[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions #97072

overmighty · 2024-06-28T15:29:36Z

Part of #93566.

overmighty · 2024-06-28T15:29:55Z

~~Requires #96787.~~

overmighty · 2024-06-28T15:32:53Z

libc/config/linux/x86_64/entrypoints.txt

      # math.h C23 mixed _Float16 and _Float128 entrypoints
+      libc.src.math.f16addf128
      libc.src.math.f16fmaf128
+      libc.src.math.f16subf128
    )


I haven't enabled these on AArch64 because I think they will cause the same Clang 11 backend error (failure to select instructions for fp_round from f128 to f16) as in #96952 due to this code:

https://github.com/llvm/llvm-project/pull/96787/files#diff-9916523646ef256733d16b860bb59b31d4cc2448632c871b6b2256fddadf685bR104-R107

libc/spec/stdc.td

overmighty · 2024-06-29T23:46:24Z

Rebased to fix merge conflicts, re-merged overmighty:libc-math-f16addf, and added this commit:

[libc][math][c23] Move f16{add,sub}{,l} specs to llvm_libc_ext.td

overmighty · 2024-07-02T16:38:20Z

cc @lntue

llvmbot · 2024-07-02T16:38:39Z

@llvm/pr-subscribers-libc

Author: OverMighty (overmighty)

Changes

Part of #93566.

Patch is 30.31 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/97072.diff

31 Files Affected:

(modified) libc/config/linux/aarch64/entrypoints.txt (+2)
(modified) libc/config/linux/x86_64/entrypoints.txt (+6)
(modified) libc/docs/math/index.rst (+2-2)
(modified) libc/spec/llvm_libc_ext.td (+4)
(modified) libc/spec/stdc.td (+4)
(modified) libc/src/math/CMakeLists.txt (+6)
(added) libc/src/math/f16add.h (+20)
(added) libc/src/math/f16addf128.h (+20)
(added) libc/src/math/f16addl.h (+20)
(added) libc/src/math/f16sub.h (+20)
(added) libc/src/math/f16subf128.h (+20)
(added) libc/src/math/f16subl.h (+20)
(modified) libc/src/math/generic/CMakeLists.txt (+78)
(added) libc/src/math/generic/f16add.cpp (+19)
(added) libc/src/math/generic/f16addf128.cpp (+19)
(added) libc/src/math/generic/f16addl.cpp (+19)
(added) libc/src/math/generic/f16sub.cpp (+19)
(added) libc/src/math/generic/f16subf128.cpp (+19)
(added) libc/src/math/generic/f16subl.cpp (+19)
(modified) libc/test/src/math/CMakeLists.txt (+52)
(added) libc/test/src/math/f16add_test.cpp (+13)
(added) libc/test/src/math/f16addl_test.cpp (+13)
(added) libc/test/src/math/f16sub_test.cpp (+13)
(added) libc/test/src/math/f16subl_test.cpp (+13)
(modified) libc/test/src/math/smoke/CMakeLists.txt (+87)
(added) libc/test/src/math/smoke/f16add_test.cpp (+13)
(added) libc/test/src/math/smoke/f16addf128_test.cpp (+13)
(added) libc/test/src/math/smoke/f16addl_test.cpp (+13)
(added) libc/test/src/math/smoke/f16sub_test.cpp (+13)
(added) libc/test/src/math/smoke/f16subf128_test.cpp (+13)
(added) libc/test/src/math/smoke/f16subl_test.cpp (+13)

diff --git a/libc/config/linux/aarch64/entrypoints.txt b/libc/config/linux/aarch64/entrypoints.txt
index ff35e8fffec19..940df63e3912b 100644
--- a/libc/config/linux/aarch64/entrypoints.txt
+++ b/libc/config/linux/aarch64/entrypoints.txt
@@ -508,12 +508,14 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.canonicalizef16
     libc.src.math.ceilf16
     libc.src.math.copysignf16
+    libc.src.math.f16add
     libc.src.math.f16addf
     libc.src.math.f16div
     libc.src.math.f16divf
     libc.src.math.f16fmaf
     libc.src.math.f16sqrt
     libc.src.math.f16sqrtf
+    libc.src.math.f16sub
     libc.src.math.f16subf
     libc.src.math.fabsf16
     libc.src.math.fdimf16
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 3eefa129c9758..09f04fb31dfd8 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -538,7 +538,9 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.canonicalizef16
     libc.src.math.ceilf16
     libc.src.math.copysignf16
+    libc.src.math.f16add
     libc.src.math.f16addf
+    libc.src.math.f16addl
     libc.src.math.f16div
     libc.src.math.f16divf
     libc.src.math.f16divl
@@ -548,7 +550,9 @@ if(LIBC_TYPES_HAS_FLOAT16)
     libc.src.math.f16sqrt
     libc.src.math.f16sqrtf
     libc.src.math.f16sqrtl
+    libc.src.math.f16sub
     libc.src.math.f16subf
+    libc.src.math.f16subl
     libc.src.math.fabsf16
     libc.src.math.fdimf16
     libc.src.math.floorf16
@@ -601,9 +605,11 @@ if(LIBC_TYPES_HAS_FLOAT16)
   if(LIBC_TYPES_HAS_FLOAT128)
     list(APPEND TARGET_LIBM_ENTRYPOINTS
       # math.h C23 mixed _Float16 and _Float128 entrypoints
+      libc.src.math.f16addf128
       libc.src.math.f16divf128
       libc.src.math.f16fmaf128
       libc.src.math.f16sqrtf128
+      libc.src.math.f16subf128
     )
   endif()
 endif()
diff --git a/libc/docs/math/index.rst b/libc/docs/math/index.rst
index e4da3d42baf7a..04f63d03778a8 100644
--- a/libc/docs/math/index.rst
+++ b/libc/docs/math/index.rst
@@ -124,13 +124,13 @@ Basic Operations
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | dsub             | N/A              | N/A             |                        | N/A                  |                        | 7.12.14.2              | F.10.11                    |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| f16add           | |check|\*        |                 |                        | N/A                  |                        | 7.12.14.1              | F.10.11                    |
+| f16add           | |check|\*        | |check|\*       | |check|\*              | N/A                  | |check|                | 7.12.14.1              | F.10.11                    |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | f16div           | |check|\*        | |check|\*       | |check|\*              | N/A                  | |check|                | 7.12.14.4              | F.10.11                    |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | f16fma           | |check|          | |check|         | |check|                | N/A                  | |check|                | 7.12.14.5              | F.10.11                    |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| f16sub           | |check|\*        |                 |                        | N/A                  |                        | 7.12.14.2              | F.10.11                    |
+| f16sub           | |check|\*        | |check|\*       | |check|\*              | N/A                  | |check|                | 7.12.14.2              | F.10.11                    |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
 | fabs             | |check|          | |check|         | |check|                | |check|              | |check|                | 7.12.7.3               | F.10.4.3                   |
 +------------------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
diff --git a/libc/spec/llvm_libc_ext.td b/libc/spec/llvm_libc_ext.td
index ff7831f8bce16..b994e7ca56a93 100644
--- a/libc/spec/llvm_libc_ext.td
+++ b/libc/spec/llvm_libc_ext.td
@@ -57,9 +57,13 @@ def LLVMLibcExt : StandardSpec<"llvm_libc_ext"> {
       [], // Types
       [], // Enumerations
       [
+          GuardedFunctionSpec<"f16add", RetValSpec<Float16Type>, [ArgSpec<DoubleType>, ArgSpec<DoubleType>], "LIBC_TYPES_HAS_FLOAT16">,
           GuardedFunctionSpec<"f16addf", RetValSpec<Float16Type>, [ArgSpec<FloatType>, ArgSpec<FloatType>], "LIBC_TYPES_HAS_FLOAT16">,
+          GuardedFunctionSpec<"f16addl", RetValSpec<Float16Type>, [ArgSpec<LongDoubleType>, ArgSpec<LongDoubleType>], "LIBC_TYPES_HAS_FLOAT16">,
 
+          GuardedFunctionSpec<"f16sub", RetValSpec<Float16Type>, [ArgSpec<DoubleType>, ArgSpec<DoubleType>], "LIBC_TYPES_HAS_FLOAT16">,
           GuardedFunctionSpec<"f16subf", RetValSpec<Float16Type>, [ArgSpec<FloatType>, ArgSpec<FloatType>], "LIBC_TYPES_HAS_FLOAT16">,
+          GuardedFunctionSpec<"f16subl", RetValSpec<Float16Type>, [ArgSpec<LongDoubleType>, ArgSpec<LongDoubleType>], "LIBC_TYPES_HAS_FLOAT16">,
 
           GuardedFunctionSpec<"f16div", RetValSpec<Float16Type>, [ArgSpec<DoubleType>, ArgSpec<DoubleType>], "LIBC_TYPES_HAS_FLOAT16">,
           GuardedFunctionSpec<"f16divf", RetValSpec<Float16Type>, [ArgSpec<FloatType>, ArgSpec<FloatType>], "LIBC_TYPES_HAS_FLOAT16">,
diff --git a/libc/spec/stdc.td b/libc/spec/stdc.td
index 6e59062dde727..9ff40bf76700c 100644
--- a/libc/spec/stdc.td
+++ b/libc/spec/stdc.td
@@ -729,6 +729,10 @@ def StdC : StandardSpec<"stdc"> {
 
           GuardedFunctionSpec<"setpayloadsigf16", RetValSpec<IntType>, [ArgSpec<Float16Ptr>, ArgSpec<Float16Type>], "LIBC_TYPES_HAS_FLOAT16">,
 
+          GuardedFunctionSpec<"f16addf128", RetValSpec<Float16Type>, [ArgSpec<Float128Type>, ArgSpec<Float128Type>], "LIBC_TYPES_HAS_FLOAT16_AND_FLOAT128">,
+
+          GuardedFunctionSpec<"f16subf128", RetValSpec<Float16Type>, [ArgSpec<Float128Type>, ArgSpec<Float128Type>], "LIBC_TYPES_HAS_FLOAT16_AND_FLOAT128">,
+
           GuardedFunctionSpec<"f16divf128", RetValSpec<Float16Type>, [ArgSpec<Float128Type>, ArgSpec<Float128Type>], "LIBC_TYPES_HAS_FLOAT16_AND_FLOAT128">,
 
           GuardedFunctionSpec<"f16sqrtf128", RetValSpec<Float16Type>, [ArgSpec<Float128Type>], "LIBC_TYPES_HAS_FLOAT16_AND_FLOAT128">,
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index 5b20913134fdf..0983d268bd4b8 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -99,7 +99,10 @@ add_math_entrypoint_object(exp10f)
 add_math_entrypoint_object(expm1)
 add_math_entrypoint_object(expm1f)
 
+add_math_entrypoint_object(f16add)
 add_math_entrypoint_object(f16addf)
+add_math_entrypoint_object(f16addl)
+add_math_entrypoint_object(f16addf128)
 
 add_math_entrypoint_object(f16div)
 add_math_entrypoint_object(f16divf)
@@ -116,7 +119,10 @@ add_math_entrypoint_object(f16sqrtf)
 add_math_entrypoint_object(f16sqrtl)
 add_math_entrypoint_object(f16sqrtf128)
 
+add_math_entrypoint_object(f16sub)
 add_math_entrypoint_object(f16subf)
+add_math_entrypoint_object(f16subl)
+add_math_entrypoint_object(f16subf128)
 
 add_math_entrypoint_object(fabs)
 add_math_entrypoint_object(fabsf)
diff --git a/libc/src/math/f16add.h b/libc/src/math/f16add.h
new file mode 100644
index 0000000000000..763a0787d860d
--- /dev/null
+++ b/libc/src/math/f16add.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16add ------------------------*- 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_F16ADD_H
+#define LLVM_LIBC_SRC_MATH_F16ADD_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16add(double x, double y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16ADD_H
diff --git a/libc/src/math/f16addf128.h b/libc/src/math/f16addf128.h
new file mode 100644
index 0000000000000..284ce1d303775
--- /dev/null
+++ b/libc/src/math/f16addf128.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16addf128 --------------------*- 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_F16ADDF128_H
+#define LLVM_LIBC_SRC_MATH_F16ADDF128_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16addf128(float128 x, float128 y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16ADDF128_H
diff --git a/libc/src/math/f16addl.h b/libc/src/math/f16addl.h
new file mode 100644
index 0000000000000..6a7267a10d0c6
--- /dev/null
+++ b/libc/src/math/f16addl.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16addl -----------------------*- 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_F16ADDL_H
+#define LLVM_LIBC_SRC_MATH_F16ADDL_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16addl(long double x, long double y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16ADDL_H
diff --git a/libc/src/math/f16sub.h b/libc/src/math/f16sub.h
new file mode 100644
index 0000000000000..66f82daada019
--- /dev/null
+++ b/libc/src/math/f16sub.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16sub ------------------------*- 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_F16SUB_H
+#define LLVM_LIBC_SRC_MATH_F16SUB_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16sub(double x, double y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16SUB_H
diff --git a/libc/src/math/f16subf128.h b/libc/src/math/f16subf128.h
new file mode 100644
index 0000000000000..eb674297ba266
--- /dev/null
+++ b/libc/src/math/f16subf128.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16subf128 --------------------*- 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_F16SUBF128_H
+#define LLVM_LIBC_SRC_MATH_F16SUBF128_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16subf128(float128 x, float128 y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16SUBF128_H
diff --git a/libc/src/math/f16subl.h b/libc/src/math/f16subl.h
new file mode 100644
index 0000000000000..43b44a57a604e
--- /dev/null
+++ b/libc/src/math/f16subl.h
@@ -0,0 +1,20 @@
+//===-- Implementation header for f16subl -----------------------*- 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_F16SUBL_H
+#define LLVM_LIBC_SRC_MATH_F16SUBL_H
+
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE {
+
+float16 f16subl(long double x, long double y);
+
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC_MATH_F16SUBL_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index d6ea8c54174b6..2e4ed8f2961da 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -3795,6 +3795,19 @@ add_entrypoint_object(
     -O3
 )
 
+add_entrypoint_object(
+  f16add
+  SRCS
+    f16add.cpp
+  HDRS
+    ../f16add.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
 add_entrypoint_object(
   f16addf
   SRCS
@@ -3808,6 +3821,45 @@ add_entrypoint_object(
     -O3
 )
 
+add_entrypoint_object(
+  f16addl
+  SRCS
+    f16addl.cpp
+  HDRS
+    ../f16addl.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
+add_entrypoint_object(
+  f16addf128
+  SRCS
+    f16addf128.cpp
+  HDRS
+    ../f16addf128.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
+add_entrypoint_object(
+  f16sub
+  SRCS
+    f16sub.cpp
+  HDRS
+    ../f16sub.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
 add_entrypoint_object(
   f16subf
   SRCS
@@ -3821,6 +3873,32 @@ add_entrypoint_object(
     -O3
 )
 
+add_entrypoint_object(
+  f16subl
+  SRCS
+    f16subl.cpp
+  HDRS
+    ../f16subl.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
+add_entrypoint_object(
+  f16subf128
+  SRCS
+    f16subf128.cpp
+  HDRS
+    ../f16subf128.h
+  DEPENDS
+    libc.src.__support.macros.properties.types
+    libc.src.__support.FPUtil.generic.add_sub
+  COMPILE_OPTIONS
+    -O3
+)
+
 add_entrypoint_object(
   f16div
   SRCS
diff --git a/libc/src/math/generic/f16add.cpp b/libc/src/math/generic/f16add.cpp
new file mode 100644
index 0000000000000..ef9b43e9f46aa
--- /dev/null
+++ b/libc/src/math/generic/f16add.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16add 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/f16add.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16add, (double x, double y)) {
+  return fputil::generic::add<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/f16addf128.cpp b/libc/src/math/generic/f16addf128.cpp
new file mode 100644
index 0000000000000..61c458f7d5de1
--- /dev/null
+++ b/libc/src/math/generic/f16addf128.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16addf128 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/f16addf128.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16addf128, (float128 x, float128 y)) {
+  return fputil::generic::add<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/f16addl.cpp b/libc/src/math/generic/f16addl.cpp
new file mode 100644
index 0000000000000..d32d09d0dbb83
--- /dev/null
+++ b/libc/src/math/generic/f16addl.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16addl 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/f16addl.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16addl, (long double x, long double y)) {
+  return fputil::generic::add<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/f16sub.cpp b/libc/src/math/generic/f16sub.cpp
new file mode 100644
index 0000000000000..114c8ad3155e1
--- /dev/null
+++ b/libc/src/math/generic/f16sub.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16sub 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/f16sub.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16sub, (double x, double y)) {
+  return fputil::generic::sub<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/f16subf128.cpp b/libc/src/math/generic/f16subf128.cpp
new file mode 100644
index 0000000000000..1f9ff28abdf29
--- /dev/null
+++ b/libc/src/math/generic/f16subf128.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16subf128 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/f16subf128.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16subf128, (float128 x, float128 y)) {
+  return fputil::generic::sub<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/src/math/generic/f16subl.cpp b/libc/src/math/generic/f16subl.cpp
new file mode 100644
index 0000000000000..31970af9a2366
--- /dev/null
+++ b/libc/src/math/generic/f16subl.cpp
@@ -0,0 +1,19 @@
+//===-- Implementation of f16subl 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/f16subl.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(float16, f16subl, (long double x, long double y)) {
+  return fputil::generic::sub<float16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/...
[truncated]

lntue · 2024-07-02T22:57:01Z

libc/config/linux/aarch64/entrypoints.txt

@@ -508,12 +508,14 @@ if(LIBC_TYPES_HAS_FLOAT16)
    libc.src.math.canonicalizef16
    libc.src.math.ceilf16
    libc.src.math.copysignf16
+    libc.src.math.f16add


no f16addl and f16addf128 for aarch64?

See my comment above:

I haven't enabled these on AArch64 because I think they will cause the same Clang 11 backend error (failure to select instructions for fp_round from f128 to f16) as in #96952 due to this code:

https://github.com/llvm/llvm-project/pull/96787/files#diff-9916523646ef256733d16b860bb59b31d4cc2448632c871b6b2256fddadf685bR104-R107

#97072 (comment)

I could still try just in case.

…97072) Part of llvm#93566.

overmighty commented Jun 28, 2024

View reviewed changes

overmighty mentioned this pull request Jun 28, 2024

[gsoc2024][libc][math][c23] Implement _Float16 basic operations #93566

Closed

70 tasks

overmighty force-pushed the libc-math-f16add-variants branch from cedb6d7 to 83d4b90 Compare June 29, 2024 23:45

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions

52ef221

overmighty force-pushed the libc-math-f16add-variants branch from 83d4b90 to 52ef221 Compare July 2, 2024 16:37

overmighty marked this pull request as ready for review July 2, 2024 16:38

llvmbot added the libc label Jul 2, 2024

lntue reviewed Jul 2, 2024

View reviewed changes

lntue approved these changes Jul 2, 2024

View reviewed changes

lntue merged commit 4e56724 into llvm:main Jul 2, 2024
10 checks passed

lravenclaw pushed a commit to lravenclaw/llvm-project that referenced this pull request Jul 3, 2024

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions (llvm#…

d63dd06

…97072) Part of llvm#93566.

kbluck pushed a commit to kbluck/llvm-project that referenced this pull request Jul 6, 2024

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions (llvm#…

f75bfe8

…97072) Part of llvm#93566.

overmighty mentioned this pull request Aug 3, 2024

[libc] enable most of the entrypoints on aarch64 #101797

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions #97072

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions #97072

Uh oh!

overmighty commented Jun 28, 2024

Uh oh!

overmighty commented Jun 28, 2024 •

edited

Loading

Uh oh!

overmighty Jun 28, 2024 •

edited

Loading

Uh oh!

Uh oh!

overmighty commented Jun 29, 2024

Uh oh!

overmighty commented Jul 2, 2024

Uh oh!

llvmbot commented Jul 2, 2024

Uh oh!

lntue Jul 2, 2024

Uh oh!

overmighty Jul 2, 2024

Uh oh!

Uh oh!

Uh oh!

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions #97072

[libc][math][c23] Add f16{add,sub}{,l,f128} C23 math functions #97072

Uh oh!

Conversation

overmighty commented Jun 28, 2024

Uh oh!

overmighty commented Jun 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

overmighty Jun 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

overmighty commented Jun 29, 2024

Uh oh!

overmighty commented Jul 2, 2024

Uh oh!

llvmbot commented Jul 2, 2024

Uh oh!

lntue Jul 2, 2024

Choose a reason for hiding this comment

Uh oh!

overmighty Jul 2, 2024

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

overmighty commented Jun 28, 2024 •

edited

Loading

overmighty Jun 28, 2024 •

edited

Loading