Skip to content

Commit

Permalink
c10::nullopt -> std::nullopt
Browse files Browse the repository at this point in the history 
Summary:
X-link: pytorch/torchrec#2515

X-link: pytorch/executorch#6461

X-link: pytorch/audio#3848

X-link: pytorch/ao#1151

X-link: facebookincubator/AITemplate#1032

Reviewed By: houseroad

Differential Revision: D64835967

fbshipit-source-id: 9f9f65335aaf5497680561027ef9314e4b36f8d9
  • Loading branch information
@r-barnes @facebook-github-bot
r-barnes authored and facebook-github-bot committed Oct 23, 2024
1 parent 2834998 commit 3375de1
Show file tree
Hide file tree
Showing 20 changed files with 59 additions and 59 deletions.
6 changes: 3 additions & 3 deletions fbgemm_gpu/codegen/training/backward/embedding_backward_dense_host_cpu.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -164,11 +164,11 @@ Tensor split_embedding_codegen_lookup_dense_function(
const std::optional<Tensor>& indice_weights,
const std::optional<Tensor>& feature_requires_grad,
int64_t /* output_dtype = static_cast<int64_t>(SparseType::FP32) */,
const std::optional<Tensor>& /* B_offsets = c10::nullopt */,
const std::optional<Tensor>& /* B_offsets = std::nullopt */,
const std::optional<
Tensor>& /* vbe_output_offsets_feature_rank = c10::nullopt */,
Tensor>& /* vbe_output_offsets_feature_rank = std::nullopt */,
const std::optional<
Tensor>& /* vbe_B_offsets_rank_per_feature = c10::nullopt */,
Tensor>& /* vbe_B_offsets_rank_per_feature = std::nullopt */,
c10::SymInt /* max_B = -1 */,
c10::SymInt /* max_B_feature_rank = -1 */,
c10::SymInt /* vbe_output_size = -1 */) {
Expand Down
12 changes: 6 additions & 6 deletions fbgemm_gpu/codegen/training/backward/embedding_backward_split_host_template.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1006,26 +1006,26 @@ Tensor {{ bwd_mdesc }}_embedding_codegen_lookup_{{ optimizer }}_function(
{{ args.split_function_args | join(", ") }},
{%- endif %}
const int64_t output_dtype = static_cast<int64_t>(SparseType::FP32),
const std::optional<Tensor>& B_offsets = c10::nullopt,
const std::optional<Tensor>& vbe_output_offsets_feature_rank = c10::nullopt,
const std::optional<Tensor>& vbe_B_offsets_rank_per_feature = c10::nullopt,
const std::optional<Tensor>& B_offsets = std::nullopt,
const std::optional<Tensor>& vbe_output_offsets_feature_rank = std::nullopt,
const std::optional<Tensor>& vbe_B_offsets_rank_per_feature = std::nullopt,
const c10::SymInt max_B = -1,
const c10::SymInt max_B_feature_rank = -1,
{%- if not dense %}
const c10::SymInt vbe_output_size = -1,
const bool is_experimental_tbe = false, // formerly named is_experimental
const bool use_uniq_cache_locations_bwd = false,
const bool use_homogeneous_placements = false,
const std::optional<Tensor>& uvm_cache_stats = c10::nullopt,
const std::optional<Tensor>& uvm_cache_stats = std::nullopt,
{%- if "prev_iter_dev" not in args.split_function_arg_names %}
const std::optional<Tensor>& prev_iter_dev = c10::nullopt,
const std::optional<Tensor>& prev_iter_dev = std::nullopt,
{%- endif %}
{%- if "iter" not in args.split_function_arg_names %}
const int64_t iter = 0,
{%- endif %}
const bool apply_global_weight_decay = false,
{%- if ssd %}
const std::optional<at::TensorList>& ssd_tensors = c10::nullopt,
const std::optional<at::TensorList>& ssd_tensors = std::nullopt,
{%- endif %}
const double gwd_lower_bound = 0
{%- else %}
Expand Down
12 changes: 6 additions & 6 deletions fbgemm_gpu/codegen/training/pt2/embedding_split_host_pt2_autograd_template.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -987,25 +987,25 @@ Tensor {{ bwd_mdesc }}_embedding_codegen_lookup_{{ optimizer }}_function_pt2(
{%- endif %}
{{ args_pt2.unified_pt2.split_function_args | join(", ") }},
const int64_t output_dtype = static_cast<int64_t>(SparseType::FP32),
const std::optional<Tensor>& B_offsets = c10::nullopt,
const std::optional<Tensor>& vbe_output_offsets_feature_rank = c10::nullopt,
const std::optional<Tensor>& vbe_B_offsets_rank_per_feature = c10::nullopt,
const std::optional<Tensor>& B_offsets = std::nullopt,
const std::optional<Tensor>& vbe_output_offsets_feature_rank = std::nullopt,
const std::optional<Tensor>& vbe_B_offsets_rank_per_feature = std::nullopt,
const c10::SymInt max_B = -1,
const c10::SymInt max_B_feature_rank = -1,
const c10::SymInt vbe_output_size = -1,
const bool is_experimental_tbe = false, // formerly named is_experimental
const bool use_uniq_cache_locations_bwd = false,
const bool use_homogeneous_placements = false,
const std::optional<Tensor>& uvm_cache_stats = c10::nullopt,
const std::optional<Tensor>& uvm_cache_stats = std::nullopt,
{%- if "prev_iter_dev" not in args_pt2.split_function_arg_names %}
const std::optional<Tensor>& prev_iter_dev = c10::nullopt,
const std::optional<Tensor>& prev_iter_dev = std::nullopt,
{%- endif %}
{%- if "iter" not in args_pt2.split_function_arg_names %}
const int64_t iter = 0,
{%- endif %}
const bool apply_global_weight_decay = false,
{%- if ssd %}
const std::optional<at::TensorList>& ssd_tensors = c10::nullopt,
const std::optional<at::TensorList>& ssd_tensors = std::nullopt,
{%- endif %}
const double gwd_lower_bound = 0
) {
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/experimental/gen_ai/src/gemm/ck_extensions.hip
View file
Original file line number Diff line number Diff line change
Expand Up @@ -211,7 +211,7 @@ at::Tensor dispatch_bf16_gemm(at::Tensor A, at::Tensor B, std::optional<at::Tens
}
}

at::Tensor bf16_gemm(at::Tensor A, at::Tensor B, std::optional<at::Tensor> bias = c10::nullopt) {
at::Tensor bf16_gemm(at::Tensor A, at::Tensor B, std::optional<at::Tensor> bias = std::nullopt) {
TORCH_CHECK(
A.dtype() == at::kBFloat16 && B.dtype() == at::kBFloat16,
"Inputs must be bfloat16.");
Expand Down
4 changes: 2 additions & 2 deletions fbgemm_gpu/experimental/gen_ai/src/gemm/gemm.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ namespace fbgemm_gpu {
at::Tensor bf16_gemm(
at::Tensor A,
at::Tensor B,
std::optional<at::Tensor> bias = c10::nullopt);
std::optional<at::Tensor> bias = std::nullopt);

TORCH_LIBRARY_FRAGMENT(fbgemm, m) {
#ifdef USE_ROCM
Expand All @@ -31,7 +31,7 @@ TORCH_LIBRARY_IMPL(fbgemm, CUDA, m) {
at::Tensor bf16_gemm_meta(
at::Tensor A,
at::Tensor B,
std::optional<at::Tensor> /* bias */ = c10::nullopt) {
std::optional<at::Tensor> /* bias */ = std::nullopt) {
const at::SymInt M = A.sym_size(0);
const at::SymInt N = B.sym_size(0);
auto C = at::empty_symint({M, N}, A.options().dtype(at::kBFloat16));
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/experimental/gen_ai/src/quantize/ck_extensions/fp8_rowwise_gemm.hip
View file
Original file line number Diff line number Diff line change
Expand Up @@ -257,7 +257,7 @@ at::Tensor f8f8bf16_rowwise(
at::Tensor w_scale,
std::optional<at::Tensor> bias,
bool use_fast_accum,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
// Check that input datatypes are valid.
TORCH_CHECK(
(XQ.dtype() == at::kFloat8_e4m3fnuz) &&
Expand Down
12 changes: 6 additions & 6 deletions fbgemm_gpu/experimental/gen_ai/src/quantize/cutlass_extensions/f8f8bf16_cublas.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,10 +20,10 @@ namespace fbgemm_gpu {
at::Tensor f8f8bf16_cublas(
at::Tensor A, // FP8
at::Tensor B, // FP8
std::optional<at::Tensor> Ainvs = c10::nullopt,
std::optional<at::Tensor> Binvs = c10::nullopt,
std::optional<at::Tensor> Ainvs = std::nullopt,
std::optional<at::Tensor> Binvs = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
auto m = A.size(0);
auto n = B.size(0);
auto k = A.size(1);
Expand Down Expand Up @@ -167,10 +167,10 @@ at::Tensor f8f8bf16_cublas(
at::Tensor f8f8bf16_cublas(
at::Tensor A, // FP8
at::Tensor B, // FP8
std::optional<at::Tensor> Ainvs = c10::nullopt,
std::optional<at::Tensor> Binvs = c10::nullopt,
std::optional<at::Tensor> Ainvs = std::nullopt,
std::optional<at::Tensor> Binvs = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
throw std::runtime_error(
"CUDA version is older than 12.0"); // requires CUDA>=12
}
Expand Down
8 changes: 4 additions & 4 deletions fbgemm_gpu/experimental/gen_ai/src/quantize/cutlass_extensions/f8f8bf16_rowwise.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -361,9 +361,9 @@ at::Tensor f8f8bf16_rowwise(
at::Tensor WQ, // FP8
at::Tensor x_scale, // FP32
at::Tensor w_scale, // FP32
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
// Check datatypes.
TORCH_CHECK(
x_scale.dtype() == at::kFloat && w_scale.dtype() == at::kFloat,
Expand Down Expand Up @@ -482,9 +482,9 @@ at::Tensor f8f8bf16_rowwise(
at::Tensor WQ, // FP8
at::Tensor x_scale,
at::Tensor w_scale,
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
throw std::runtime_error(
"CUDA version is older than 12.0"); // requires CUDA>=12
}
Expand Down
8 changes: 4 additions & 4 deletions fbgemm_gpu/experimental/gen_ai/src/quantize/cutlass_extensions/f8f8bf16_rowwise_batched.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -599,9 +599,9 @@ at::Tensor f8f8bf16_rowwise_batched(
at::Tensor WQ, // FP8
at::Tensor x_scale, // FP32
at::Tensor w_scale, // FP32
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
// Check datatypes.
TORCH_CHECK(
x_scale.dtype() == at::kFloat && w_scale.dtype() == at::kFloat,
Expand Down Expand Up @@ -720,9 +720,9 @@ at::Tensor f8f8bf16_rowwise_batched(
at::Tensor WQ, // FP8
at::Tensor x_scale,
at::Tensor w_scale,
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt) {
std::optional<at::Tensor> output = std::nullopt) {
throw std::runtime_error(
"CUDA version is older than 12.0"); // requires CUDA>=12
}
Expand Down
14 changes: 7 additions & 7 deletions fbgemm_gpu/experimental/gen_ai/src/quantize/quantize.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -59,17 +59,17 @@ at::Tensor f8f8bf16_rowwise(
at::Tensor WQ,
at::Tensor x_scale,
at::Tensor w_scale,
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt);
std::optional<at::Tensor> output = std::nullopt);
at::Tensor f8f8bf16_rowwise_batched(
at::Tensor XQ,
at::Tensor WQ,
at::Tensor x_scale,
at::Tensor w_scale,
std::optional<at::Tensor> bias = c10::nullopt,
std::optional<at::Tensor> bias = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt);
std::optional<at::Tensor> output = std::nullopt);
at::Tensor f8f8bf16_blockwise(
at::Tensor XQ,
at::Tensor WQ,
Expand All @@ -81,10 +81,10 @@ at::Tensor f8f8bf16_blockwise(
at::Tensor f8f8bf16_cublas(
at::Tensor A,
at::Tensor B,
std::optional<at::Tensor> Ainvs = c10::nullopt,
std::optional<at::Tensor> Binvs = c10::nullopt,
std::optional<at::Tensor> Ainvs = std::nullopt,
std::optional<at::Tensor> Binvs = std::nullopt,
bool use_fast_accum = true,
std::optional<at::Tensor> output = c10::nullopt);
std::optional<at::Tensor> output = std::nullopt);
at::Tensor f8i4bf16_rowwise(
at::Tensor XQ,
at::Tensor WQ,
Expand Down
12 changes: 6 additions & 6 deletions fbgemm_gpu/include/fbgemm_gpu/embedding_inplace_update.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -54,8 +54,8 @@ void embedding_inplace_update_cuda(
Tensor update_row_idx,
Tensor update_offsets,
const int64_t row_alignment,
std::optional<Tensor> lxu_cache_weights = c10::nullopt,
std::optional<Tensor> lxu_cache_locations = c10::nullopt);
std::optional<Tensor> lxu_cache_weights = std::nullopt,
std::optional<Tensor> lxu_cache_locations = std::nullopt);

void embedding_inplace_update_single_placement_cuda(
Tensor& dev_weights,
Expand All @@ -69,8 +69,8 @@ void embedding_inplace_update_single_placement_cuda(
const Tensor& update_row_idx,
const Tensor& update_offsets,
const int64_t row_alignment,
std::optional<Tensor> lxu_cache_weights = c10::nullopt,
std::optional<Tensor> lxu_cache_locations = c10::nullopt);
std::optional<Tensor> lxu_cache_weights = std::nullopt,
std::optional<Tensor> lxu_cache_locations = std::nullopt);

void embedding_inplace_update_cpu(
Tensor dev_weights,
Expand All @@ -85,9 +85,9 @@ void embedding_inplace_update_cpu(
Tensor update_offsets,
const int64_t row_alignment,
std::optional<Tensor> lxu_cache_weights =
c10::nullopt, // Not used, to match cache interface for CUDA op
std::nullopt, // Not used, to match cache interface for CUDA op
std::optional<Tensor> lxu_cache_locations =
c10::nullopt // Not used, to match cache interface for CUDA op
std::nullopt // Not used, to match cache interface for CUDA op
);

/**
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/include/fbgemm_gpu/sparse_ops.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -987,7 +987,7 @@ std::vector<at::Tensor> jagged_index_select_2d(
const at::Tensor& values,
const at::Tensor& lengths,
const at::Tensor& indices,
const std::optional<int64_t> num_dense_output_rows = c10::nullopt);
const std::optional<int64_t> num_dense_output_rows = std::nullopt);

at::Tensor jagged_index_select_2d_forward_cpu(
const at::Tensor& values,
Expand Down
4 changes: 2 additions & 2 deletions fbgemm_gpu/src/quantize_ops/quantize_ops_cpu.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -44,7 +44,7 @@ Tensor& _float_to_fused8bitrowwise_cpu_out_t(

auto output_dims = input_sizes.vec();
output_dims[last_dim] = output_columns;
at::native::resize_(output, output_dims, c10::nullopt);
at::native::resize_(output, output_dims, std::nullopt);

const auto input_data = static_cast<input_t*>(
input.data_ptr()); // input.data_ptr<input_t>(); -> Yields
Expand Down Expand Up @@ -73,7 +73,7 @@ Tensor& _fused8bitrowwise_to_float_cpu_out_t(

auto output_dims = input_sizes.vec();
output_dims[last_dim] = output_columns;
at::native::resize_(output, output_dims, c10::nullopt);
at::native::resize_(output, output_dims, std::nullopt);

auto output_data = static_cast<output_t*>(
output.data_ptr()); // output.data_ptr<output_t>(); -> Yields
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/src/sparse_ops/common.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ inline Tensor native_empty_like(const Tensor& self) {
self.options().layout_opt(),
self.options().device_opt(),
self.options().pinned_memory_opt(),
c10::nullopt);
std::nullopt);
}

} // namespace
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/src/sparse_ops/sparse_async_cumsum.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -39,7 +39,7 @@ void asynchronous_exclusive_cumsum_cpu_out(Tensor& t_out, const Tensor& t_in) {
TENSOR_ON_CPU(t_out);

const auto t_in_contig = t_in.expect_contiguous();
at::native::resize_(t_out, t_in_contig->sizes(), c10::nullopt);
at::native::resize_(t_out, t_in_contig->sizes(), std::nullopt);

FBGEMM_DISPATCH_ALL_TYPES(
t_in_contig->scalar_type(),
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/src/sparse_ops/sparse_ops_cpu.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2354,7 +2354,7 @@ Tensor& lengths_range_out(
});
}

at::native::resize_(output, {output_size}, c10::nullopt);
at::native::resize_(output, {output_size}, std::nullopt);

AT_DISPATCH_INDEX_TYPES(
t_in_contig->scalar_type(), "lengths_range_compute", [&]() {
Expand Down
4 changes: 2 additions & 2 deletions fbgemm_gpu/src/sparse_ops/sparse_permute_1d.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -87,8 +87,8 @@ permute_1D_sparse_data_cuda(
return {
lengths.view({-1}).clone(),
indices.clone(),
weights.has_value() ? c10::make_optional(weights->clone())
: c10::nullopt};
weights.has_value() ? std::make_optional(weights->clone())
: std::nullopt};
}

Tensor permuted_lengths;
Expand Down
4 changes: 2 additions & 2 deletions fbgemm_gpu/src/sparse_ops/sparse_permute_2d.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -93,8 +93,8 @@ permute_2D_sparse_data_cuda(
return {
lengths.clone(),
indices.clone(),
weights.has_value() ? c10::make_optional(weights->clone())
: c10::nullopt,
weights.has_value() ? std::make_optional(weights->clone())
: std::nullopt,
};
}

Expand Down
4 changes: 2 additions & 2 deletions fbgemm_gpu/src/split_embeddings_cache/linearize_cache_indices.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -217,8 +217,8 @@ get_unique_indices_cuda_impl(
auto unique_indices = at::empty_like(linear_indices);
auto unique_indices_length =
at::empty({1}, linear_indices.options().dtype(at::kInt));
std::optional<Tensor> unique_indices_count = c10::nullopt;
std::optional<Tensor> linear_index_positions_sorted = c10::nullopt;
std::optional<Tensor> unique_indices_count = std::nullopt;
std::optional<Tensor> linear_index_positions_sorted = std::nullopt;
Tensor linear_index_positions;
if (compute_inverse_indices) {
Expand Down
2 changes: 1 addition & 1 deletion fbgemm_gpu/src/split_embeddings_cache/lru_cache_find.cu
View file
Original file line number Diff line number Diff line change
Expand Up @@ -186,7 +186,7 @@ lru_cache_find_uncached_cuda(
auto cache_set_sorted_unique_indices = empty_like(unique_indices);
Tensor cache_sets_positions;
std::optional<Tensor> cache_set_inverse_indices = c10::nullopt;
std::optional<Tensor> cache_set_inverse_indices = std::nullopt;
if (compute_inverse_indices) {
TORCH_CHECK(
cache_sets.numel() <=
Expand Down

0 comments on commit 3375de1

Please sign in to comment.