[NVPTX] Lower integer absolute value idiom to abs instruction.

Summary: Previously we lowered it literally, to shifts and xors.

Reviewers: tra

Subscribers: jholewinski, llvm-commits

Differential Revision: https://reviews.llvm.org/D28722

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292303 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/NVPTX/NVPTXInstrInfo.td b/lib/Target/NVPTX/NVPTXInstrInfo.td
index 179830817ed21313fd6270664134f0716d7f5f2e..ac7b7e0b9f470d99545aec1d3984edbcc7c3c776 100644 (file)
--- a/lib/Target/NVPTX/NVPTXInstrInfo.td
+++ b/lib/Target/NVPTX/NVPTXInstrInfo.td
@@ -516,6 +516,18 @@ defm UDIV : I3<"div.u", udiv>;
 defm SREM : I3<"rem.s", srem>;
 defm UREM : I3<"rem.u", urem>;
 
+// Integer absolute value.  NumBits should be one minus the bit width of RC.
+// This idiom implements the algorithm at
+// http://graphics.stanford.edu/~seander/bithacks.html#IntegerAbs.
+multiclass ABS<RegisterClass RC, int NumBits, string SizeName> {
+  def : NVPTXInst<(outs RC:$dst), (ins RC:$a),
+                  !strconcat("abs", SizeName, " \t$dst, $a;"),
+                  [(set RC:$dst, (xor (add (sra RC:$a, (i32 NumBits)), RC:$a),
+                                      (sra RC:$a, (i32 NumBits))))]>;
+}
+defm ABS_16 : ABS<Int16Regs, 15, ".s16">;
+defm ABS_32 : ABS<Int32Regs, 31, ".s32">;
+defm ABS_64 : ABS<Int64Regs, 63, ".s64">;
 
 //
 // Wide multiplication

diff --git a/test/CodeGen/NVPTX/idioms.ll b/test/CodeGen/NVPTX/idioms.ll
new file mode 100644 (file)
index 0000000..047325c
--- /dev/null
+++ b/test/CodeGen/NVPTX/idioms.ll
@@ -0,0 +1,31 @@
+; Check that various LLVM idioms get lowered to NVPTX as expected.
+
+; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s
+; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 | FileCheck %s
+
+; CHECK-LABEL: abs_i16(
+define i16 @abs_i16(i16 %a) {
+; CHECK: abs.s16
+  %neg = sub i16 0, %a
+  %abs.cond = icmp sge i16 %a, 0
+  %abs = select i1 %abs.cond, i16 %a, i16 %neg
+  ret i16 %abs
+}
+
+; CHECK-LABEL: abs_i32(
+define i32 @abs_i32(i32 %a) {
+; CHECK: abs.s32
+  %neg = sub i32 0, %a
+  %abs.cond = icmp sge i32 %a, 0
+  %abs = select i1 %abs.cond, i32 %a, i32 %neg
+  ret i32 %abs
+}
+
+; CHECK-LABEL: abs_i64(
+define i64 @abs_i64(i64 %a) {
+; CHECK: abs.s64
+  %neg = sub i64 0, %a
+  %abs.cond = icmp sge i64 %a, 0
+  %abs = select i1 %abs.cond, i64 %a, i64 %neg
+  ret i64 %abs
+}