mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 19:23:23 +01:00
bc658bf60a
This adds support for the new 128-bit vector float instructions of z14. Note that these instructions actually only operate on the f128 type, since only each 128-bit vector register can hold only one 128-bit float value. However, this is still preferable to the legacy 128-bit float instructions, since those operate on pairs of floating-point registers (so we can hold at most 8 values in registers), while the new instructions use single vector registers (so we hold up to 32 value in registers). Adding support includes: - Enabling the instructions for the assembler/disassembler. - CodeGen for the instructions. This includes allocating the f128 type now to the VR128BitRegClass instead of FP128BitRegClass. - Scheduler description support for the instructions. Note that for a small number of operations, we have no new vector instructions (like integer <-> 128-bit float conversions), and so we use the legacy instruction and then reformat the operand (i.e. copy between a pair of floating-point registers and a vector register). llvm-svn: 308196
33 lines
914 B
LLVM
33 lines
914 B
LLVM
; Test 128-bit floating-point multiplication on z14.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z14 | FileCheck %s
|
|
|
|
define void @f1(fp128 *%ptr1, fp128 *%ptr2) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK-DAG: vl [[REG1:%v[0-9]+]], 0(%r2)
|
|
; CHECK-DAG: vl [[REG2:%v[0-9]+]], 0(%r3)
|
|
; CHECK: wfmxb [[RES:%v[0-9]+]], [[REG1]], [[REG2]]
|
|
; CHECK: vst [[RES]], 0(%r2)
|
|
; CHECK: br %r14
|
|
%f1 = load fp128, fp128 *%ptr1
|
|
%f2 = load fp128, fp128 *%ptr2
|
|
%sum = fmul fp128 %f1, %f2
|
|
store fp128 %sum, fp128 *%ptr1
|
|
ret void
|
|
}
|
|
|
|
define void @f2(double %f1, double %f2, fp128 *%dst) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK-DAG: wflld [[REG1:%v[0-9]+]], %f0
|
|
; CHECK-DAG: wflld [[REG2:%v[0-9]+]], %f2
|
|
; CHECK: wfmxb [[RES:%v[0-9]+]], [[REG1]], [[REG2]]
|
|
; CHECK: vst [[RES]], 0(%r2)
|
|
; CHECK: br %r14
|
|
%f1x = fpext double %f1 to fp128
|
|
%f2x = fpext double %f2 to fp128
|
|
%res = fmul fp128 %f1x, %f2x
|
|
store fp128 %res, fp128 *%dst
|
|
ret void
|
|
}
|
|
|