caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
ec056f54589d937ae3f74d795b94de886cbb7ec7
clang-p2996/llvm/test/CodeGen/RISCV/rvv/vreductions-fp-sdnode.ll
Philip Reames 859c871184 [RISCV] Default to MicroOpBufferSize = 1 for scheduling purposes (#126608) 
This change introduces a default schedule model for the RISCV target
which leaves everything unchanged except the MicroOpBufferSize. The
default value of this flag in NoSched is 0. Both configurations
represent in order cores (i.e. no reorder window), the difference
between them comes down to whether heuristics other than latency are
allowed to apply. (Implementation details below)

I left the processor models which explicitly set MicroOpBufferSize=0
unchanged in this patch, but strongly suspect we should change those
too. Honestly, I think the LLVM wide default for this flag should be
changed, but don't have the energy to manage the updates for all
targets.

Implementation wise, the effect of this change is that schedule units
which are ready to run *except that* one of their predecessors may not
have completed yet are added to the Available list, not the Pending one.
The result of this is that it becomes possible to chose to schedule a
node before it's ready cycle if the heuristics prefer. This is
essentially chosing to insert a resource stall instead of e.g.
increasing register pressure.

Note that I was initially concerned there might be a correctness aspect
(as in some kind of exposed pipeline design), but the generic scheduler
doesn't seem to know how to insert noop instructions. Without that, a
program wouldn't be guaranteed to schedule on an exposed pipeline
depending on the program and schedule model in question.

The effect of this is that we sometimes prefer register pressure in
codegen results. This is mostly churn (or small wins) on scalar because
we have many more registers, but is of major importance on vector -
particularly high LMUL - because we effectively have many fewer
registers and the relative cost of spilling is much higher. This is a
significant improvement on high LMUL code quality for default rva23u
configurations - or any non -mcpu vector configuration for that matter.

Fixes #107532
2025-02-12 12:31:39 -08:00

1041 lines
37 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+m -target-abi=ilp32d \
; RUN: -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+m -target-abi=lp64d \
; RUN: -verify-machineinstrs < %s | FileCheck %s
declare half @llvm.vector.reduce.fadd.nxv1f16(half, <vscale x 1 x half>)
define half @vreduce_fadd_nxv1f16(<vscale x 1 x half> %v, half %s) {
; CHECK-LABEL: vreduce_fadd_nxv1f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc half @llvm.vector.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v)
ret half %red
}
define half @vreduce_ord_fadd_nxv1f16(<vscale x 1 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv1f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fadd.nxv2f16(half, <vscale x 2 x half>)
define half @vreduce_fadd_nxv2f16(<vscale x 2 x half> %v, half %s) {
; CHECK-LABEL: vreduce_fadd_nxv2f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc half @llvm.vector.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v)
ret half %red
}
define half @vreduce_ord_fadd_nxv2f16(<vscale x 2 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv2f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fadd.nxv4f16(half, <vscale x 4 x half>)
define half @vreduce_fadd_nxv4f16(<vscale x 4 x half> %v, half %s) {
; CHECK-LABEL: vreduce_fadd_nxv4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc half @llvm.vector.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v)
ret half %red
}
define half @vreduce_ord_fadd_nxv4f16(<vscale x 4 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v)
ret half %red
}
declare float @llvm.vector.reduce.fadd.nxv1f32(float, <vscale x 1 x float>)
define float @vreduce_fadd_nxv1f32(<vscale x 1 x float> %v, float %s) {
; CHECK-LABEL: vreduce_fadd_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_ord_fadd_nxv1f32(<vscale x 1 x float> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_fwadd_nxv1f32(<vscale x 1 x half> %v, float %s) {
; CHECK-LABEL: vreduce_fwadd_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 1 x half> %v to <vscale x 1 x float>
%red = call reassoc float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %e)
ret float %red
}
define float @vreduce_ord_fwadd_nxv1f32(<vscale x 1 x half> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 1 x half> %v to <vscale x 1 x float>
%red = call float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %e)
ret float %red
}
declare float @llvm.vector.reduce.fadd.nxv2f32(float, <vscale x 2 x float>)
define float @vreduce_fadd_nxv2f32(<vscale x 2 x float> %v, float %s) {
; CHECK-LABEL: vreduce_fadd_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v)
ret float %red
}
define float @vreduce_ord_fadd_nxv2f32(<vscale x 2 x float> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v)
ret float %red
}
define float @vreduce_fwadd_nxv2f32(<vscale x 2 x half> %v, float %s) {
; CHECK-LABEL: vreduce_fwadd_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 2 x half> %v to <vscale x 2 x float>
%red = call reassoc float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %e)
ret float %red
}
define float @vreduce_ord_fwadd_nxv2f32(<vscale x 2 x half> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 2 x half> %v to <vscale x 2 x float>
%red = call float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %e)
ret float %red
}
declare float @llvm.vector.reduce.fadd.nxv4f32(float, <vscale x 4 x float>)
define float @vreduce_fadd_nxv4f32(<vscale x 4 x float> %v, float %s) {
; CHECK-LABEL: vreduce_fadd_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v)
ret float %red
}
define float @vreduce_ord_fadd_nxv4f32(<vscale x 4 x float> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v)
ret float %red
}
define float @vreduce_fwadd_nxv4f32(<vscale x 4 x half> %v, float %s) {
; CHECK-LABEL: vreduce_fwadd_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, m2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 4 x half> %v to <vscale x 4 x float>
%red = call reassoc float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %e)
ret float %red
}
define float @vreduce_ord_fwadd_nxv4f32(<vscale x 4 x half> %v, float %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e32, m2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 4 x half> %v to <vscale x 4 x float>
%red = call float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %e)
ret float %red
}
declare double @llvm.vector.reduce.fadd.nxv1f64(double, <vscale x 1 x double>)
define double @vreduce_fadd_nxv1f64(<vscale x 1 x double> %v, double %s) {
; CHECK-LABEL: vreduce_fadd_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_ord_fadd_nxv1f64(<vscale x 1 x double> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_fwadd_nxv1f64(<vscale x 1 x float> %v, double %s) {
; CHECK-LABEL: vreduce_fwadd_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 1 x float> %v to <vscale x 1 x double>
%red = call reassoc double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %e)
ret double %red
}
define double @vreduce_ord_fwadd_nxv1f64(<vscale x 1 x float> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 1 x float> %v to <vscale x 1 x double>
%red = call double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %e)
ret double %red
}
declare double @llvm.vector.reduce.fadd.nxv2f64(double, <vscale x 2 x double>)
define double @vreduce_fadd_nxv2f64(<vscale x 2 x double> %v, double %s) {
; CHECK-LABEL: vreduce_fadd_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v)
ret double %red
}
define double @vreduce_ord_fadd_nxv2f64(<vscale x 2 x double> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v)
ret double %red
}
define double @vreduce_fwadd_nxv2f64(<vscale x 2 x float> %v, double %s) {
; CHECK-LABEL: vreduce_fwadd_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e64, m2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 2 x float> %v to <vscale x 2 x double>
%red = call reassoc double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %e)
ret double %red
}
define double @vreduce_ord_fwadd_nxv2f64(<vscale x 2 x float> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v9
; CHECK-NEXT: vsetvli zero, zero, e64, m2, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 2 x float> %v to <vscale x 2 x double>
%red = call double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %e)
ret double %red
}
declare double @llvm.vector.reduce.fadd.nxv4f64(double, <vscale x 4 x double>)
define double @vreduce_fadd_nxv4f64(<vscale x 4 x double> %v, double %s) {
; CHECK-LABEL: vreduce_fadd_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v)
ret double %red
}
define double @vreduce_ord_fadd_nxv4f64(<vscale x 4 x double> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vfredosum.vs v8, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v)
ret double %red
}
define double @vreduce_fwadd_nxv4f64(<vscale x 4 x float> %v, double %s) {
; CHECK-LABEL: vreduce_fwadd_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfwredusum.vs v8, v8, v10
; CHECK-NEXT: vsetvli zero, zero, e64, m4, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 4 x float> %v to <vscale x 4 x double>
%red = call reassoc double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %e)
ret double %red
}
define double @vreduce_ord_fwadd_nxv4f64(<vscale x 4 x float> %v, double %s) {
; CHECK-LABEL: vreduce_ord_fwadd_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfwredosum.vs v8, v8, v10
; CHECK-NEXT: vsetvli zero, zero, e64, m4, ta, ma
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%e = fpext <vscale x 4 x float> %v to <vscale x 4 x double>
%red = call double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %e)
ret double %red
}
declare half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half>)
define half @vreduce_fmin_nxv1f16(<vscale x 1 x half> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
define half @vreduce_fmin_nxv1f16_nonans(<vscale x 1 x half> %v) #0 {
; CHECK-LABEL: vreduce_fmin_nxv1f16_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
define half @vreduce_fmin_nxv1f16_nonans_noinfs(<vscale x 1 x half> %v) #1 {
; CHECK-LABEL: vreduce_fmin_nxv1f16_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmin.nxv2f16(<vscale x 2 x half>)
define half @vreduce_fmin_nxv2f16(<vscale x 2 x half> %v) {
; CHECK-LABEL: vreduce_fmin_nxv2f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmin.nxv2f16(<vscale x 2 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmin.nxv4f16(<vscale x 4 x half>)
define half @vreduce_fmin_nxv4f16(<vscale x 4 x half> %v) {
; CHECK-LABEL: vreduce_fmin_nxv4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmin.nxv4f16(<vscale x 4 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmin.nxv64f16(<vscale x 64 x half>)
define half @vreduce_fmin_nxv64f16(<vscale x 64 x half> %v) {
; CHECK-LABEL: vreduce_fmin_nxv64f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, ma
; CHECK-NEXT: vfmin.vv v8, v8, v16
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmin.nxv64f16(<vscale x 64 x half> %v)
ret half %red
}
declare float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float>)
define float @vreduce_fmin_nxv1f32(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_fmin_nxv1f32_nonans(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f32_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_fmin_nxv1f32_nonans_noinfs(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f32_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmin.nxv2f32(<vscale x 2 x float>)
define float @vreduce_fmin_nxv2f32(<vscale x 2 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmin.nxv2f32(<vscale x 2 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmin.nxv4f32(<vscale x 4 x float>)
define float @vreduce_fmin_nxv4f32(<vscale x 4 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmin.nxv4f32(<vscale x 4 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float>)
define float @vreduce_fmin_nxv32f32(<vscale x 32 x float> %v) {
; CHECK-LABEL: vreduce_fmin_nxv32f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m8, ta, ma
; CHECK-NEXT: vfmin.vv v8, v8, v16
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float> %v)
ret float %red
}
declare double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double>)
define double @vreduce_fmin_nxv1f64(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_fmin_nxv1f64_nonans(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f64_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_fmin_nxv1f64_nonans_noinfs(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv1f64_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmin.nxv2f64(<vscale x 2 x double>)
define double @vreduce_fmin_nxv2f64(<vscale x 2 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmin.nxv2f64(<vscale x 2 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmin.nxv4f64(<vscale x 4 x double>)
define double @vreduce_fmin_nxv4f64(<vscale x 4 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, ma
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmin.nxv4f64(<vscale x 4 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmin.nxv16f64(<vscale x 16 x double>)
define double @vreduce_fmin_nxv16f64(<vscale x 16 x double> %v) {
; CHECK-LABEL: vreduce_fmin_nxv16f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m8, ta, ma
; CHECK-NEXT: vfmin.vv v8, v8, v16
; CHECK-NEXT: vfredmin.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmin.nxv16f64(<vscale x 16 x double> %v)
ret double %red
}
declare half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half>)
define half @vreduce_fmax_nxv1f16(<vscale x 1 x half> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
define half @vreduce_fmax_nxv1f16_nonans(<vscale x 1 x half> %v) #0 {
; CHECK-LABEL: vreduce_fmax_nxv1f16_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
define half @vreduce_fmax_nxv1f16_nonans_noinfs(<vscale x 1 x half> %v) #1 {
; CHECK-LABEL: vreduce_fmax_nxv1f16_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmax.nxv2f16(<vscale x 2 x half>)
define half @vreduce_fmax_nxv2f16(<vscale x 2 x half> %v) {
; CHECK-LABEL: vreduce_fmax_nxv2f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmax.nxv2f16(<vscale x 2 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmax.nxv4f16(<vscale x 4 x half>)
define half @vreduce_fmax_nxv4f16(<vscale x 4 x half> %v) {
; CHECK-LABEL: vreduce_fmax_nxv4f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmax.nxv4f16(<vscale x 4 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half>)
define half @vreduce_fmax_nxv64f16(<vscale x 64 x half> %v) {
; CHECK-LABEL: vreduce_fmax_nxv64f16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, ma
; CHECK-NEXT: vfmax.vv v8, v8, v16
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half> %v)
ret half %red
}
declare float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float>)
define float @vreduce_fmax_nxv1f32(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_fmax_nxv1f32_nonans(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f32_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
define float @vreduce_fmax_nxv1f32_nonans_noinfs(<vscale x 1 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f32_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmax.nxv2f32(<vscale x 2 x float>)
define float @vreduce_fmax_nxv2f32(<vscale x 2 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmax.nxv2f32(<vscale x 2 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmax.nxv4f32(<vscale x 4 x float>)
define float @vreduce_fmax_nxv4f32(<vscale x 4 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmax.nxv4f32(<vscale x 4 x float> %v)
ret float %red
}
declare float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float>)
define float @vreduce_fmax_nxv32f32(<vscale x 32 x float> %v) {
; CHECK-LABEL: vreduce_fmax_nxv32f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, m8, ta, ma
; CHECK-NEXT: vfmax.vv v8, v8, v16
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float> %v)
ret float %red
}
declare double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double>)
define double @vreduce_fmax_nxv1f64(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_fmax_nxv1f64_nonans(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f64_nonans:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
define double @vreduce_fmax_nxv1f64_nonans_noinfs(<vscale x 1 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv1f64_nonans_noinfs:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call nnan ninf double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmax.nxv2f64(<vscale x 2 x double>)
define double @vreduce_fmax_nxv2f64(<vscale x 2 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmax.nxv2f64(<vscale x 2 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmax.nxv4f64(<vscale x 4 x double>)
define double @vreduce_fmax_nxv4f64(<vscale x 4 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, ma
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmax.nxv4f64(<vscale x 4 x double> %v)
ret double %red
}
declare double @llvm.vector.reduce.fmax.nxv16f64(<vscale x 16 x double>)
define double @vreduce_fmax_nxv16f64(<vscale x 16 x double> %v) {
; CHECK-LABEL: vreduce_fmax_nxv16f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e64, m8, ta, ma
; CHECK-NEXT: vfmax.vv v8, v8, v16
; CHECK-NEXT: vfredmax.vs v8, v8, v8
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call double @llvm.vector.reduce.fmax.nxv16f64(<vscale x 16 x double> %v)
ret double %red
}
define float @vreduce_nsz_fadd_nxv1f32(<vscale x 1 x float> %v, float %s) {
; CHECK-LABEL: vreduce_nsz_fadd_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vfredusum.vs v8, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v8
; CHECK-NEXT: ret
%red = call reassoc nsz float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v)
ret float %red
}
; Test Widen VECREDUCE_SEQ_FADD
declare half @llvm.vector.reduce.fadd.nxv3f16(half, <vscale x 3 x half>)
define half @vreduce_ord_fadd_nxv3f16(<vscale x 3 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv3f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: slli a1, a0, 1
; CHECK-NEXT: add a0, a1, a0
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; CHECK-NEXT: vfredosum.vs v9, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v9
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv3f16(half %s, <vscale x 3 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fadd.nxv6f16(half, <vscale x 6 x half>)
define half @vreduce_ord_fadd_nxv6f16(<vscale x 6 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv6f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: srli a1, a0, 3
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: sub a0, a0, a1
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; CHECK-NEXT: vfredosum.vs v10, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v10
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv6f16(half %s, <vscale x 6 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fadd.nxv10f16(half, <vscale x 10 x half>)
define half @vreduce_ord_fadd_nxv10f16(<vscale x 10 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv10f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: li a1, 10
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: mul a0, a0, a1
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; CHECK-NEXT: vfredosum.vs v12, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v12
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv10f16(half %s, <vscale x 10 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fadd.nxv12f16(half, <vscale x 12 x half>)
define half @vreduce_ord_fadd_nxv12f16(<vscale x 12 x half> %v, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_nxv12f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: slli a1, a0, 2
; CHECK-NEXT: slli a0, a0, 4
; CHECK-NEXT: sub a0, a0, a1
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; CHECK-NEXT: vfredosum.vs v12, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v12
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fadd.nxv12f16(half %s, <vscale x 12 x half> %v)
ret half %red
}
; Test Widen vector reduce type (fadd/fmin/fmax)
define half @vreduce_fadd_nxv3f16(<vscale x 3 x half> %v, half %s) {
; CHECK-LABEL: vreduce_fadd_nxv3f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v9, fa0
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: slli a1, a0, 1
; CHECK-NEXT: add a0, a1, a0
; CHECK-NEXT: lui a1, 1048568
; CHECK-NEXT: vmv.s.x v10, a1
; CHECK-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; CHECK-NEXT: vfredusum.vs v10, v8, v9
; CHECK-NEXT: vfmv.f.s fa0, v10
; CHECK-NEXT: ret
%red = call reassoc half @llvm.vector.reduce.fadd.nxv3f16(half %s, <vscale x 3 x half> %v)
ret half %red
}
define half @vreduce_fadd_nxv6f16(<vscale x 6 x half> %v, half %s) {
; CHECK-LABEL: vreduce_fadd_nxv6f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: srli a1, a0, 3
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: sub a0, a0, a1
; CHECK-NEXT: lui a1, 1048568
; CHECK-NEXT: vmv.s.x v11, a1
; CHECK-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; CHECK-NEXT: vfredusum.vs v11, v8, v10
; CHECK-NEXT: vfmv.f.s fa0, v11
; CHECK-NEXT: ret
%red = call reassoc half @llvm.vector.reduce.fadd.nxv6f16(half %s, <vscale x 6 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmin.nxv10f16(<vscale x 10 x half>)
define half @vreduce_fmin_nxv10f16(<vscale x 10 x half> %v) {
; CHECK-LABEL: vreduce_fmin_nxv10f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: li a1, 10
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: mul a0, a0, a1
; CHECK-NEXT: lui a1, %hi(.LCPI73_0)
; CHECK-NEXT: addi a1, a1, %lo(.LCPI73_0)
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vle16.v v12, (a1)
; CHECK-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; CHECK-NEXT: vfredmin.vs v12, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v12
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmin.nxv10f16(<vscale x 10 x half> %v)
ret half %red
}
declare half @llvm.vector.reduce.fmax.nxv12f16(<vscale x 12 x half>)
define half @vreduce_fmax_nxv12f16(<vscale x 12 x half> %v) {
; CHECK-LABEL: vreduce_fmax_nxv12f16:
; CHECK: # %bb.0:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: srli a0, a0, 3
; CHECK-NEXT: slli a1, a0, 2
; CHECK-NEXT: slli a0, a0, 4
; CHECK-NEXT: sub a0, a0, a1
; CHECK-NEXT: li a1, -512
; CHECK-NEXT: vsetivli zero, 1, e16, m1, ta, ma
; CHECK-NEXT: vmv.s.x v12, a1
; CHECK-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; CHECK-NEXT: vfredmax.vs v12, v8, v12
; CHECK-NEXT: vfmv.f.s fa0, v12
; CHECK-NEXT: ret
%red = call half @llvm.vector.reduce.fmax.nxv12f16(<vscale x 12 x half> %v)
ret half %red
}
Reference in New Issue View Git Blame Copy Permalink