Skip to content

Add UInt16 support to Q and DQ #6891

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

Merged
facebook-github-bot merged 1 commit into from
Nov 15, 2024
Merged

Add UInt16 support to Q and DQ #6891

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 3 additions & 0 deletions kernels/quantized/cpu/op_dequantize.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,7 @@ void check_dequantize_per_tensor_args(
input.scalar_type() == ScalarType::Byte ||
input.scalar_type() == ScalarType::Char ||
input.scalar_type() == ScalarType::Bits16 ||
input.scalar_type() == ScalarType::UInt16 ||
input.scalar_type() == ScalarType::Short ||
input.scalar_type() == ScalarType::Int,
"input.scalar_type() %" PRId8 " is not supported:",
Expand Down Expand Up @@ -120,6 +121,7 @@ Tensor& dequantize_per_tensor_out(
switch (input.scalar_type()) {
ET_FORALL_INT_TYPES(CALCULATE_INT_TYPE);
CALCULATE_INT_TYPE(uint16_t, Bits16);
CALCULATE_INT_TYPE(uint16_t, UInt16);
default:
ET_CHECK_MSG(
false,
Expand Down Expand Up @@ -315,6 +317,7 @@ Tensor& dequantize_per_channel_out(
switch (input.scalar_type()) {
ET_FORALL_INT_TYPES(CALCULATE_FLOAT_TYPE);
CALCULATE_INT_TYPE(uint16_t, Bits16);
CALCULATE_INT_TYPE(uint16_t, UInt16);
default:
ET_CHECK_MSG(
false,
Expand Down
4 changes: 3 additions & 1 deletion kernels/quantized/cpu/op_quantize.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ void check_quantize_per_tensor_args(
static_cast<int32_t>(std::numeric_limits<int8_t>::min());
quant_max_upper_bound =
static_cast<int32_t>(std::numeric_limits<int8_t>::max());
} else if (dtype == ScalarType::Bits16) {
} else if (dtype == ScalarType::Bits16 || dtype == ScalarType::UInt16) {
quant_min_lower_bound = std::numeric_limits<uint16_t>::min();
quant_max_upper_bound = std::numeric_limits<uint16_t>::max();
} else if (dtype == ScalarType::Short) {
Expand Down Expand Up @@ -139,6 +139,7 @@ Tensor& quantize_per_tensor_out(
switch (out.scalar_type()) { \
ET_FORALL_INT_TYPES_WITH(IN_CTYPE, QUANTIZE_IMPL); \
QUANTIZE_IMPL(IN_CTYPE, uint16_t, Bits16) \
QUANTIZE_IMPL(IN_CTYPE, uint16_t, UInt16) \
default: \
ET_CHECK_MSG( \
false, \
Expand Down Expand Up @@ -334,6 +335,7 @@ Tensor& quantize_per_channel_out(
switch (out.scalar_type()) { \
ET_FORALL_INT_TYPES_WITH(CTYPE_IN, QUANTIZE_IMPL); \
QUANTIZE_IMPL(CTYPE_IN, uint16_t, Bits16) \
QUANTIZE_IMPL(CTYPE_IN, uint16_t, UInt16) \
default: \
ET_CHECK_MSG( \
false, \
Expand Down
1 change: 1 addition & 0 deletions kernels/quantized/test/op_dequantize_test.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,6 +63,7 @@ TEST(OpDequantizeOutTest, AllDtypesSupported) {
test_dtype<ScalarType::Char>();
test_dtype<ScalarType::Short>();
test_dtype<ScalarType::Bits16>();
test_dtype<ScalarType::UInt16>();
test_dtype<ScalarType::Int>();
}

Expand Down
1 change: 1 addition & 0 deletions kernels/quantized/test/op_quantize_test.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,6 +54,7 @@ TEST(OpQuantizeOutTest, AllDtypesSupported) {
test_dtype<ScalarType::Char>();
test_dtype<ScalarType::Short>();
test_dtype<ScalarType::Bits16>();
test_dtype<ScalarType::UInt16>();
test_dtype<ScalarType::Int>();
}

Expand Down
7 changes: 7 additions & 0 deletions runtime/core/exec_aten/testing_util/tensor_factory.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -650,6 +650,13 @@ struct ScalarTypeToCppTypeWrapper<torch::executor::ScalarType::Bits16> {
using ctype = uint16_t;
};

// Use a C type of `uint16_t` instead of `UInt16` to simplify code reuse when
// testing multiple integer types.
template <>
struct ScalarTypeToCppTypeWrapper<torch::executor::ScalarType::UInt16> {
using ctype = uint16_t;
};

// To allow implicit conversion between simple types to `ctype`
#define SPECIALIZE_ScalarTypeToCppTypeWrapper(CTYPE, DTYPE) \
template <> \
Expand Down
Loading