[llvm] [RISCV] Re-model RVV comparison instructions (PR #88868)

[Pengcheng Wang via llvm-commits]

================
@@ -605,21 +593,19 @@ declare <8 x double> @llvm.vp.floor.v8f64(<8 x double>, <8 x i1>, i32)
 define <8 x double> @vp_floor_v8f64(<8 x double> %va, <8 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vp_floor_v8f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vmv1r.v v12, v0
 ; CHECK-NEXT:    lui a1, %hi(.LCPI20_0)
 ; CHECK-NEXT:    fld fa5, %lo(.LCPI20_0)(a1)
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
-; CHECK-NEXT:    vfabs.v v16, v8, v0.t
+; CHECK-NEXT:    vfabs.v v12, v8, v0.t
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m4, ta, mu
-; CHECK-NEXT:    vmflt.vf v12, v16, fa5, v0.t
+; CHECK-NEXT:    vmflt.vf v0, v12, fa5, v0.t
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-NEXT:    fsrmi a0, 2
-; CHECK-NEXT:    vmv1r.v v0, v12
-; CHECK-NEXT:    vfcvt.x.f.v v16, v8, v0.t
+; CHECK-NEXT:    vfcvt.x.f.v v12, v8, v0.t
----------------
wangpc-pp wrote:

An example of improvement here.

https://github.com/llvm/llvm-project/pull/88868