Skip to content

Commit 88ee511

Browse files
sergey-kozubcopybara-github
authored andcommitted
PR #19096: Add F4E2M1FN and F8E8M0FNU types
Imported from GitHub PR openxla/xla#19096 This PR adds F4E2M1FN primitive type (4-bit float with 2 bits exponent and 1 bit mantissa), F8E8M0FNU primitive type (8-bit float with 8 bits exponent, no mantissa and no sign) and enables loads/stores in the same way S4/U4 type is implemented. This will enable using microscaling (MX) formats ([RFC](openxla/xla#18085)), such as MXFP4. ```c F4E2M1FN - Exponent bias: 1 - Maximum stored exponent value: 3 (binary 11) - Maximum unbiased exponent value: 3 - 1 = 2 - Minimum stored exponent value: 1 (binary 01) - Minimum unbiased exponent value: 1 − 1 = 0 - Has Positive and Negative zero - Doesn't have infinity - Doesn't have NaNs Additional details: - Zeros (+/-): S.00.0 - Max normal number: S.11.1 = ±2^(2) x (1 + 0.5) = ±6.0 - Min normal number: S.01.0 = ±2^(0) = ±1.0 - Min subnormal number: S.00.1 = ±2^(0) x 0.5 = ±0.5 F8E8M0FNU - Exponent bias: 127 - Maximum stored exponent value: 254 (binary 1111'1110) - Maximum unbiased exponent value: 254 - 127 = 127 - Minimum stored exponent value: 0 (binary 0000'0000) - Minimum unbiased exponent value: 0 − 127 = -127 - Doesn't have zero - Doesn't have infinity - NaN is encoded as binary 1111'1111 Additional details: - Zeros cannot be represented - Negative values cannot be represented - Mantissa is always 1 ``` Related PRs: - openxla/stablehlo#2582 - jax-ml/ml_dtypes#181 - llvm/llvm-project#95392 - llvm/llvm-project#108877 - jax-ml/ml_dtypes#166 - llvm/llvm-project#107127 - llvm/llvm-project#111028 The PR is split into multiple commits just to make the review easier, it is possible that some tests could fail if only some (i.e. not all) of these commits are applied. Copybara import of the project: -- f493e4803eaa5ff3da3ceb130e9348c014b4a2e8 by Sergey Kozub <[email protected]>: Add F4E2M1FN type: import mxfloat.h -- 87d005630b310a355d7c30b22828c35237373f17 by Sergey Kozub <[email protected]>: Add F4E2M1FN type: primitive type -- 70ca82093faeec98f2dc5e8b82f617d99ca96849 by Sergey Kozub <[email protected]>: Add F4E2M1FN type: literal support -- c479f0940da490e9668e2f48e14a7466f0c4a97f by Sergey Kozub <[email protected]>: Add F4E2M1FN type: conversion codegen -- daaa3af3ce3af456f2ef44dbc291ebeb09e86d9b by Sergey Kozub <[email protected]>: Add F4E2M1FN type: python interface -- 1f0e19ff14733eff790726936b68ef0cf607a766 by Sergey Kozub <[email protected]>: Add F4E2M1FN type: FFI -- 999bf96092e57c7b3039811f2887281f347ff17a by Sergey Kozub <[email protected]>: Add F4E2M1FN type: HLO evaluator -- d7d5af74c5f8a94522779a121c0a4a962156fb64 by Sergey Kozub <[email protected]>: Add F4E2M1FN type: add tests -- 9e8c7bc02849f241d0f05941221d99f1d08d9e67 by Sergey Kozub <[email protected]>: Add F8E8M0FNU type -- 1e344174b931cea4978770ab740dfed67186c2f4 by Sergey Kozub <[email protected]>: Addressing PR#19096 review comments -- d4de0a369d9dc853f34f3cf3bf7dcc5a47502106 by Sergey Kozub <[email protected]>: Addressing PR#19096 review comments (round 2) Merging this change closes #19096 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#19096 from openxla:skozub/e2m1 d4de0a369d9dc853f34f3cf3bf7dcc5a47502106 PiperOrigin-RevId: 707638099
1 parent 6a6aae2 commit 88ee511

File tree

2 files changed

+4
-0
lines changed

2 files changed

+4
-0
lines changed

tsl/platform/BUILD

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -985,6 +985,7 @@ cc_library(
985985
deps = [
986986
"@ml_dtypes//:float8",
987987
"@ml_dtypes//:intn",
988+
"@ml_dtypes//:mxfloat",
988989
],
989990
)
990991

tsl/platform/ml_dtypes.h

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -18,15 +18,18 @@ limitations under the License.
1818

1919
#include "ml_dtypes/include/float8.h" // from @ml_dtypes
2020
#include "ml_dtypes/include/intn.h" // from @ml_dtypes
21+
#include "ml_dtypes/include/mxfloat.h" // from @ml_dtypes
2122

2223
namespace tsl {
24+
using float4_e2m1fn = ::ml_dtypes::float4_e2m1fn;
2325
using float8_e3m4 = ::ml_dtypes::float8_e3m4;
2426
using float8_e4m3 = ::ml_dtypes::float8_e4m3;
2527
using float8_e4m3fn = ::ml_dtypes::float8_e4m3fn;
2628
using float8_e4m3fnuz = ::ml_dtypes::float8_e4m3fnuz;
2729
using float8_e4m3b11fnuz = ::ml_dtypes::float8_e4m3b11fnuz;
2830
using float8_e5m2 = ::ml_dtypes::float8_e5m2;
2931
using float8_e5m2fnuz = ::ml_dtypes::float8_e5m2fnuz;
32+
using float8_e8m0fnu = ::ml_dtypes::float8_e8m0fnu;
3033

3134
using int1 = ::ml_dtypes::int1;
3235
using uint1 = ::ml_dtypes::uint1;

0 commit comments

Comments
 (0)