[libc][math][c23] Add fmaf16 C23 math function. #130757

harrisonGPU · 2025-03-11T11:49:37Z

Implementation of fmaf16 function for 16-bit inputs.

libc/test/src/math/CMakeLists.txt

llvmbot · 2025-03-12T02:04:53Z

@llvm/pr-subscribers-backend-amdgpu

@llvm/pr-subscribers-libc

Author: Harrison Hao (harrisonGPU)

Changes

Implementation of fmaf16 function for 16-bit inputs.

Full diff: https://github.com/llvm/llvm-project/pull/130757.diff

17 Files Affected:

(modified) libc/config/gpu/amdgpu/entrypoints.txt (+1)
(modified) libc/config/gpu/nvptx/entrypoints.txt (+1)
(modified) libc/config/linux/aarch64/entrypoints.txt (+1)
(modified) libc/config/linux/x86_64/entrypoints.txt (+1)
(modified) libc/docs/headers/math/index.rst (+1-1)
(modified) libc/include/math.yaml (+9)
(modified) libc/src/math/CMakeLists.txt (+1)
(added) libc/src/math/fmaf16.h (+21)
(modified) libc/src/math/generic/CMakeLists.txt (+10)
(added) libc/src/math/generic/fmaf16.cpp (+20)
(modified) libc/test/src/math/CMakeLists.txt (+15)
(added) libc/test/src/math/fmaf16_test.cpp (+13)
(modified) libc/test/src/math/smoke/CMakeLists.txt (+12)
(added) libc/test/src/math/smoke/fmaf16_test.cpp (+13)
(modified) libc/utils/MPFRWrapper/MPFRUtils.cpp (+5)
(modified) utils/bazel/llvm-project-overlay/libc/BUILD.bazel (+2)
(modified) utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel (+5)

diff --git a/libc/config/gpu/amdgpu/entrypoints.txt b/libc/config/gpu/amdgpu/entrypoints.txt
index 291d86b4dd587..a5d32c8eda39f 100644
--- a/libc/config/gpu/amdgpu/entrypoints.txt
+++ b/libc/config/gpu/amdgpu/entrypoints.txt
@@ -548,6 +548,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.fabsf16
     libc.src.math.fdimf16
     libc.src.math.floorf16
+    libc.src.math.fmaf16
     libc.src.math.fmaxf16
     libc.src.math.fmaximum_mag_numf16
     libc.src.math.fmaximum_magf16
diff --git a/libc/config/gpu/nvptx/entrypoints.txt b/libc/config/gpu/nvptx/entrypoints.txt
index 1ea0d9b03b37e..f553f7cc2b210 100644
--- a/libc/config/gpu/nvptx/entrypoints.txt
+++ b/libc/config/gpu/nvptx/entrypoints.txt
@@ -550,6 +550,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.fabsf16
     libc.src.math.fdimf16
     libc.src.math.floorf16
+    libc.src.math.fmaf16
     libc.src.math.fmaxf16
     libc.src.math.fmaximum_mag_numf16
     libc.src.math.fmaximum_magf16
diff --git a/libc/config/linux/aarch64/entrypoints.txt b/libc/config/linux/aarch64/entrypoints.txt
index c7beb3ef3fdfc..f32e03ff2524f 100644
--- a/libc/config/linux/aarch64/entrypoints.txt
+++ b/libc/config/linux/aarch64/entrypoints.txt
@@ -676,6 +676,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.ffma
     libc.src.math.ffmal
     libc.src.math.floorf16
+    libc.src.math.fmaf16
     libc.src.math.fmaxf16
     libc.src.math.fmaximum_mag_numf16
     libc.src.math.fmaximum_magf16
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 12dc87bf945fd..c73690b7864b4 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -686,6 +686,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.fabsf16
     libc.src.math.fdimf16
     libc.src.math.floorf16
+    libc.src.math.fmaf16
     libc.src.math.fmaxf16
     libc.src.math.fmaximum_mag_numf16
     libc.src.math.fmaximum_magf16
diff --git a/libc/docs/headers/math/index.rst b/libc/docs/headers/math/index.rst
index 5b855ce4881c3..dda8855df1a5a 100644
--- a/libc/docs/headers/math/index.rst
+++ b/libc/docs/headers/math/index.rst
@@ -299,7 +299,7 @@ Higher Math Functions
 +-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | expm1     | |check|          | |check|         |                        | |check|              |                        | 7.12.6.6               | F.10.3.6                   |
 +-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| fma       | |check|          | |check|         |                        |                      |                        | 7.12.13.1              | F.10.10.1                  |
+| fma       | |check|          | |check|         |                        | |check|              |                        | 7.12.13.1              | F.10.10.1                  |
 +-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | f16sqrt   | |check|\*        | |check|\*       | |check|\*              | N/A                  | |check|                | 7.12.14.6              | F.10.11                    |
 +-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
diff --git a/libc/include/math.yaml b/libc/include/math.yaml
index a66f981030864..08a28263a186d 100644
--- a/libc/include/math.yaml
+++ b/libc/include/math.yaml
@@ -736,6 +736,15 @@ functions:
       - type: float
       - type: float
       - type: float
+  - name: fmaf16
+    standards:
+      - stdc
+    return_type: _Float16
+    arguments:
+      - type: _Float16
+      - type: _Float16
+      - type: _Float16
+    guard: LIBC_TYPES_HAS_FLOAT16
   - name: fmax
     standards:
       - stdc
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index f18a73d46f9aa..a42421614661a 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -209,6 +209,7 @@ add_math_entrypoint_object(floorf128)
 
 add_math_entrypoint_object(fma)
 add_math_entrypoint_object(fmaf)
+add_math_entrypoint_object(fmaf16)
 
 add_math_entrypoint_object(fmax)
 add_math_entrypoint_object(fmaxf)
diff --git a/libc/src/math/fmaf16.h b/libc/src/math/fmaf16.h
new file mode 100644
index 0000000000000..1c4d468e67a7a
--- /dev/null
+++ b/libc/src/math/fmaf16.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for fmaf16 ------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_FMAF16_H
+#define LLVM_LIBC_SRC_MATH_FMAF16_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+float16 fmaf16(float16 x, float16 y, float16 z);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_FMAF16_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index 3114289bad486..781e6e7e04906 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -4241,6 +4241,16 @@ add_entrypoint_object(
     libc.src.__support.FPUtil.fma
 )
 
+add_entrypoint_object(
+  fmaf16
+  SRCS
+    fmaf16.cpp
+  HDRS
+    ../fmaf16.h
+  DEPENDS
+    libc.src.__support.FPUtil.fma
+)
+
 add_entrypoint_object(
   fma
   SRCS
diff --git a/libc/src/math/generic/fmaf16.cpp b/libc/src/math/generic/fmaf16.cpp
new file mode 100644
index 0000000000000..4f712f5de764f
--- /dev/null
+++ b/libc/src/math/generic/fmaf16.cpp
@@ -0,0 +1,20 @@
+//===-- Implementation of fmaf16 function ---------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/fmaf16.h"
+#include "src/__support/FPUtil/FMA.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(float16, fmaf16, (float16 x, float16 y, float16 z)) {
+  return fputil::fma<float16>(x, y, z);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/src/math/CMakeLists.txt
index 53ddd301900c0..44e499b6297f4 100644
--- a/libc/test/src/math/CMakeLists.txt
+++ b/libc/test/src/math/CMakeLists.txt
@@ -1776,6 +1776,21 @@ add_fp_unittest(
     FMA_OPT__ONLY
 )
 
+add_fp_unittest(
+  fmaf16_test
+  NEED_MPFR
+  SUITE
+    libc-math-unittests
+  SRCS
+    fmaf16_test.cpp
+  HDRS
+    FmaTest.h
+  DEPENDS
+    libc.src.math.fmaf16
+    libc.src.stdlib.rand
+    libc.src.stdlib.srand
+)
+
 add_fp_unittest(
   fma_test
   NEED_MPFR
diff --git a/libc/test/src/math/fmaf16_test.cpp b/libc/test/src/math/fmaf16_test.cpp
new file mode 100644
index 0000000000000..233d3a7c54cf4
--- /dev/null
+++ b/libc/test/src/math/fmaf16_test.cpp
@@ -0,0 +1,13 @@
+//===-- Unittests for fmaf16 ----------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "FmaTest.h"
+
+#include "src/math/fmaf16.h"
+
+LIST_FMA_TESTS(float16, LIBC_NAMESPACE::fmaf16)
diff --git a/libc/test/src/math/smoke/CMakeLists.txt b/libc/test/src/math/smoke/CMakeLists.txt
index 6f94440d826d9..e515633ad6720 100644
--- a/libc/test/src/math/smoke/CMakeLists.txt
+++ b/libc/test/src/math/smoke/CMakeLists.txt
@@ -3562,6 +3562,18 @@ add_fp_unittest(
     libc.src.__support.macros.properties.types
 )
 
+add_fp_unittest(
+  fmaf16_test
+  SUITE
+    libc-math-smoke-tests
+  SRCS
+    fmaf16_test.cpp
+  HDRS
+    FmaTest.h
+  DEPENDS
+    libc.src.math.fmaf16
+)
+
 add_fp_unittest(
   fma_test
   SUITE
diff --git a/libc/test/src/math/smoke/fmaf16_test.cpp b/libc/test/src/math/smoke/fmaf16_test.cpp
new file mode 100644
index 0000000000000..cc1f01e2f9541
--- /dev/null
+++ b/libc/test/src/math/smoke/fmaf16_test.cpp
@@ -0,0 +1,13 @@
+//===-- Unittests for fmaxf16 --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "FmaTest.h"
+
+#include "src/math/fmaf16.h"
+
+LIST_FMA_TESTS(float16, LIBC_NAMESPACE::fmaf16)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index 775a3d1a31964..f29fe43ad8c88 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -449,6 +449,8 @@ explain_ternary_operation_one_output_error(Operation,
                                            long double, double, RoundingMode);
 
 #ifdef LIBC_TYPES_HAS_FLOAT16
+template void explain_ternary_operation_one_output_error(
+    Operation, const TernaryInput<float16> &, float16, double, RoundingMode);
 template void explain_ternary_operation_one_output_error(
     Operation, const TernaryInput<float> &, float16, double, RoundingMode);
 template void explain_ternary_operation_one_output_error(
@@ -660,6 +662,9 @@ compare_ternary_operation_one_output(Operation,
                                      long double, double, RoundingMode);
 
 #ifdef LIBC_TYPES_HAS_FLOAT16
+template bool
+compare_ternary_operation_one_output(Operation, const TernaryInput<float16> &,
+                                     float16, double, RoundingMode);
 template bool compare_ternary_operation_one_output(Operation,
                                                    const TernaryInput<float> &,
                                                    float16, double,
diff --git a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
index 757db65ca8a77..849f4b728898a 100644
--- a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
@@ -2826,6 +2826,8 @@ libc_math_function(name = "floorf16")
 
 # TODO: Add fma, fmaf, fmal, fmaf128 functions.
 
+libc_math_function(name = "fmaf16")
+
 libc_math_function(name = "fmax")
 
 libc_math_function(name = "fmaxf")
diff --git a/utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel b/utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel
index 803548178b0e3..6b959fe559e45 100644
--- a/utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel
@@ -474,6 +474,11 @@ math_test(
 
 # TODO: Add fma, fmaf, fmal, fmaf128 tests.
 
+math_test(
+    name = "fmaf16",
+    hdrs = ["FMaxTest.h"],
+)
+
 math_test(
     name = "fmax",
     hdrs = ["FMaxTest.h"],

overmighty · 2025-03-12T18:01:00Z

Why did the bot add the backend:AMDGPU label?

libc/src/math/generic/CMakeLists.txt

libc/test/src/math/smoke/CMakeLists.txt

overmighty · 2025-03-12T18:31:24Z

utils/bazel/llvm-project-overlay/libc/BUILD.bazel

 # TODO: Add fma, fmaf, fmal, fmaf128 functions.

+libc_math_function(name = "fmaf16")
+


@lntue Why the TODO?

It was probably when I didn't set up the target dependency selection for fma.

@harrisonGPU you'll need to add extra dependency similar to ffma as above.

Okay, Thanks. :-)

@lntue , I have updated it, what do you think about it? :-)

I don't see changes to this file. Did you forget to push the update?

Sorry, I have updated it. What do you think about it? @lntue

overmighty · 2025-03-12T18:32:22Z

utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel

 # TODO: Add fma, fmaf, fmal, fmaf128 tests.

+math_test(
+    name = "fmaf16",
+    hdrs = ["FMaxTest.h"],
+)
+


Same here, why was there a TODO for the other fma* functions?

libc/test/src/math/smoke/fmaf16_test.cpp

utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel

harrisonGPU · 2025-03-13T01:42:14Z

Why did the bot add the backend:AMDGPU label?

Because I added entrypoints to amdgpu? Right?

overmighty · 2025-03-14T14:39:15Z

My understanding was that the backend:* labels were for LLVM code gen backends. I'm not sure if this label was intended to be applied to PRs that make changes to libc/config/gpu/amdgpu/entrypoints.txt.

harrisonGPU · 2025-03-14T14:53:45Z

My understanding was that the backend:* labels were for LLVM code gen backends. I'm not sure if this label was intended to be applied to PRs that make changes to libc/config/gpu/amdgpu/entrypoints.txt.

I am also not sure why not add this label. So what else can I do? :-)

overmighty · 2025-03-16T21:55:03Z

libc/test/src/math/smoke/CMakeLists.txt

+  DEPENDS
+    libc.src.math.fmaf16
+    libc.src.__support.macros.properties.types
+)


Suggested change

DEPENDS

libc.src.math.fmaf16

libc.src.__support.macros.properties.types

)

DEPENDS

libc.src.math.fmaf16

libc.src.__support.CPP.type_traits

libc.src.__support.FPUtil.cast

libc.src.__support.macros.properties.types

)

Or maybe we should fix this now that we have fputil::cast:

llvm-project/libc/test/src/math/smoke/FmaTest.h

Lines 93 to 98 in 3e6f618

#if defined(LIBC_TYPES_HAS_FLOAT16) && !defined(__LIBC_USE_FLOAT16_CONVERSION)

// Rounding modes other than the default might not be usable with float16.

if constexpr (LIBC_NAMESPACE::cpp::is_same_v<OutType, float16>)

EXPECT_FP_EQ(OutType(0.75) * z, func(InType(1.75), in_z, -in_z));

else

#endif

Then FmaTest.h would only need libc.src.__support.FPUtil.cast.

Hi @overmighty , I have updated it, please take a look. :-)

arsenm · 2025-03-17T03:46:26Z

My understanding was that the backend:* labels were for LLVM code gen backends.

No, not really. It's fuzzier than that, relevant to the target can spread to many more components and are not localized to the codegen. The auto-labeling rules we have though are too aggressive (in particular many of the test directories are mostly noise)

github-actions · 2025-03-19T10:14:19Z

✅ With the latest revision this PR passed the C/C++ code formatter.

lntue · 2025-03-21T15:50:03Z

Can you investigate the precommit failures?

harrisonGPU · 2025-03-21T16:02:58Z

Can you investigate the precommit failures?

Okay.

harrisonGPU · 2025-03-22T11:00:14Z

Hi, @lntue @overmighty , I found an issue with the FMA tests. When we enable the fmaf16 type test, the values of IN_MAX_SUBNORMAL_U and IN_MIN_SUBNORMAL_U are 1023 and 1 respectively:

llvm-project/libc/test/src/math/FmaTest.h

Lines 37 to 46 in 2a45222

    
           static constexpr OutStorageType OUT_MIN_NORMAL_U = 
        
               OutFPBits::min_normal().uintval(); 
        
           static constexpr InStorageType IN_MAX_NORMAL_U = 
        
               InFPBits::max_normal().uintval(); 
        
           static constexpr InStorageType IN_MIN_NORMAL_U = 
        
               InFPBits::min_normal().uintval(); 
        
           static constexpr InStorageType IN_MAX_SUBNORMAL_U = 
        
               InFPBits::max_subnormal().uintval(); 
        
           static constexpr InStorageType IN_MIN_SUBNORMAL_U = 
        
               InFPBits::min_subnormal().uintval();

If we calculate the step as follows:

constexpr InStorageType STEP =
        (IN_MAX_SUBNORMAL_U - IN_MIN_SUBNORMAL_U) / COUNT;

this will result in STEP being zero due to integer division, causing the loop condition never to be met and the loop to run indefinitely:

llvm-project/libc/test/src/math/FmaTest.h

Lines 64 to 73 in 2a45222

    
           for (InStorageType v = IN_MIN_SUBNORMAL_U, w = IN_MAX_SUBNORMAL_U; 
        
                v <= IN_MAX_SUBNORMAL_U && w >= IN_MIN_SUBNORMAL_U; 
        
                v += STEP, w -= STEP) { 
        
             InType x = InFPBits(get_random_bit_pattern()).get_val(); 
        
             InType y = InFPBits(v).get_val(); 
        
             InType z = InFPBits(w).get_val(); 
        
             mpfr::TernaryInput<InType> input{x, y, z}; 
        
             ASSERT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Fma, input, func(x, y, z), 
        
                                            0.5); 
        
           }

,
I have made some modifications in this PR to address this issue. Could you please take a look? :-)

lntue

LGTM. Please wait to see if @overmighty has any other comments.

overmighty

We can remove some includes from libc/test/src/math/smoke/FmaTest.h now that the rounding issues should be solved by fputil::cast. Other than that, LGTM.

libc/test/src/math/smoke/FmaTest.h

libc/test/src/math/smoke/CMakeLists.txt

harrisonGPU · 2025-03-23T02:48:50Z

Thanks!

These tests failed at Google after #130757. Disable them in bazel for the time being.

harrisonGPU requested a review from overmighty March 11, 2025 11:49

harrisonGPU self-assigned this Mar 11, 2025

harrisonGPU requested a review from lntue March 11, 2025 11:50

lntue reviewed Mar 11, 2025

View reviewed changes

libc/test/src/math/CMakeLists.txt Outdated Show resolved Hide resolved

harrisonGPU marked this pull request as ready for review March 12, 2025 02:04

harrisonGPU requested review from rupprecht, keith and aaronmondal as code owners March 12, 2025 02:04

llvmbot added backend:AMDGPU libc bazel "Peripheral" support tier build system: utils/bazel labels Mar 12, 2025

harrisonGPU requested a review from lntue March 12, 2025 02:05

overmighty mentioned this pull request Mar 12, 2025

[gsoc2024][libc][math][c23] Implement _Float16 higher math functions #95250

Open

52 tasks

overmighty requested changes Mar 12, 2025

View reviewed changes

harrisonGPU requested a review from overmighty March 13, 2025 03:23

jhuber6 removed the backend:AMDGPU label Mar 14, 2025

harrisonGPU requested a review from jhuber6 March 14, 2025 15:14

overmighty reviewed Mar 16, 2025

View reviewed changes

harrisonGPU force-pushed the libcFmaf16 branch from a879a22 to 5f2b384 Compare March 19, 2025 10:10

llvmbot added the backend:AMDGPU label Mar 19, 2025

harrisonGPU requested a review from overmighty March 19, 2025 10:12

harrisonGPU removed the backend:AMDGPU label Mar 19, 2025

llvmbot added the backend:AMDGPU label Mar 19, 2025

harrisonGPU added 5 commits March 20, 2025 12:43

[libc][math][c23] Add fmaf16 C23 math function.

7e03a43

[libc][math][c23] Remove FMA_OPT flag.

71149fd

[libc][math][c23] Do some change for comments.

adef262

[libc][math][c23] Update cmake.

2fb8aa8

[libc][math][c23] Fix clang format.

fd02a05

harrisonGPU force-pushed the libcFmaf16 branch from 450d6bb to fd02a05 Compare March 20, 2025 12:43

llvmbot added the backend:AMDGPU label Mar 20, 2025

harrisonGPU removed the backend:AMDGPU label Mar 20, 2025

[libc][math][c23] Update for comments.

b69e80e

llvmbot added the backend:AMDGPU label Mar 20, 2025

harrisonGPU removed the backend:AMDGPU label Mar 21, 2025

[libc][math][c23] Fix test test failure.

f118c0f

llvmbot added the backend:AMDGPU label Mar 22, 2025

harrisonGPU added c23 and removed backend:AMDGPU labels Mar 22, 2025

lntue approved these changes Mar 22, 2025

View reviewed changes

overmighty reviewed Mar 22, 2025

View reviewed changes

libc/test/src/math/smoke/FmaTest.h Show resolved Hide resolved

libc/test/src/math/smoke/CMakeLists.txt Outdated Show resolved Hide resolved

[libc][math][c23] Update for comments.

da6625b

llvmbot added the backend:AMDGPU label Mar 22, 2025

harrisonGPU requested a review from overmighty March 22, 2025 16:49

harrisonGPU removed the backend:AMDGPU label Mar 22, 2025

overmighty approved these changes Mar 22, 2025

View reviewed changes

harrisonGPU merged commit 445837a into llvm:main Mar 23, 2025
18 checks passed

harrisonGPU deleted the libcFmaf16 branch March 23, 2025 02:49

googlewalt mentioned this pull request Mar 25, 2025

Disable some tests on bazel #132951

Merged

googlewalt added a commit that referenced this pull request Mar 25, 2025

Disable some tests on bazel (#132951)

6ddc071

These tests failed at Google after #130757. Disable them in bazel for the time being.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[libc][math][c23] Add fmaf16 C23 math function. #130757

[libc][math][c23] Add fmaf16 C23 math function. #130757

harrisonGPU commented Mar 11, 2025 •

edited

Loading

llvmbot commented Mar 12, 2025 •

edited

Loading

overmighty commented Mar 12, 2025

overmighty Mar 12, 2025

lntue Mar 13, 2025

lntue Mar 13, 2025

harrisonGPU Mar 13, 2025

harrisonGPU Mar 13, 2025

lntue Mar 20, 2025

harrisonGPU Mar 20, 2025 •

edited

Loading

overmighty Mar 12, 2025

harrisonGPU commented Mar 13, 2025

overmighty commented Mar 14, 2025

harrisonGPU commented Mar 14, 2025

overmighty Mar 16, 2025

harrisonGPU Mar 19, 2025

arsenm commented Mar 17, 2025

github-actions bot commented Mar 19, 2025 •

edited

Loading

lntue commented Mar 21, 2025

harrisonGPU commented Mar 21, 2025

harrisonGPU commented Mar 22, 2025

lntue left a comment

overmighty left a comment

harrisonGPU commented Mar 23, 2025

		# TODO: Add fma, fmaf, fmal, fmaf128 functions.

		libc_math_function(name = "fmaf16")

	#if defined(LIBC_TYPES_HAS_FLOAT16) && !defined(__LIBC_USE_FLOAT16_CONVERSION)
	// Rounding modes other than the default might not be usable with float16.
	if constexpr (LIBC_NAMESPACE::cpp::is_same_v<OutType, float16>)
	EXPECT_FP_EQ(OutType(0.75) * z, func(InType(1.75), in_z, -in_z));
	else
	#endif

[libc][math][c23] Add fmaf16 C23 math function. #130757

[libc][math][c23] Add fmaf16 C23 math function. #130757

Conversation

harrisonGPU commented Mar 11, 2025 • edited Loading

llvmbot commented Mar 12, 2025 • edited Loading

overmighty commented Mar 12, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

harrisonGPU Mar 20, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

harrisonGPU commented Mar 13, 2025

overmighty commented Mar 14, 2025

harrisonGPU commented Mar 14, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

arsenm commented Mar 17, 2025

github-actions bot commented Mar 19, 2025 • edited Loading

lntue commented Mar 21, 2025

harrisonGPU commented Mar 21, 2025

harrisonGPU commented Mar 22, 2025

lntue left a comment

Choose a reason for hiding this comment

overmighty left a comment

Choose a reason for hiding this comment

harrisonGPU commented Mar 23, 2025

harrisonGPU commented Mar 11, 2025 •

edited

Loading

llvmbot commented Mar 12, 2025 •

edited

Loading

harrisonGPU Mar 20, 2025 •

edited

Loading

github-actions bot commented Mar 19, 2025 •

edited

Loading