Skip to content

Add the SIMD table generator script (#858) #858

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
terrelln wants to merge 2 commits into
base: dev
Choose a base branch
from
Open

Add the SIMD table generator script (#858) #858

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c5078bb
Remove partitionLeft / mergeVectorConstant (#900)
terrelln Jul 23, 2026
5c0d165
Add the SIMD table generator script (#858)
terrelln Jul 23, 2026
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
105 changes: 0 additions & 105 deletions benchmark/micro/micro_pivco_huffman.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -216,30 +216,6 @@ void verifyPartitionFull(const EncodeArch& arch, const PartitionData& data)
requireEqualPrefix(rhs, expectedRhs, expectedOnes);
}

void verifyPartitionLeft(const EncodeArch& arch, const PartitionData& data)
{
std::vector<uint8_t> expectedBitmap(data.bitmap.size());
std::vector<uint8_t> expectedLhs(data.lhs.size());
size_t const expectedOnes = ZL_PivCoHuffmanEncode_generic.partitionLeft(
expectedBitmap.data(),
expectedLhs.data(),
data.ranks.data(),
data.size,
data.rightRank);

std::vector<uint8_t> bitmap(data.bitmap.size());
std::vector<uint8_t> lhs(data.lhs.size());
size_t const ones = arch.kernels->partitionLeft(
bitmap.data(),
lhs.data(),
data.ranks.data(),
data.size,
data.rightRank);
ZL_REQUIRE_EQ(ones, expectedOnes);
requireEqualPrefix(bitmap, expectedBitmap, bitmapBytes(data.size));
requireEqualPrefix(lhs, expectedLhs, data.size - expectedOnes);
}

void verifyPartitionRight(const EncodeArch& arch, const PartitionData& data)
{
std::vector<uint8_t> expectedBitmap(data.bitmap.size());
Expand Down Expand Up @@ -356,35 +332,6 @@ void verifyMergeConstantVector(const DecodeArch& arch, const MergeData& data)
requireEqualPrefix(out, expectedOut, data.totalSize);
}

void verifyMergeVectorConstant(const DecodeArch& arch, const MergeData& data)
{
std::vector<uint8_t> expectedOut(data.out.size());
uint8_t const rhs = 0xC3;
size_t const expectedOnes =
ZL_PivCoHuffmanDecode_generic.mergeVectorConstant(
expectedOut.data(),
expectedOut.size(),
data.bitmap.data(),
data.bitmap.size(),
data.lhs.data(),
data.lhsSize,
rhs,
data.rhsSize);

std::vector<uint8_t> out(data.out.size());
size_t const ones = arch.kernels->mergeVectorConstant(
out.data(),
out.size(),
data.bitmap.data(),
data.bitmap.size(),
data.lhs.data(),
data.lhsSize,
rhs,
data.rhsSize);
ZL_REQUIRE_EQ(ones, expectedOnes);
requireEqualPrefix(out, expectedOut, data.totalSize);
}

void verifyMergeFlatDepth(const DecodeArch& arch, const FlatData& data)
{
std::vector<uint8_t> expectedOut(data.out.size());
Expand Down Expand Up @@ -437,28 +384,6 @@ void registerPartitionFullBenchmark(EncodeArch arch)
});
}

void registerPartitionLeftBenchmark(EncodeArch arch)
{
auto data = std::make_shared<PartitionData>(64 * 1024);
verifyPartitionLeft(arch, *data);
RegisterBenchmark(
benchName(arch.name, "partitionLeft"),
[arch, data](benchmark::State& state) {
size_t ones = 0;
for (auto _ : state) {
ones += arch.kernels->partitionLeft(
data->bitmap.data(),
data->lhs.data(),
data->ranks.data(),
data->size,
data->rightRank);
}
benchmark::DoNotOptimize(ones);
state.SetBytesProcessed(
(int64_t)data->size * (int64_t)state.iterations());
});
}

void registerPartitionRightBenchmark(EncodeArch arch)
{
auto data = std::make_shared<PartitionData>(64 * 1024);
Expand Down Expand Up @@ -574,32 +499,6 @@ void registerMergeConstantVectorBenchmark(DecodeArch arch)
});
}

void registerMergeVectorConstantBenchmark(DecodeArch arch)
{
auto data = std::make_shared<MergeData>(64 * 1024);
verifyMergeVectorConstant(arch, *data);
RegisterBenchmark(
benchName(arch.name, "mergeVectorConstant"),
[arch, data](benchmark::State& state) {
size_t ones = 0;
uint8_t const rhs = 0xC3;
for (auto _ : state) {
ones += arch.kernels->mergeVectorConstant(
data->out.data(),
data->out.size(),
data->bitmap.data(),
data->bitmap.size(),
data->lhs.data(),
data->lhsSize,
rhs,
data->rhsSize);
}
benchmark::DoNotOptimize(ones);
state.SetBytesProcessed(
(int64_t)data->totalSize * (int64_t)state.iterations());
});
}

void registerMergeFlatDepthBenchmark(DecodeArch arch, size_t depth)
{
auto data = std::make_shared<FlatData>(64 * 1024, depth);
Expand Down Expand Up @@ -628,12 +527,10 @@ void registerMergeFlatDepthBenchmark(DecodeArch arch, size_t depth)
void registerEncodeBenchmarks(EncodeArch arch)
{
ZL_REQUIRE_NN(arch.kernels->partitionFull);
ZL_REQUIRE_NN(arch.kernels->partitionLeft);
ZL_REQUIRE_NN(arch.kernels->partitionRight);
ZL_REQUIRE_NN(arch.kernels->partitionNone);
ZL_REQUIRE_NN(arch.kernels->packFlatDepth);
registerPartitionFullBenchmark(arch);
registerPartitionLeftBenchmark(arch);
registerPartitionRightBenchmark(arch);
registerPartitionNoneBenchmark(arch);
for (size_t depth = 1; depth <= 8; ++depth) {
Expand All @@ -645,11 +542,9 @@ void registerDecodeBenchmarks(DecodeArch arch)
{
ZL_REQUIRE_NN(arch.kernels->mergeVectorVector);
ZL_REQUIRE_NN(arch.kernels->mergeConstantVector);
ZL_REQUIRE_NN(arch.kernels->mergeVectorConstant);
ZL_REQUIRE_NN(arch.kernels->mergeFlatDepth);
registerMergeVectorVectorBenchmark(arch);
registerMergeConstantVectorBenchmark(arch);
registerMergeVectorConstantBenchmark(arch);
for (size_t depth = 1; depth <= 8; ++depth) {
registerMergeFlatDepthBenchmark(arch, depth);
}
Expand Down
Loading
Loading