[llvm] d4563bf - [NFC][InstCombine] Add some tests for open-coded round-up of an integer w/ power-of-2 target alignment
Roman Lebedev via llvm-commits
llvm-commits at lists.llvm.org
Wed Apr 27 07:28:36 PDT 2022
Author: Roman Lebedev
Date: 2022-04-27T17:24:42+03:00
New Revision: d4563bfeb940256384efce7dbb58502a83726170
URL: https://github.com/llvm/llvm-project/commit/d4563bfeb940256384efce7dbb58502a83726170
DIFF: https://github.com/llvm/llvm-project/commit/d4563bfeb940256384efce7dbb58502a83726170.diff
LOG: [NFC][InstCombine] Add some tests for open-coded round-up of an integer w/ power-of-2 target alignment
Added:
llvm/test/Transforms/InstCombine/integer-round-up-pow2-alignment.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/Transforms/InstCombine/integer-round-up-pow2-alignment.ll b/llvm/test/Transforms/InstCombine/integer-round-up-pow2-alignment.ll
new file mode 100644
index 000000000000..eeeb9fb1e622
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/integer-round-up-pow2-alignment.ll
@@ -0,0 +1,541 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=instcombine -S | FileCheck %s
+
+declare void @use.i8(i8)
+declare void @use.v2i4(<2 x i4>)
+declare void @use.i1(i1)
+
+declare void @llvm.assume(i1)
+
+; Basic pattern
+define i8 @t0(i8 %x) {
+; CHECK-LABEL: @t0(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Another alignment is fine
+define i8 @t1(i8 %x) {
+; CHECK-LABEL: @t1(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 31
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 32
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -32
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 31
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 32
+ %x.biased.highbits = and i8 %x.biased, -32
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Bias can be either the alignment or alignment-1
+define i8 @t2(i8 %x) {
+; CHECK-LABEL: @t2(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 15
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 15
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; select is commutative
+define i8 @t3_commutative(i8 %x) {
+; CHECK-LABEL: @t3_commutative(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_NOT_ZERO:%.*]] = icmp ne i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: call void @use.i1(i1 [[X_LOWBITS_ARE_NOT_ZERO]])
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_NOT_ZERO]], i8 [[X_BIASED_HIGHBITS]], i8 [[X]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.not.zero = icmp ne i8 %x.lowbits, 0
+ call void @use.i1(i1 %x.lowbits.are.not.zero)
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.not.zero, i8 %x.biased.highbits, i8 %x
+ ret i8 %x.roundedup
+}
+
+; Basic splat vector test
+define <2 x i8> @t4_splat(<2 x i8> %x) {
+; CHECK-LABEL: @t4_splat(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 15>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 16>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -16>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 15>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 16>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -16>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+
+; Splat-with-undef
+define <2 x i8> @t5_splat_undef_0b0001(<2 x i8> %x) {
+; CHECK-LABEL: @t5_splat_undef_0b0001(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 15>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 16>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 undef>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 15>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 16>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 undef>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t5_splat_undef_0b0010(<2 x i8> %x) {
+; CHECK-LABEL: @t5_splat_undef_0b0010(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 15>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 undef>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -16>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 15>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 undef>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -16>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t5_splat_undef_0b0100(<2 x i8> %x) {
+; CHECK-LABEL: @t5_splat_undef_0b0100(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 15>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], <i8 0, i8 undef>
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 16>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -16>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 15>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 undef>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 16>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -16>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t5_splat_undef_0b1000(<2 x i8> %x) {
+; CHECK-LABEL: @t5_splat_undef_0b1000(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 undef>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 16>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -16>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 undef>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 16>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -16>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+
+; Basic non-splat vector test
+define <2 x i8> @t6_nonsplat(<2 x i8> %x) {
+; CHECK-LABEL: @t6_nonsplat(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 31>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 32>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -32>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 31>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 32>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -32>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+
+; Even if the alignment (and masks) are splat, the bias could be non-splat
+define <2 x i8> @t7_nonsplat_bias(<2 x i8> %x) {
+; CHECK-LABEL: @t7_nonsplat_bias(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 15>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 15, i8 16>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -16>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 15>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 15, i8 16>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -16>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+
+; Splat-in-disguise vector tests
+define <2 x i8> @t8_nonsplat_masked_by_undef_0b0001(<2 x i8> %x) {
+; CHECK-LABEL: @t8_nonsplat_masked_by_undef_0b0001(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 31>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 32>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 undef>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 31>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 32>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 undef>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t8_nonsplat_masked_by_undef_0b0010(<2 x i8> %x) {
+; CHECK-LABEL: @t8_nonsplat_masked_by_undef_0b0010(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 31>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 undef>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -32>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 31>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 undef>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -32>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t8_nonsplat_masked_by_undef_0b0100(<2 x i8> %x) {
+; CHECK-LABEL: @t8_nonsplat_masked_by_undef_0b0100(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 31>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], <i8 0, i8 undef>
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 32>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -32>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 31>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 undef>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 32>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -32>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+define <2 x i8> @t8_nonsplat_masked_by_undef_0b1000(<2 x i8> %x) {
+; CHECK-LABEL: @t8_nonsplat_masked_by_undef_0b1000(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i8> [[X:%.*]], <i8 15, i8 undef>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i8> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i8> [[X]], <i8 16, i8 32>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i8> [[X_BIASED]], <i8 -16, i8 -32>
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i8> [[X]], <2 x i8> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i8> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i8> %x, <i8 15, i8 undef>
+ %x.lowbits.are.zero = icmp eq <2 x i8> %x.lowbits, <i8 0, i8 0>
+ %x.biased = add <2 x i8> %x, <i8 16, i8 32>
+ %x.biased.highbits = and <2 x i8> %x.biased, <i8 -16, i8 -32>
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i8> %x, <2 x i8> %x.biased.highbits
+ ret <2 x i8> %x.roundedup
+}
+
+; The X are
diff erent
+define i8 @n9_wrong_x0(i8 %x.0, i8 %x.1) {
+; CHECK-LABEL: @n9_wrong_x0(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X_0:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X_0]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X_1:%.*]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x.0, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x.0, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x.1, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+define i8 @n9_wrong_x1(i8 %x.0, i8 %x.1) {
+; CHECK-LABEL: @n9_wrong_x1(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X_0:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X_1:%.*]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X_0]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x.0, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x.1, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x.0, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+define i8 @n9_wrong_x2(i8 %x.0, i8 %x.1) {
+; CHECK-LABEL: @n9_wrong_x2(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X_1:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X_0:%.*]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X_0]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x.1, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x.0, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x.0, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Wrong low-bit mask
+define i8 @n10_wrong_low_bit_mask(i8 %x) {
+; CHECK-LABEL: @n10_wrong_low_bit_mask(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 31
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 31
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Wrong high-bit mask
+define i8 @n11_wrong_high_bit_mask(i8 %x) {
+; CHECK-LABEL: @n11_wrong_high_bit_mask(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -32
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -32
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Wrong bias
+define i8 @n12_wrong_bias(i8 %x) {
+; CHECK-LABEL: @n12_wrong_bias(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 32
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 32
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Wrong constants
+define i8 @n13_wrong_constants_alignment_is_not_power_of_two(i8 %x) {
+; CHECK-LABEL: @n13_wrong_constants_alignment_is_not_power_of_two(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 2
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 3
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -3
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 2
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 3
+ %x.biased.highbits = and i8 %x.biased, -3
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Comparison is not with zero
+define i8 @n14_wrong_comparison_constant(i8 %x) {
+; CHECK-LABEL: @n14_wrong_comparison_constant(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 1
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 1
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Wrong comparison
+define i8 @n15_wrong_comparison_predicate_and_constant(i8 %x) {
+; CHECK-LABEL: @n15_wrong_comparison_predicate_and_constant(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 14
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp ult i8 %x.lowbits, 2
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; %x.biased.highbits must not have other uses
+define i8 @n16_oneuse(i8 %x) {
+; CHECK-LABEL: @n16_oneuse(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 16
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: call void @use.i8(i8 [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 16
+ %x.biased.highbits = and i8 %x.biased, -16
+ call void @use.i8(i8 %x.biased.highbits)
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; But if bias is equal to low-bit mask, then we *could* just replace %x.roundedup with %x.biased.highbits
+define i8 @t17_oneuse(i8 %x) {
+; CHECK-LABEL: @t17_oneuse(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and i8 [[X:%.*]], 15
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq i8 [[X_LOWBITS]], 0
+; CHECK-NEXT: [[X_BIASED:%.*]] = add i8 [[X]], 15
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and i8 [[X_BIASED]], -16
+; CHECK-NEXT: call void @use.i8(i8 [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select i1 [[X_LOWBITS_ARE_ZERO]], i8 [[X]], i8 [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret i8 [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and i8 %x, 15
+ %x.lowbits.are.zero = icmp eq i8 %x.lowbits, 0
+ %x.biased = add i8 %x, 15
+ %x.biased.highbits = and i8 %x.biased, -16
+ call void @use.i8(i8 %x.biased.highbits)
+ %x.roundedup = select i1 %x.lowbits.are.zero, i8 %x, i8 %x.biased.highbits
+ ret i8 %x.roundedup
+}
+
+; Bias is equal to the alignment-1 (as opposed to alignment),
+; so we can just replace %x.roundedup with %x.biased.highbits
+define <2 x i4> @t18_replacement_0b0001(<2 x i4> %x) {
+; CHECK-LABEL: @t18_replacement_0b0001(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i4> [[X:%.*]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i4> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i4> [[X]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i4> [[X_BIASED]], <i4 -4, i4 undef>
+; CHECK-NEXT: call void @use.v2i4(<2 x i4> [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i4> [[X]], <2 x i4> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i4> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i4> %x, <i4 3, i4 3>
+ %x.lowbits.are.zero = icmp eq <2 x i4> %x.lowbits, <i4 0, i4 0>
+ %x.biased = add <2 x i4> %x, <i4 3, i4 3>
+ %x.biased.highbits = and <2 x i4> %x.biased, <i4 -4, i4 undef>
+ call void @use.v2i4(<2 x i4> %x.biased.highbits)
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i4> %x, <2 x i4> %x.biased.highbits
+ ret <2 x i4> %x.roundedup
+}
+define <2 x i4> @t18_replacement_0b0010(<2 x i4> %x) {
+; CHECK-LABEL: @t18_replacement_0b0010(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i4> [[X:%.*]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i4> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i4> [[X]], <i4 3, i4 undef>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i4> [[X_BIASED]], <i4 -4, i4 -4>
+; CHECK-NEXT: call void @use.v2i4(<2 x i4> [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i4> [[X]], <2 x i4> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i4> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i4> %x, <i4 3, i4 3>
+ %x.lowbits.are.zero = icmp eq <2 x i4> %x.lowbits, <i4 0, i4 0>
+ %x.biased = add <2 x i4> %x, <i4 3, i4 undef>
+ %x.biased.highbits = and <2 x i4> %x.biased, <i4 -4, i4 -4>
+ call void @use.v2i4(<2 x i4> %x.biased.highbits)
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i4> %x, <2 x i4> %x.biased.highbits
+ ret <2 x i4> %x.roundedup
+}
+define <2 x i4> @t18_replacement_0b0100(<2 x i4> %x) {
+; CHECK-LABEL: @t18_replacement_0b0100(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i4> [[X:%.*]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i4> [[X_LOWBITS]], <i4 0, i4 undef>
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i4> [[X]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i4> [[X_BIASED]], <i4 -4, i4 -4>
+; CHECK-NEXT: call void @use.v2i4(<2 x i4> [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i4> [[X]], <2 x i4> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i4> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i4> %x, <i4 3, i4 3>
+ %x.lowbits.are.zero = icmp eq <2 x i4> %x.lowbits, <i4 0, i4 undef>
+ %x.biased = add <2 x i4> %x, <i4 3, i4 3>
+ %x.biased.highbits = and <2 x i4> %x.biased, <i4 -4, i4 -4>
+ call void @use.v2i4(<2 x i4> %x.biased.highbits)
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i4> %x, <2 x i4> %x.biased.highbits
+ ret <2 x i4> %x.roundedup
+}
+define <2 x i4> @t18_replacement_0b1000(<2 x i4> %x) {
+; CHECK-LABEL: @t18_replacement_0b1000(
+; CHECK-NEXT: [[X_LOWBITS:%.*]] = and <2 x i4> [[X:%.*]], <i4 3, i4 undef>
+; CHECK-NEXT: [[X_LOWBITS_ARE_ZERO:%.*]] = icmp eq <2 x i4> [[X_LOWBITS]], zeroinitializer
+; CHECK-NEXT: [[X_BIASED:%.*]] = add <2 x i4> [[X]], <i4 3, i4 3>
+; CHECK-NEXT: [[X_BIASED_HIGHBITS:%.*]] = and <2 x i4> [[X_BIASED]], <i4 -4, i4 -4>
+; CHECK-NEXT: call void @use.v2i4(<2 x i4> [[X_BIASED_HIGHBITS]])
+; CHECK-NEXT: [[X_ROUNDEDUP:%.*]] = select <2 x i1> [[X_LOWBITS_ARE_ZERO]], <2 x i4> [[X]], <2 x i4> [[X_BIASED_HIGHBITS]]
+; CHECK-NEXT: ret <2 x i4> [[X_ROUNDEDUP]]
+;
+ %x.lowbits = and <2 x i4> %x, <i4 3, i4 undef>
+ %x.lowbits.are.zero = icmp eq <2 x i4> %x.lowbits, <i4 0, i4 0>
+ %x.biased = add <2 x i4> %x, <i4 3, i4 3>
+ %x.biased.highbits = and <2 x i4> %x.biased, <i4 -4, i4 -4>
+ call void @use.v2i4(<2 x i4> %x.biased.highbits)
+ %x.roundedup = select <2 x i1> %x.lowbits.are.zero, <2 x i4> %x, <2 x i4> %x.biased.highbits
+ ret <2 x i4> %x.roundedup
+}
More information about the llvm-commits
mailing list