llvm/test/CodeGen/SystemZ/fp-mul-07.ll
Ulrich Weigand cf0fa9b9dd [SystemZ] Add CodeGen support for scalar f64 ops in vector registers
The z13 vector facility includes some instructions that operate only on the
high f64 in a v2f64, effectively extending the FP register set from 16
to 32 registers.  It's still better to use the old instructions if the
operands happen to fit though, since the older instructions have a shorter
encoding.

Based on a patch by Richard Sandiford.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@236524 91177308-0d34-0410-b5e6-96231b3b80d8
2015-05-05 19:28:34 +00:00

107 lines
3.4 KiB
LLVM

; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 \
; RUN: | FileCheck -check-prefix=CHECK -check-prefix=CHECK-SCALAR %s
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 \
; RUN: | FileCheck -check-prefix=CHECK -check-prefix=CHECK-VECTOR %s
declare double @llvm.fma.f64(double %f1, double %f2, double %f3)
define double @f1(double %f1, double %f2, double %acc) {
; CHECK-LABEL: f1:
; CHECK-SCALAR: madbr %f4, %f0, %f2
; CHECK-SCALAR: ldr %f0, %f4
; CHECK-VECTOR: wfmadb %f0, %f0, %f2, %f4
; CHECK: br %r14
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f2(double %f1, double *%ptr, double %acc) {
; CHECK-LABEL: f2:
; CHECK: madb %f2, %f0, 0(%r2)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f3(double %f1, double *%base, double %acc) {
; CHECK-LABEL: f3:
; CHECK: madb %f2, %f0, 4088(%r2)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 511
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f4(double %f1, double *%base, double %acc) {
; The important thing here is that we don't generate an out-of-range
; displacement. Other sequences besides this one would be OK.
;
; CHECK-LABEL: f4:
; CHECK: aghi %r2, 4096
; CHECK: madb %f2, %f0, 0(%r2)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 512
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f5(double %f1, double *%base, double %acc) {
; Here too the important thing is that we don't generate an out-of-range
; displacement. Other sequences besides this one would be OK.
;
; CHECK-LABEL: f5:
; CHECK: aghi %r2, -8
; CHECK: madb %f2, %f0, 0(%r2)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 -1
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f6(double %f1, double *%base, i64 %index, double %acc) {
; CHECK-LABEL: f6:
; CHECK: sllg %r1, %r3, 3
; CHECK: madb %f2, %f0, 0(%r1,%r2)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 %index
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f7(double %f1, double *%base, i64 %index, double %acc) {
; CHECK-LABEL: f7:
; CHECK: sllg %r1, %r3, 3
; CHECK: madb %f2, %f0, 4088({{%r1,%r2|%r2,%r1}})
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%index2 = add i64 %index, 511
%ptr = getelementptr double, double *%base, i64 %index2
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}
define double @f8(double %f1, double *%base, i64 %index, double %acc) {
; CHECK-LABEL: f8:
; CHECK: sllg %r1, %r3, 3
; CHECK: lay %r1, 4096({{%r1,%r2|%r2,%r1}})
; CHECK: madb %f2, %f0, 0(%r1)
; CHECK: ldr %f0, %f2
; CHECK: br %r14
%index2 = add i64 %index, 512
%ptr = getelementptr double, double *%base, i64 %index2
%f2 = load double , double *%ptr
%res = call double @llvm.fma.f64 (double %f1, double %f2, double %acc)
ret double %res
}