This is an archive of the discontinued LLVM Phabricator instance.

Differential D131876

[mlir][math] Lower math.floor,ceil to libm
ClosedPublic

Authored by Lewuathe on Aug 15 2022, 12:10 AM.

Details

Reviewers
ftynse
Commits
rGad714d5b7420: [mlir][math] Lower math.floor,ceil to libm
Summary

Lower math.floor and math.ceil to libm

Diff Detail

Event Timeline

Lewuathe created this revision.Aug 15 2022, 12:10 AM
Herald added a project: Restricted Project. · View Herald TranscriptAug 15 2022, 12:10 AM
Herald added subscribers: bzcheeseman, sdasgup3, wenzhicui and 18 others. · View Herald Transcript
Lewuathe requested review of this revision.Aug 15 2022, 12:10 AM
Herald added a project: Restricted Project. · View Herald TranscriptAug 15 2022, 12:10 AM
Herald added subscribers: stephenneuendorffer, nicolasvasilache. · View Herald Transcript
Harbormaster completed remote builds in B181223: Diff 452594.Aug 15 2022, 12:44 AM
ftynse accepted this revision.Aug 23 2022, 4:29 AM
ftynse added inline comments.
mlir/test/Conversion/MathToLibm/convert-to-libm.mlir
334

Please add the newline.

This revision is now accepted and ready to land.Aug 23 2022, 4:29 AM
Lewuathe updated this revision to Diff 455045.Aug 23 2022, 7:27 PM

Add new line

Lewuathe updated this revision to Diff 455047.Aug 23 2022, 7:28 PM

Remove unnecessary newline.

Harbormaster completed remote builds in B182998: Diff 455047.Aug 23 2022, 9:23 PM
Closed by commit rGad714d5b7420: [mlir][math] Lower math.floor,ceil to libm (authored by Lewuathe). · Explain WhyAug 23 2022, 9:38 PM
This revision was automatically updated to reflect the committed changes.
Lewuathe added a commit: rGad714d5b7420: [mlir][math] Lower math.floor,ceil to libm.

Revision Contents

PathSize
mlir/
lib/
Conversion/
MathToLibm/
4 lines
test/
Conversion/
MathToLibm/
28 lines

Diff 452594

mlir/lib/Conversion/MathToLibm/MathToLibm.cpp

Show First 20 Lines • Show All 162 Lines • ▼ Show 20 Linesvoid mlir::populateMathToLibmConversionPatterns(RewritePatternSet &patterns,
patterns.add<ScalarOpToLibmCall<math::TanhOp>>(patterns.getContext(), "tanhf", patterns.add<ScalarOpToLibmCall<math::TanhOp>>(patterns.getContext(), "tanhf",
"tanh", benefit); "tanh", benefit);
patterns.add<ScalarOpToLibmCall<math::RoundOp>>(patterns.getContext(), patterns.add<ScalarOpToLibmCall<math::RoundOp>>(patterns.getContext(),
"roundf", "round", benefit); "roundf", "round", benefit);
patterns.add<ScalarOpToLibmCall<math::CosOp>>(patterns.getContext(), "cosf", patterns.add<ScalarOpToLibmCall<math::CosOp>>(patterns.getContext(), "cosf",
"cos", benefit); "cos", benefit);
patterns.add<ScalarOpToLibmCall<math::SinOp>>(patterns.getContext(), "sinf", patterns.add<ScalarOpToLibmCall<math::SinOp>>(patterns.getContext(), "sinf",
"sin", benefit); "sin", benefit);
patterns.add<ScalarOpToLibmCall<math::FloorOp>>(patterns.getContext(),
"floorf", "floor", benefit);
patterns.add<ScalarOpToLibmCall<math::CeilOp>>(patterns.getContext(), "ceilf",
"ceil", benefit);
} }
namespace { namespace {
struct ConvertMathToLibmPass struct ConvertMathToLibmPass
: public ConvertMathToLibmBase<ConvertMathToLibmPass> { : public ConvertMathToLibmBase<ConvertMathToLibmPass> {
void runOnOperation() override; void runOnOperation() override;
}; };
} // namespace } // namespace
Show All 18 Lines

mlir/test/Conversion/MathToLibm/convert-to-libm.mlir

Show All 12 Lines
// CHECK-DAG: @tanh(f64) -> f64 // CHECK-DAG: @tanh(f64) -> f64
// CHECK-DAG: @tanhf(f32) -> f32 // CHECK-DAG: @tanhf(f32) -> f32
// CHECK-DAG: @round(f64) -> f64 // CHECK-DAG: @round(f64) -> f64
// CHECK-DAG: @roundf(f32) -> f32 // CHECK-DAG: @roundf(f32) -> f32
// CHECK-DAG: @cos(f64) -> f64 // CHECK-DAG: @cos(f64) -> f64
// CHECK-DAG: @cosf(f32) -> f32 // CHECK-DAG: @cosf(f32) -> f32
// CHECK-DAG: @sin(f64) -> f64 // CHECK-DAG: @sin(f64) -> f64
// CHECK-DAG: @sinf(f32) -> f32 // CHECK-DAG: @sinf(f32) -> f32
// CHECK-DAG: @floor(f64) -> f64
// CHECK-DAG: @floorf(f32) -> f32
// CHECK-DAG: @ceil(f64) -> f64
// CHECK-DAG: @ceilf(f32) -> f32
// CHECK-LABEL: func @atan_caller // CHECK-LABEL: func @atan_caller
// CHECK-SAME: %[[FLOAT:.*]]: f32 // CHECK-SAME: %[[FLOAT:.*]]: f32
// CHECK-SAME: %[[DOUBLE:.*]]: f64 // CHECK-SAME: %[[DOUBLE:.*]]: f64
// CHECK-SAME: %[[HALF:.*]]: f16 // CHECK-SAME: %[[HALF:.*]]: f16
// CHECK-SAME: %[[BFLOAT:.*]]: bf16 // CHECK-SAME: %[[BFLOAT:.*]]: bf16
func.func @atan_caller(%float: f32, %double: f64, %half: f16, %bfloat: bf16) -> (f32, f64, f16, bf16) { func.func @atan_caller(%float: f32, %double: f64, %half: f16, %bfloat: bf16) -> (f32, f64, f16, bf16) {
// CHECK: %[[FLOAT_RESULT:.*]] = call @atanf(%[[FLOAT]]) : (f32) -> f32 // CHECK: %[[FLOAT_RESULT:.*]] = call @atanf(%[[FLOAT]]) : (f32) -> f32
▲ Show 20 Lines • Show All 269 Lines • ▼ Show 20 Lines
// CHECK: %[[VAL_17:.*]] = vector.insert %[[OUT1_F64]], %[[VAL_14]] [1] : f64 into vector<2xf64> // CHECK: %[[VAL_17:.*]] = vector.insert %[[OUT1_F64]], %[[VAL_14]] [1] : f64 into vector<2xf64>
// CHECK: return %[[VAL_11]], %[[VAL_17]] : vector<2xf32>, vector<2xf64> // CHECK: return %[[VAL_11]], %[[VAL_17]] : vector<2xf32>, vector<2xf64>
// CHECK: } // CHECK: }
func.func @tan_vec_caller(%float: vector<2xf32>, %double: vector<2xf64>) -> (vector<2xf32>, vector<2xf64>) { func.func @tan_vec_caller(%float: vector<2xf32>, %double: vector<2xf64>) -> (vector<2xf32>, vector<2xf64>) {
%float_result = math.tan %float : vector<2xf32> %float_result = math.tan %float : vector<2xf32>
%double_result = math.tan %double : vector<2xf64> %double_result = math.tan %double : vector<2xf64>
return %float_result, %double_result : vector<2xf32>, vector<2xf64> return %float_result, %double_result : vector<2xf32>, vector<2xf64>
} }
// CHECK-LABEL: func @floor_caller
// CHECK-SAME: %[[FLOAT:.*]]: f32
// CHECK-SAME: %[[DOUBLE:.*]]: f64
func.func @floor_caller(%float: f32, %double: f64) -> (f32, f64) {
// CHECK-DAG: %[[FLOAT_RESULT:.*]] = call @floorf(%[[FLOAT]]) : (f32) -> f32
%float_result = math.floor %float : f32
// CHECK-DAG: %[[DOUBLE_RESULT:.*]] = call @floor(%[[DOUBLE]]) : (f64) -> f64
%double_result = math.floor %double : f64
// CHECK: return %[[FLOAT_RESULT]], %[[DOUBLE_RESULT]]
return %float_result, %double_result : f32, f64
}
// CHECK-LABEL: func @ceil_caller
// CHECK-SAME: %[[FLOAT:.*]]: f32
// CHECK-SAME: %[[DOUBLE:.*]]: f64
func.func @ceil_caller(%float: f32, %double: f64) -> (f32, f64) {
// CHECK-DAG: %[[FLOAT_RESULT:.*]] = call @ceilf(%[[FLOAT]]) : (f32) -> f32
%float_result = math.ceil %float : f32
// CHECK-DAG: %[[DOUBLE_RESULT:.*]] = call @ceil(%[[DOUBLE]]) : (f64) -> f64
%double_result = math.ceil %double : f64
// CHECK: return %[[FLOAT_RESULT]], %[[DOUBLE_RESULT]]
return %float_result, %double_result : f32, f64
}
No newline at end of file
ftynseUnsubmitted
Not Done
ReplyInline Actions

Please add the newline.

ftynse: Please add the newline.