biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[MISC] Replace c10::optional with std::optional (#11730)

Browse Source 
Signed-off-by: Lu Fang <lufang@fb.com>
This commit is contained in:
Lu Fang
2025-01-04 17:20:34 -08:00
committed by GitHub
parent 47831430cc
commit 4068f4b5b5
24 changed files with 136 additions and 136 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
csrc/quantization/compressed_tensors/int8_quant_kernels.cu
View File

@@ -226,7 +226,7 @@ __global__ void dynamic_scaled_int8_azp_quant_kernel(
void static_scaled_int8_quant(torch::Tensor& out, // [..., hidden_size]
torch::Tensor const& input, // [..., hidden_size]
torch::Tensor const& scale,
c10::optional<torch::Tensor> const& azp) {
std::optional<torch::Tensor> const& azp) {
TORCH_CHECK(input.is_contiguous());
TORCH_CHECK(out.is_contiguous());
TORCH_CHECK(scale.numel() == 1);
@@ -257,7 +257,7 @@ void static_scaled_int8_quant(torch::Tensor& out, // [..., hidden_size]
void dynamic_scaled_int8_quant(
torch::Tensor& out, // [..., hidden_size]
torch::Tensor const& input, // [..., hidden_size]
torch::Tensor& scales, c10::optional<torch::Tensor> const& azp) {
torch::Tensor& scales, std::optional<torch::Tensor> const& azp) {
TORCH_CHECK(input.is_contiguous());
TORCH_CHECK(out.is_contiguous());
TORCH_CHECK(scales.is_contiguous());

18
csrc/quantization/cutlass_w8a8/scaled_mm_c2x.cu
View File

@@ -39,7 +39,7 @@ void cutlass_scaled_mm_sm75(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
if (bias) {
@@ -58,8 +58,8 @@ void cutlass_scaled_mm_azp_sm75(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
@@ -94,7 +94,7 @@ void cutlass_scaled_mm_sm80(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
if (bias) {
@@ -113,8 +113,8 @@ void cutlass_scaled_mm_azp_sm80(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
@@ -165,7 +165,7 @@ void cutlass_scaled_mm_sm89(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
if (bias) {
@@ -184,8 +184,8 @@ void cutlass_scaled_mm_azp_sm89(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);

6
csrc/quantization/cutlass_w8a8/scaled_mm_c3x.cu
View File

@@ -51,7 +51,7 @@ void cutlass_scaled_mm_sm90(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);
if (bias) {
@@ -70,8 +70,8 @@ void cutlass_scaled_mm_azp_sm90(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias) {
TORCH_CHECK(a_scales.dtype() == torch::kFloat32);
TORCH_CHECK(b_scales.dtype() == torch::kFloat32);

30
csrc/quantization/cutlass_w8a8/scaled_mm_entry.cu
View File

@@ -9,26 +9,26 @@ void cutlass_scaled_mm_sm75(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_sm80(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_sm89(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& bias);
#if defined ENABLE_SCALED_MM_C3X && ENABLE_SCALED_MM_C3X
void cutlass_scaled_mm_sm90(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& bias);
#endif
void cutlass_scaled_mm_azp_sm75(torch::Tensor& c, torch::Tensor const& a,
@@ -36,24 +36,24 @@ void cutlass_scaled_mm_azp_sm75(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_azp_sm80(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias);
void cutlass_scaled_mm_azp_sm89(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias);
#if defined CUDA_VERSION && CUDA_VERSION >= 12000
void cutlass_scaled_mm_azp_sm90(torch::Tensor& c, torch::Tensor const& a,
@@ -61,8 +61,8 @@ void cutlass_scaled_mm_azp_sm90(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias);
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias);
#endif
bool cutlass_scaled_mm_supports_fp8(int64_t cuda_device_capability) {
@@ -84,7 +84,7 @@ bool cutlass_scaled_mm_supports_fp8(int64_t cuda_device_capability) {
void cutlass_scaled_mm(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& b, torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& bias) {
// Checks for conformality
TORCH_CHECK(a.dim() == 2 && b.dim() == 2 && c.dim() == 2);
TORCH_CHECK(c.size(0) == a.size(0) && a.size(1) == b.size(0) &&
@@ -148,8 +148,8 @@ void cutlass_scaled_mm_azp(torch::Tensor& c, torch::Tensor const& a,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
torch::Tensor const& azp_adj,
c10::optional<torch::Tensor> const& azp,
c10::optional<torch::Tensor> const& bias) {
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias) {
// Checks for conformality
TORCH_CHECK(a.dim() == 2 && b.dim() == 2 && c.dim() == 2);
TORCH_CHECK(c.size(0) == a.size(0) && a.size(1) == b.size(0) &&

2
csrc/quantization/machete/generate.py
View File

@@ -63,7 +63,7 @@ torch::Tensor mm_dispatch_{{type_sig}}(MMArgs args) {
static inline std::optional<at::ScalarType> maybe_scalartype(
c10::optional<at::Tensor> const& t) {
std::optional<at::Tensor> const& t) {
if (!t) {
return std::nullopt;
} else {

10
csrc/quantization/machete/machete_mm_kernel.cuh
View File

@@ -183,11 +183,11 @@ struct MacheteKernelTemplate {
torch::Tensor const& A, // MxK matrix
torch::Tensor const& B, // KxN prepacked matrix
torch::Tensor& D, // MxN matrix
c10::optional<torch::Tensor> const& maybe_g_scales, // scale_KxN matrix
c10::optional<torch::Tensor> const& maybe_g_zeros, // scale_KxN matrix
c10::optional<int64_t> maybe_group_size,
c10::optional<torch::Tensor> const& maybe_ch_scales, // len N vector
c10::optional<torch::Tensor> const& maybe_tok_scales) // len M vector
std::optional<torch::Tensor> const& maybe_g_scales, // scale_KxN matrix
std::optional<torch::Tensor> const& maybe_g_zeros, // scale_KxN matrix
std::optional<int64_t> maybe_group_size,
std::optional<torch::Tensor> const& maybe_ch_scales, // len N vector
std::optional<torch::Tensor> const& maybe_tok_scales) // len M vector
{
static_assert(!with_group_zeropoints || with_group_scales);

24
csrc/quantization/machete/machete_mm_launcher.cuh
View File

@@ -13,23 +13,23 @@ struct MMArgs {
torch::Tensor const& A;
torch::Tensor const& B;
vllm::ScalarType const& b_type;
c10::optional<at::ScalarType> const& maybe_out_type;
c10::optional<torch::Tensor> const& maybe_group_scales;
c10::optional<torch::Tensor> const& maybe_group_zeros;
c10::optional<int64_t> maybe_group_size;
c10::optional<torch::Tensor> const& maybe_channel_scales;
c10::optional<torch::Tensor> const& maybe_token_scales;
c10::optional<std::string> maybe_schedule;
std::optional<at::ScalarType> const& maybe_out_type;
std::optional<torch::Tensor> const& maybe_group_scales;
std::optional<torch::Tensor> const& maybe_group_zeros;
std::optional<int64_t> maybe_group_size;
std::optional<torch::Tensor> const& maybe_channel_scales;
std::optional<torch::Tensor> const& maybe_token_scales;
std::optional<std::string> maybe_schedule;
};
struct SupportedSchedulesArgs {
at::ScalarType a_type;
vllm::ScalarType b_type;
c10::optional<at::ScalarType> maybe_group_scales_type;
c10::optional<at::ScalarType> maybe_group_zeros_type;
c10::optional<at::ScalarType> maybe_channel_scales_type;
c10::optional<at::ScalarType> maybe_token_scales_type;
c10::optional<at::ScalarType> maybe_out_type;
std::optional<at::ScalarType> maybe_group_scales_type;
std::optional<at::ScalarType> maybe_group_zeros_type;
std::optional<at::ScalarType> maybe_channel_scales_type;
std::optional<at::ScalarType> maybe_token_scales_type;
std::optional<at::ScalarType> maybe_out_type;
};
torch::Tensor mm_dispatch(MMArgs args);

2
csrc/quantization/machete/machete_prepack_launcher.cuh
View File

@@ -10,7 +10,7 @@ struct PrepackBArgs {
torch::Tensor const& B;
at::ScalarType a_type;
vllm::ScalarType b_type;
c10::optional<at::ScalarType> maybe_group_scales_type;
std::optional<at::ScalarType> maybe_group_scales_type;
};
template <typename PrepackedLayoutB>

26
csrc/quantization/machete/machete_pytorch.cu
View File

@@ -10,11 +10,11 @@ using namespace vllm;
std::vector<std::string> supported_schedules(
at::ScalarType a_type, int64_t b_type_id,
c10::optional<at::ScalarType> maybe_group_scales_type,
c10::optional<at::ScalarType> maybe_group_zeros_type,
c10::optional<at::ScalarType> maybe_channel_scales_type,
c10::optional<at::ScalarType> maybe_token_scales_type,
c10::optional<at::ScalarType> maybe_out_type) {
std::optional<at::ScalarType> maybe_group_scales_type,
std::optional<at::ScalarType> maybe_group_zeros_type,
std::optional<at::ScalarType> maybe_channel_scales_type,
std::optional<at::ScalarType> maybe_token_scales_type,
std::optional<at::ScalarType> maybe_out_type) {
ScalarType const b_type = ScalarType::from_id(b_type_id);
return supported_schedules_dispatch({
.a_type = a_type,
@@ -29,13 +29,13 @@ std::vector<std::string> supported_schedules(
torch::Tensor mm(torch::Tensor const& A, torch::Tensor const& B,
int64_t b_type_id,
c10::optional<at::ScalarType> const& maybe_out_type,
c10::optional<torch::Tensor> const& maybe_group_scales,
c10::optional<torch::Tensor> const& maybe_group_zeros,
c10::optional<int64_t> maybe_group_size,
c10::optional<torch::Tensor> const& maybe_channel_scales,
c10::optional<torch::Tensor> const& maybe_token_scales,
c10::optional<std::string> maybe_schedule) {
std::optional<at::ScalarType> const& maybe_out_type,
std::optional<torch::Tensor> const& maybe_group_scales,
std::optional<torch::Tensor> const& maybe_group_zeros,
std::optional<int64_t> maybe_group_size,
std::optional<torch::Tensor> const& maybe_channel_scales,
std::optional<torch::Tensor> const& maybe_token_scales,
std::optional<std::string> maybe_schedule) {
ScalarType const b_type = ScalarType::from_id(b_type_id);
return mm_dispatch({.A = A,
.B = B,
@@ -51,7 +51,7 @@ torch::Tensor mm(torch::Tensor const& A, torch::Tensor const& B,
torch::Tensor prepack_B(
torch::Tensor const& B, at::ScalarType const& a_type, int64_t b_type_id,
c10::optional<at::ScalarType> const& maybe_group_scales_type) {
std::optional<at::ScalarType> const& maybe_group_scales_type) {
ScalarType const b_type = ScalarType::from_id(b_type_id);
return prepack_B_dispatch(
{.B = B,