[llvm] [libc][bazel] Add targets for float16 math (PR #129965)
via llvm-commits
llvm-commits at lists.llvm.org
Wed Mar 5 17:17:08 PST 2025
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-libc
Author: Michael Jones (michaelrj-google)
<details>
<summary>Changes</summary>
Add targets for some float16 math functions that were missing them, but
mostly add targets for the smoke tests.
---
Full diff: https://github.com/llvm/llvm-project/pull/129965.diff
2 Files Affected:
- (modified) utils/bazel/llvm-project-overlay/libc/BUILD.bazel (+58)
- (modified) utils/bazel/llvm-project-overlay/libc/test/src/math/smoke/BUILD.bazel (+432-32)
``````````diff
diff --git a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
index d6eb07511cf77..5339b30154ed1 100644
--- a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
@@ -2294,6 +2294,26 @@ libc_math_function(name = "canonicalizef128")
libc_math_function(name = "canonicalizef16")
+libc_math_function(name = "iscanonical")
+
+libc_math_function(name = "iscanonicalf")
+
+libc_math_function(name = "iscanonicall")
+
+libc_math_function(name = "iscanonicalf128")
+
+libc_math_function(name = "iscanonicalf16")
+
+libc_math_function(name = "issignaling")
+
+libc_math_function(name = "issignalingf")
+
+libc_math_function(name = "issignalingl")
+
+libc_math_function(name = "issignalingf128")
+
+libc_math_function(name = "issignalingf16")
+
libc_math_function(
name = "cbrt",
additional_deps = [
@@ -2352,6 +2372,15 @@ libc_math_function(
],
)
+libc_math_function(
+ name = "cosf16",
+ additional_deps = [
+ ":__support_fputil_multiply_add",
+ ":__support_macros_optimization",
+ ":sincosf16_utils",
+ ],
+)
+
libc_math_function(
name = "coshf",
additional_deps = [
@@ -3445,6 +3474,15 @@ libc_math_function(
],
)
+libc_math_function(
+ name = "sinf16",
+ additional_deps = [
+ ":__support_fputil_nearest_integer",
+ ":__support_fputil_polyeval",
+ ":sincosf16_utils",
+ ],
+)
+
libc_math_function(
name = "sincos",
additional_deps = [
@@ -3565,6 +3603,15 @@ libc_math_function(
],
)
+libc_math_function(
+ name = "tanf16",
+ additional_deps = [
+ ":__support_fputil_nearest_integer",
+ ":__support_fputil_polyeval",
+ ":sincosf16_utils",
+ ],
+)
+
libc_math_function(
name = "tanhf",
additional_deps = [
@@ -3587,6 +3634,17 @@ libc_math_function(
],
)
+libc_math_function(
+ name = "tanpif16",
+ additional_deps = [
+ ":sincosf16_utils",
+ ":hdr_errno_macros",
+ ":hdr_fenv_macros",
+ ":__support_fputil_cast",
+ ":__support_fputil_multiply_add",
+ ],
+)
+
libc_math_function(name = "totalorder")
libc_math_function(name = "totalorderf")
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 8f519e1a15117..c56f3c7cf4f34 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
@@ -58,6 +58,19 @@ math_test(
],
)
+math_test(
+ name = "canonicalizef16",
+ hdrs = ["CanonicalizeTest.h"],
+ deps = [
+ "//libc:__support_integer_literals",
+ ],
+)
+
+math_test(
+ name = "iscanonicalf16",
+ hdrs = ["IsCanonicalTest.h"],
+)
+
math_test(name = "cbrt")
math_test(name = "cbrtf")
@@ -82,6 +95,11 @@ math_test(
hdrs = ["CeilTest.h"],
)
+math_test(
+ name = "ceilf16",
+ hdrs = ["CeilTest.h"],
+)
+
math_test(
name = "copysign",
hdrs = ["CopySignTest.h"],
@@ -102,6 +120,11 @@ math_test(
hdrs = ["CopySignTest.h"],
)
+math_test(
+ name = "copysignf16",
+ hdrs = ["CopySignTest.h"],
+)
+
math_test(name = "cos")
math_test(name = "cosf")
@@ -125,6 +148,21 @@ math_test(
hdrs = ["AddTest.h"],
)
+math_test(
+ name = "f16add",
+ hdrs = ["AddTest.h"],
+)
+
+math_test(
+ name = "f16addf",
+ hdrs = ["AddTest.h"],
+)
+
+math_test(
+ name = "f16addl",
+ hdrs = ["AddTest.h"],
+)
+
math_test(
name = "ddivl",
hdrs = ["DivTest.h"],
@@ -135,6 +173,21 @@ math_test(
hdrs = ["DivTest.h"],
)
+math_test(
+ name = "f16div",
+ hdrs = ["DivTest.h"],
+)
+
+math_test(
+ name = "f16divf",
+ hdrs = ["DivTest.h"],
+)
+
+math_test(
+ name = "f16divl",
+ hdrs = ["DivTest.h"],
+)
+
math_test(
name = "dfmal",
hdrs = ["FmaTest.h"],
@@ -145,6 +198,21 @@ math_test(
hdrs = ["FmaTest.h"],
)
+math_test(
+ name = "f16fma",
+ hdrs = ["FmaTest.h"],
+)
+
+math_test(
+ name = "f16fmaf",
+ hdrs = ["FmaTest.h"],
+)
+
+math_test(
+ name = "f16fmal",
+ hdrs = ["FmaTest.h"],
+)
+
math_test(
name = "dmull",
hdrs = ["MulTest.h"],
@@ -155,6 +223,36 @@ math_test(
hdrs = ["MulTest.h"],
)
+math_test(
+ name = "f16mul",
+ hdrs = ["MulTest.h"],
+)
+
+math_test(
+ name = "f16mulf",
+ hdrs = ["MulTest.h"],
+)
+
+math_test(
+ name = "f16mull",
+ hdrs = ["MulTest.h"],
+)
+
+math_test(
+ name = "fsqrt",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "fsqrtl",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "fsqrtf128",
+ hdrs = ["SqrtTest.h"],
+)
+
math_test(
name = "dsqrtl",
hdrs = ["SqrtTest.h"],
@@ -165,6 +263,46 @@ math_test(
hdrs = ["SqrtTest.h"],
)
+math_test(
+ name = "sqrt",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "sqrtf",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "sqrtl",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "sqrtf128",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "f16sqrt",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "f16sqrtf",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "sqrtf16",
+ hdrs = ["SqrtTest.h"],
+)
+
+math_test(
+ name = "f16sqrtl",
+ hdrs = ["SqrtTest.h"],
+)
+
math_test(
name = "dsubl",
hdrs = ["SubTest.h"],
@@ -175,6 +313,21 @@ math_test(
hdrs = ["SubTest.h"],
)
+math_test(
+ name = "f16sub",
+ hdrs = ["SubTest.h"],
+)
+
+math_test(
+ name = "f16subf",
+ hdrs = ["SubTest.h"],
+)
+
+math_test(
+ name = "f16subl",
+ hdrs = ["SubTest.h"],
+)
+
math_test(name = "erff")
math_test(name = "exp")
@@ -215,6 +368,11 @@ math_test(
hdrs = ["FAbsTest.h"],
)
+math_test(
+ name = "fabsf16",
+ hdrs = ["FAbsTest.h"],
+)
+
math_test(
name = "fadd",
hdrs = ["AddTest.h"],
@@ -250,6 +408,11 @@ math_test(
hdrs = ["FDimTest.h"],
)
+math_test(
+ name = "fdimf16",
+ hdrs = ["FDimTest.h"],
+)
+
math_test(
name = "fdiv",
hdrs = ["DivTest.h"],
@@ -300,6 +463,11 @@ math_test(
hdrs = ["FloorTest.h"],
)
+math_test(
+ name = "floorf16",
+ hdrs = ["FloorTest.h"],
+)
+
# TODO: Add fma, fmaf, fmal, fmaf128 tests.
math_test(
@@ -322,6 +490,11 @@ math_test(
hdrs = ["FMaxTest.h"],
)
+math_test(
+ name = "fmaxf16",
+ hdrs = ["FMaxTest.h"],
+)
+
math_test(
name = "fmaximum",
hdrs = ["FMaximumTest.h"],
@@ -402,6 +575,26 @@ math_test(
hdrs = ["FMaximumNumTest.h"],
)
+math_test(
+ name = "fmaximum_mag_numf16",
+ hdrs = ["FMaximumMagNumTest.h"],
+)
+
+math_test(
+ name = "fmaximum_magf16",
+ hdrs = ["FMaximumMagTest.h"],
+)
+
+math_test(
+ name = "fmaximum_numf16",
+ hdrs = ["FMaximumNumTest.h"],
+)
+
+math_test(
+ name = "fmaximumf16",
+ hdrs = ["FMaximumTest.h"],
+)
+
math_test(
name = "fmin",
hdrs = ["FMinTest.h"],
@@ -422,6 +615,11 @@ math_test(
hdrs = ["FMinTest.h"],
)
+math_test(
+ name = "fminf16",
+ hdrs = ["FMinTest.h"],
+)
+
math_test(
name = "fminimum",
hdrs = ["FMinimumTest.h"],
@@ -462,6 +660,11 @@ math_test(
hdrs = ["FMinimumMagTest.h"],
)
+math_test(
+ name = "fminimum_magf16",
+ hdrs = ["FMinimumMagTest.h"],
+)
+
math_test(
name = "fminimum_mag_num",
hdrs = ["FMinimumMagNumTest.h"],
@@ -482,6 +685,11 @@ math_test(
hdrs = ["FMinimumMagNumTest.h"],
)
+math_test(
+ name = "fminimum_mag_numf16",
+ hdrs = ["FMinimumMagNumTest.h"],
+)
+
math_test(
name = "fminimum_num",
hdrs = ["FMinimumNumTest.h"],
@@ -502,6 +710,16 @@ math_test(
hdrs = ["FMinimumNumTest.h"],
)
+math_test(
+ name = "fminimum_numf16",
+ hdrs = ["FMinimumNumTest.h"],
+)
+
+math_test(
+ name = "fminimumf16",
+ hdrs = ["FMinimumTest.h"],
+)
+
math_test(
name = "fmod",
hdrs = ["FModTest.h"],
@@ -522,6 +740,11 @@ math_test(
hdrs = ["FModTest.h"],
)
+math_test(
+ name = "fmodf16",
+ hdrs = ["FModTest.h"],
+)
+
math_test(
name = "fmul",
hdrs = ["MulTest.h"],
@@ -557,6 +780,11 @@ math_test(
hdrs = ["FrexpTest.h"],
)
+math_test(
+ name = "frexpf16",
+ hdrs = ["FrexpTest.h"],
+)
+
math_test(
name = "fromfp",
hdrs = ["FromfpTest.h"],
@@ -577,6 +805,11 @@ math_test(
hdrs = ["FromfpTest.h"],
)
+math_test(
+ name = "fromfpf16",
+ hdrs = ["FromfpTest.h"],
+)
+
math_test(
name = "fromfpx",
hdrs = ["FromfpxTest.h"],
@@ -598,18 +831,8 @@ math_test(
)
math_test(
- name = "fsqrt",
- hdrs = ["SqrtTest.h"],
-)
-
-math_test(
- name = "fsqrtl",
- hdrs = ["SqrtTest.h"],
-)
-
-math_test(
- name = "fsqrtf128",
- hdrs = ["SqrtTest.h"],
+ name = "fromfpxf16",
+ hdrs = ["FromfpxTest.h"],
)
math_test(
@@ -647,6 +870,11 @@ math_test(
hdrs = ["GetPayloadTest.h"],
)
+math_test(
+ name = "getpayloadf16",
+ hdrs = ["GetPayloadTest.h"],
+)
+
math_test(
name = "hypot",
hdrs = ["HypotTest.h"],
@@ -697,6 +925,11 @@ math_test(
hdrs = ["LdExpTest.h"],
)
+math_test(
+ name = "ldexpf16",
+ hdrs = ["LdExpTest.h"],
+)
+
math_test(
name = "llogb",
hdrs = ["ILogbTest.h"],
@@ -717,6 +950,16 @@ math_test(
hdrs = ["ILogbTest.h"],
)
+math_test(
+ name = "ilogbf16",
+ hdrs = ["ILogbTest.h"],
+)
+
+math_test(
+ name = "llogbf16",
+ hdrs = ["ILogbTest.h"],
+)
+
math_test(
name = "llrint",
hdrs = ["RoundToIntegerTest.h"],
@@ -737,6 +980,16 @@ math_test(
hdrs = ["RoundToIntegerTest.h"],
)
+math_test(
+ name = "llrintf16",
+ hdrs = ["RoundToIntegerTest.h"],
+)
+
+math_test(
+ name = "lrintf16",
+ hdrs = ["RoundToIntegerTest.h"],
+)
+
math_test(
name = "llround",
hdrs = ["RoundToIntegerTest.h"],
@@ -757,6 +1010,16 @@ math_test(
hdrs = ["RoundToIntegerTest.h"],
)
+math_test(
+ name = "llroundf16",
+ hdrs = ["RoundToIntegerTest.h"],
+)
+
+math_test(
+ name = "lroundf16",
+ hdrs = ["RoundToIntegerTest.h"],
+)
+
math_test(name = "log")
math_test(name = "logf")
@@ -793,6 +1056,11 @@ math_test(
hdrs = ["LogbTest.h"],
)
+math_test(
+ name = "logbf16",
+ hdrs = ["LogbTest.h"],
+)
+
math_test(
name = "lrint",
hdrs = ["RoundToIntegerTest.h"],
@@ -853,6 +1121,11 @@ math_test(
hdrs = ["ModfTest.h"],
)
+math_test(
+ name = "modff16",
+ hdrs = ["ModfTest.h"],
+)
+
# TODO: add nan tests.
math_test(
@@ -875,6 +1148,11 @@ math_test(
hdrs = ["NearbyIntTest.h"],
)
+math_test(
+ name = "nearbyintf16",
+ hdrs = ["NearbyIntTest.h"],
+)
+
math_test(
name = "nextafter",
hdrs = ["NextAfterTest.h"],
@@ -895,6 +1173,11 @@ math_test(
hdrs = ["NextAfterTest.h"],
)
+math_test(
+ name = "nextafterf16",
+ hdrs = ["NextAfterTest.h"],
+)
+
math_test(
name = "nextdown",
hdrs = ["NextDownTest.h"],
@@ -915,6 +1198,11 @@ math_test(
hdrs = ["NextDownTest.h"],
)
+math_test(
+ name = "nextdownf16",
+ hdrs = ["NextDownTest.h"],
+)
+
math_test(
name = "nexttoward",
hdrs = ["NextTowardTest.h"],
@@ -930,6 +1218,11 @@ math_test(
hdrs = ["NextTowardTest.h"],
)
+math_test(
+ name = "nexttowardf16",
+ hdrs = ["NextTowardTest.h"],
+)
+
math_test(
name = "nextup",
hdrs = ["NextUpTest.h"],
@@ -950,6 +1243,11 @@ math_test(
hdrs = ["NextUpTest.h"],
)
+math_test(
+ name = "nextupf16",
+ hdrs = ["NextUpTest.h"],
+)
+
math_test(name = "pow")
math_test(name = "powf")
@@ -974,6 +1272,11 @@ math_test(
hdrs = ["RemQuoTest.h"],
)
+math_test(
+ name = "remquof16",
+ hdrs = ["RemQuoTest.h"],
+)
+
math_test(
name = "rint",
hdrs = ["RIntTest.h"],
@@ -994,6 +1297,11 @@ math_test(
hdrs = ["RIntTest.h"],
)
+math_test(
+ name = "rintf16",
+ hdrs = ["RIntTest.h"],
+)
+
math_test(
name = "roundeven",
hdrs = ["RoundEvenTest.h"],
@@ -1014,6 +1322,11 @@ math_test(
hdrs = ["RoundEvenTest.h"],
)
+math_test(
+ name = "roundevenf16",
+ hdrs = ["RoundEvenTest.h"],
+)
+
math_test(
name = "round",
hdrs = ["RoundTest.h"],
@@ -1034,6 +1347,11 @@ math_test(
hdrs = ["RoundTest.h"],
)
+math_test(
+ name = "roundf16",
+ hdrs = ["RoundTest.h"],
+)
+
math_test(
name = "scalbn",
hdrs = [
@@ -1066,6 +1384,22 @@ math_test(
],
)
+math_test(
+ name = "scalblnf16",
+ hdrs = [
+ "LdExpTest.h",
+ "ScalbnTest.h",
+ ],
+)
+
+math_test(
+ name = "scalbnf16",
+ hdrs = [
+ "LdExpTest.h",
+ "ScalbnTest.h",
+ ],
+)
+
math_test(
name = "scalbln",
hdrs = [
@@ -1118,6 +1452,11 @@ math_test(
hdrs = ["SetPayloadTest.h"],
)
+math_test(
+ name = "setpayloadf16",
+ hdrs = ["SetPayloadTest.h"],
+)
+
math_test(
name = "setpayloadsig",
hdrs = ["SetPayloadSigTest.h"],
@@ -1138,6 +1477,11 @@ math_test(
hdrs = ["SetPayloadSigTest.h"],
)
+math_test(
+ name = "setpayloadsigf16",
+ hdrs = ["SetPayloadSigTest.h"],
+)
+
math_test(name = "sin")
math_test(name = "sinf")
@@ -1155,26 +1499,6 @@ math_test(
math_test(name = "sinpif")
-math_test(
- name = "sqrt",
- hdrs = ["SqrtTest.h"],
-)
-
-math_test(
- name = "sqrtf",
- hdrs = ["SqrtTest.h"],
-)
-
-math_test(
- name = "sqrtl",
- hdrs = ["SqrtTest.h"],
-)
-
-math_test(
- name = "sqrtf128",
- hdrs = ["SqrtTest.h"],
-)
-
math_test(name = "tan")
math_test(name = "tanf")
@@ -1201,6 +1525,11 @@ math_test(
hdrs = ["TotalOrderTest.h"],
)
+math_test(
+ name = "totalorderf16",
+ hdrs = ["TotalOrderTest.h"],
+)
+
math_test(
name = "totalordermag",
hdrs = ["TotalOrderMagTest.h"],
@@ -1221,6 +1550,11 @@ math_test(
hdrs = ["TotalOrderMagTest.h"],
)
+math_test(
+ name = "totalordermagf16",
+ hdrs = ["TotalOrderMagTest.h"],
+)
+
math_test(
name = "trunc",
hdrs = ["TruncTest.h"],
@@ -1241,6 +1575,11 @@ math_test(
hdrs = ["TruncTest.h"],
)
+math_test(
+ name = "truncf16",
+ hdrs = ["TruncTest.h"],
+)
+
math_test(
name = "ufromfp",
hdrs = ["UfromfpTest.h"],
@@ -1261,6 +1600,11 @@ math_test(
hdrs = ["UfromfpTest.h"],
)
+math_test(
+ name = "ufromfpf16",
+ hdrs = ["UfromfpTest.h"],
+)
+
math_test(
name = "ufromfpx",
hdrs = ["UfromfpxTest.h"],
@@ -1280,3 +1624,59 @@ math_test(
name = "ufromfpxf128",
hdrs = ["UfromfpxTest.h"],
)
+
+math_test(
+ name = "ufromfpxf16",
+ hdrs = ["UfromfpxTest.h"],
+)
+
+math_test(name = "cosf16")
+
+math_test(name = "coshf16")
+
+math_test(name = "cospif16")
+
+math_test(name = "exp10f16")
+
+math_test(name = "exp10m1f16")
+
+math_test(name = "exp2f16")
+
+math_test(name = "exp2m1f16")
+
+math_test(name = "expf16")
+
+math_test(name = "expm1f16")
+
+math_test(
+ name = "issignalingf16",
+ hdrs = ["IsSignalingTest.h"],
+)
+
+math_test(name = "log10f16")
+
+math_test(name = "log2f16")
+
+math_test(name = "logf16")
+
+math_test(
+ name = "nanf16",
+ deps = [
+ "//libc:__support_macros_sanitizer",
+ "//libc:hdr_signal_macros",
+ ],
+)
+
+# math_test(name = "remainderf16") #TODO: add remainderf16 tests
+
+math_test(name = "sinf16")
+
+math_test(name = "sinhf16")
+
+math_test(name = "sinpif16")
+
+math_test(name = "tanf16")
+
+math_test(name = "tanhf16")
+
+math_test(name = "tanpif16")
``````````
</details>
https://github.com/llvm/llvm-project/pull/129965
More information about the llvm-commits
mailing list