llvm-6502/test/Transforms/InstCombine/fmul.ll

73 lines
1.8 KiB
LLVM
Raw Normal View History

1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; RUN: opt -S -instcombine < %s | FileCheck %s
; (-0.0 - X) * C => X * -C
define float @test1(float %x) {
%sub = fsub float -0.000000e+00, %x
%mul = fmul float %sub, 2.0e+1
ret float %mul
; CHECK-LABEL: @test1(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: fmul float %x, -2.000000e+01
}
; (0.0 - X) * C => X * -C
define float @test2(float %x) {
%sub = fsub nsz float 0.000000e+00, %x
%mul = fmul float %sub, 2.0e+1
ret float %mul
; CHECK-LABEL: @test2(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: fmul float %x, -2.000000e+01
}
; (-0.0 - X) * (-0.0 - Y) => X * Y
define float @test3(float %x, float %y) {
%sub1 = fsub float -0.000000e+00, %x
%sub2 = fsub float -0.000000e+00, %y
%mul = fmul float %sub1, %sub2
ret float %mul
; CHECK-LABEL: @test3(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: fmul float %x, %y
}
; (0.0 - X) * (0.0 - Y) => X * Y
define float @test4(float %x, float %y) {
%sub1 = fsub nsz float 0.000000e+00, %x
%sub2 = fsub nsz float 0.000000e+00, %y
%mul = fmul float %sub1, %sub2
ret float %mul
; CHECK-LABEL: @test4(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: fmul float %x, %y
}
; (-0.0 - X) * Y => -0.0 - (X * Y)
define float @test5(float %x, float %y) {
%sub1 = fsub float -0.000000e+00, %x
%mul = fmul float %sub1, %y
ret float %mul
; CHECK-LABEL: @test5(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: %1 = fmul float %x, %y
; CHECK: %mul = fsub float -0.000000e+00, %1
}
; (0.0 - X) * Y => 0.0 - (X * Y)
define float @test6(float %x, float %y) {
%sub1 = fsub nsz float 0.000000e+00, %x
%mul = fmul float %sub1, %y
ret float %mul
; CHECK-LABEL: @test6(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: %1 = fmul float %x, %y
; CHECK: %mul = fsub float -0.000000e+00, %1
}
; "(-0.0 - X) * Y => -0.0 - (X * Y)" is disabled if expression "-0.0 - X"
; has multiple uses.
define float @test7(float %x, float %y) {
%sub1 = fsub float -0.000000e+00, %x
%mul = fmul float %sub1, %y
%mul2 = fmul float %mul, %sub1
ret float %mul2
; CHECK-LABEL: @test7(
1. Hoist minus sign as high as possible in an attempt to reveal some optimization opportunities (in the enclosing supper-expressions). rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y) if expression "-0.0 - X" has only one reference. rule 2. (0.0 - X ) * Y => -0.0 - (X * Y) if expression "0.0 - X" has only one reference, and the instruction is marked "noSignedZero". 2. Eliminate negation (The compiler was already able to handle these opt if the 0.0s are replaced with -0.0.) rule 3: (0.0 - X) * (0.0 - Y) => X * Y rule 4: (0.0 - X) * C => X * -C if the expr is flagged "noSignedZero". 3. Rule 5: (X*Y) * X => (X*X) * Y if X!=Y and the expression is flagged with "UnsafeAlgebra". The purpose of this transformation is two-fold: a) to form a power expression (of X). b) potentially shorten the critical path: After transformation, the latency of the instruction Y is amortized by the expression of X*X, and therefore Y is in a "less critical" position compared to what it was before the transformation. 4. Remove the InstCombine code about simplifiying "X * select". The reasons are following: a) The "select" is somewhat architecture-dependent, therefore the higher level optimizers are not able to precisely predict if the simplification really yields any performance improvement or not. b) The "select" operator is bit complicate, and tends to obscure optimization opportunities. It is btter to keep it as low as possible in expr tree, and let CodeGen to tackle the optimization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-15 21:09:32 +00:00
; CHECK: fsub float -0.000000e+00, %x
}