>(0)), "");
static_assert(round_trip<packed_vec_t<4, 2>>(static_cast<unsigned char>(1)), "");
static_assert(round_trip<packed_vec_t<4, 2>>(static_cast<unsigned char>(0x55)), "");
static_assert(round_trip<packed_vec_t<4, 4>>(static_cast<short>(0)), "");
static_assert(round_trip<packed_vec_t<4, 4>>(static_cast<short>(-1)), "");
static_assert(round_trip<packed_vec_t<4, 4>>(static_cast<short>(0x5555)), "");

static_assert(bit_cast<unsigned char>(packed_vec_t<4, 2>{-4, -7}) == (LITTLE_END ? 0x9C : 0xC9), "");
static_assert(bit_cast<unsigned short>(packed_vec_t<4, 4>{3, -5, -1, 7}) == (LITTLE_END ? 0x7FB3 : 0x3BF7), "");

}

namespace test_complex {
Expand Down
30 changes: 30 additions & 0 deletions clang/test/SemaCXX/ext-int.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -293,3 +293,33 @@ void FromPaper1() {
void FromPaper2(_BitInt(8) a1, _BitInt(24) a2) {
static_assert(is_same<decltype(a1 * (_BitInt(32))a2), _BitInt(32)>::value, "");
}

// Check sub-byte integer vector size and alignment, expecting packing.
template <int Bits, int N>
using packed_vec_t = _BitInt(Bits) __attribute__((ext_vector_type(N)));
void SubByteVecPacking() {
static_assert(sizeof(packed_vec_t<2, 2>) == 1);
static_assert(sizeof(packed_vec_t<2, 3>) == 1);
static_assert(sizeof(packed_vec_t<2, 4>) == 1);
static_assert(sizeof(packed_vec_t<2, 8>) == 2);
static_assert(sizeof(packed_vec_t<2, 16>) == 4);
static_assert(sizeof(packed_vec_t<2, 32>) == 8);
static_assert(sizeof(packed_vec_t<4, 2>) == 1);
static_assert(sizeof(packed_vec_t<4, 4>) == 2);
static_assert(sizeof(packed_vec_t<4, 8>) == 4);
static_assert(sizeof(packed_vec_t<4, 16>) == 8);
static_assert(sizeof(packed_vec_t<4, 32>) == 16);

static_assert(alignof(packed_vec_t<2, 2>) == 1);
static_assert(alignof(packed_vec_t<2, 3>) == 1);
static_assert(alignof(packed_vec_t<2, 4>) == 1);
static_assert(alignof(packed_vec_t<2, 8>) == 2);
static_assert(alignof(packed_vec_t<2, 16>) == 4);
static_assert(alignof(packed_vec_t<2, 32>) == 8);
static_assert(alignof(packed_vec_t<4, 2>) == 1);
static_assert(alignof(packed_vec_t<4, 3>) == 2);
static_assert(alignof(packed_vec_t<4, 4>) == 2);
static_assert(alignof(packed_vec_t<4, 8>) == 4);
static_assert(alignof(packed_vec_t<4, 16>) == 8);
static_assert(alignof(packed_vec_t<4, 32>) == 16);
}