mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-17 21:36:39 +00:00
80267f8148
The recently announced IBM z17 processor implements the architecture already supported as "arch15" in LLVM. This patch adds support for "z17" as an alternate architecture name for arch15. This patch also add the scheduler description for the z17 processor, provided by Jonas Paulsson.
160 lines
4.7 KiB
LLVM
160 lines
4.7 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
|
|
; Test high-part i128->i256 multiplications on z17.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z17 | FileCheck %s
|
|
|
|
; Multiply high signed.
|
|
define i128 @f1(i128 %a, i128 %b) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vmhq %v0, %v1, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = sext i128 %a to i256
|
|
%extb = sext i128 %b to i256
|
|
%extres = mul i256 %exta, %extb
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply high unsigned.
|
|
define i128 @f2(i128 %a, i128 %b) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vmlhq %v0, %v1, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = zext i128 %a to i256
|
|
%extb = zext i128 %b to i256
|
|
%extres = mul i256 %exta, %extb
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high signed.
|
|
define i128 @f3(i128 %a, i128 %b, i128 %add) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vl %v0, 0(%r5), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r3), 3
|
|
; CHECK-NEXT: vmahq %v0, %v2, %v1, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = sext i128 %a to i256
|
|
%extb = sext i128 %b to i256
|
|
%extadd = sext i128 %add to i256
|
|
%extmul = mul i256 %exta, %extb
|
|
%extres = add i256 %extmul, %extadd
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high unsigned.
|
|
define i128 @f4(i128 %a, i128 %b, i128 %add) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vl %v0, 0(%r5), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r3), 3
|
|
; CHECK-NEXT: vmalhq %v0, %v2, %v1, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = zext i128 %a to i256
|
|
%extb = zext i128 %b to i256
|
|
%extadd = zext i128 %add to i256
|
|
%extmul = mul i256 %exta, %extb
|
|
%extres = add i256 %extmul, %extadd
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high signed with immediate operand to multiply.
|
|
define i128 @f5(i128 %a, i128 %add) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: larl %r1, .LCPI4_0
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r1), 3
|
|
; CHECK-NEXT: vmahq %v0, %v1, %v2, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = sext i128 %a to i256
|
|
%extadd = sext i128 %add to i256
|
|
%extmul = mul i256 %exta, 12345
|
|
%extres = add i256 %extmul, %extadd
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high unsigned with immediate operand to multiply.
|
|
define i128 @f6(i128 %a, i128 %add) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: larl %r1, .LCPI5_0
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r1), 3
|
|
; CHECK-NEXT: vmalhq %v0, %v1, %v2, %v0
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = zext i128 %a to i256
|
|
%extadd = zext i128 %add to i256
|
|
%extmul = mul i256 %exta, 12345
|
|
%extres = add i256 %extmul, %extadd
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high signed with immediate operand to addition.
|
|
define i128 @f7(i128 %a, i128 %b) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: larl %r1, .LCPI6_0
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r1), 3
|
|
; CHECK-NEXT: vmahq %v0, %v1, %v0, %v2
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = sext i128 %a to i256
|
|
%extb = sext i128 %b to i256
|
|
%extmul = mul i256 %exta, %extb
|
|
%extres = add i256 %extmul, 12345
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|
|
; Multiply-and-add high unsigned with immediate operand to addition.
|
|
define i128 @f8(i128 %a, i128 %b) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: larl %r1, .LCPI7_0
|
|
; CHECK-NEXT: vl %v0, 0(%r4), 3
|
|
; CHECK-NEXT: vl %v1, 0(%r3), 3
|
|
; CHECK-NEXT: vl %v2, 0(%r1), 3
|
|
; CHECK-NEXT: vmalhq %v0, %v1, %v0, %v2
|
|
; CHECK-NEXT: vst %v0, 0(%r2), 3
|
|
; CHECK-NEXT: br %r14
|
|
%exta = zext i128 %a to i256
|
|
%extb = zext i128 %b to i256
|
|
%extmul = mul i256 %exta, %extb
|
|
%extres = add i256 %extmul, 12345
|
|
%shiftres = lshr i256 %extres, 128
|
|
%res = trunc i256 %shiftres to i128
|
|
ret i128 %res
|
|
}
|
|
|