Skip to content

Add quantized embedding kernels to torchao #1018

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
merged 1 commit into from
Oct 17, 2024
Merged

Add quantized embedding kernels to torchao #1018

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
12 changes: 6 additions & 6 deletions torchao/experimental/kernels/cpu/aarch64/bitpacking/bitpack.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,13 +9,13 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint1.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint2.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint3.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint4.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint5.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/uint6.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>
#include <cassert>

namespace torchao {
Expand Down Expand Up @@ -142,7 +142,7 @@ TORCHAO_ALWAYS_INLINE inline void vec_pack_32_lowbit_values(
break;
case 6:
torchao::bitpacking::internal::vec_pack_32_uint6_values(
packed, shifted0, shifted1);
packed, shifted0, shifted1);
break;
default:
assert(false);
Expand All @@ -153,7 +153,7 @@ template <int nbit>
TORCHAO_ALWAYS_INLINE inline void vec_unpack_32_lowbit_values(
int8x16_t& unpacked0,
int8x16_t& unpacked1,
uint8_t* packed) {
const uint8_t* packed) {
static_assert(nbit < 8);
static_assert(nbit >= 1);

Expand Down Expand Up @@ -217,7 +217,7 @@ TORCHAO_ALWAYS_INLINE inline void vec_unpack_32_lowbit_values(
break;
case 6:
torchao::bitpacking::internal::vec_unpack_32_uint6_values(
shifted0, shifted1, packed);
shifted0, shifted1, packed);
break;
default:
assert(false);
Expand Down Expand Up @@ -288,7 +288,7 @@ TORCHAO_ALWAYS_INLINE inline void vec_unpack_64_lowbit_values(
int8x16_t& unpacked1,
int8x16_t& unpacked2,
int8x16_t& unpacked3,
uint8_t* packed) {
const uint8_t* packed) {
static_assert(nbit < 8);
static_assert(nbit >= 1);

Expand Down Expand Up @@ -443,7 +443,7 @@ TORCHAO_ALWAYS_INLINE inline void vec_unpack_128_lowbit_values(
int8x16_t& unpacked5,
int8x16_t& unpacked6,
int8x16_t& unpacked7,
uint8_t* packed) {
const uint8_t* packed) {
static_assert(nbit < 8);
static_assert(nbit >= 1);

Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint1.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@

#if defined(__aarch64__) || defined(__ARM_NEON)
#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint1.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint2.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint4.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint3.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint3.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint4.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint4.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint5.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint5.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
2 changes: 1 addition & 1 deletion torchao/experimental/kernels/cpu/aarch64/bitpacking/uint6.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
#if defined(__aarch64__) || defined(__ARM_NEON)

#include <arm_neon.h>
#include <torchao/experimental/kernels/cpu/aarch64/bitpacking/macro.h>
#include <torchao/experimental/kernels/cpu/aarch64/macro.h>

// This file contains bitpacking and unpacking methods for uint5.
// These are not inteded to be used outside of bitpacking directory.
Expand Down
Loading
Loading