diff --git a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-muladdsub.ll b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-muladdsub.ll
index 4201b27a6beba..a1c6a37933254 100644
--- a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-muladdsub.ll
+++ b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-muladdsub.ll
@@ -3,226 +3,442 @@
 
 target triple = "aarch64-unknown-linux-gnu"
 
-define <vscale x 8 x half> @combine_fmla(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmla(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
+; fadd(a, fmul(b, c)) -> fmla(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_1(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Test highlights an invalid combine!
+; fadd(a, fmul_u(b, c)) -> fmla(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_2(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_2(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; fadd_u(a, fmul_u(b, c)) -> fmla_u(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_3(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_3(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fadd_u(a, fmul(b, c)) -> fmla_u(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_4(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_4(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; fadd(fmul(b, c), a) -> fmad(b, c, a)
+define <vscale x 8 x half> @combine_fmuladd_5(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_5(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fadd(fmul_u(b, c), a) -> fmla_u(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_6(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_6(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[A:%.*]])
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
 }
 
-define <vscale x 8 x half> @combine_fmla_u(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmla_u(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmla.u.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
+; TODO: Missing combine!
+; fadd_u(fmul_u(b, c), a) -> fmla_u(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_7(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_7(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[A:%.*]])
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fadd_u(fmul(b, c), a) -> fmla_u(a, b, c)
+define <vscale x 8 x half> @combine_fmuladd_8(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmuladd_8(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; fsub(a, fmul(b, c)) -> fmls(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_1(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Test highlights an invalid combine!
+; fsub(a, fmul_u(b, c)) -> fmls(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_2(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_2(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
 }
 
-define <vscale x 16 x i8> @combine_mla_i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @combine_mla_i8(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[C:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]])
+; fsub_u(a, fmul_u(b, c)) -> fmls_u(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_3(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_3(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fsub_u(a, fmul(b, c)) -> fmls_u(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_4(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_4(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+; fsub(mul(b, c), a) -> fnmsb(b, c, a)
+define <vscale x 8 x half> @combine_fmulsub_5(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_5(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fsub(fmul_u(b, c), a) -> fnmls_u(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_6(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_6(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fsub_u(fmul_u(b, c), a) -> fnmls_u(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_7(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_7(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; TODO: Missing combine!
+; fsub_u(fmul(b, c), a) -> fnmls_u(a, b, c)
+define <vscale x 8 x half> @combine_fmulsub_8(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @combine_fmulsub_8(
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
+;
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %1, <vscale x 8 x half> %a)
+  ret <vscale x 8 x half> %2
+}
+
+; add(a, mul(b, c)) -> mla(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_1(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %c, <vscale x 16 x i8> %1)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 16 x i8> @combine_mla_i8_u(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @combine_mla_i8_u(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[C:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]])
+; TODO: Missing combine!
+; add(a, mul_u(b, c)) -> mla(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_2(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_2(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
+;
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
+  ret <vscale x 16 x i8> %2
+}
+
+; add_u(a, mul_u(b, c)) -> mla_u(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_3(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_3(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %c, <vscale x 16 x i8> %1)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @combine_fmad(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmad(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+; TODO: Missing combine!
+; add_u(a, mul(b, c)) -> mla_u(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_4(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_4(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %2, <vscale x 8 x half> %c)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
+  ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 16 x i8> @combine_mad_i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @combine_mad_i8(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; add(mul(b, c), a) -> mad(b, c, c)
+define <vscale x 16 x i8> @combine_muladd_5(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_5(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]], <vscale x 16 x i8> [[A:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %c)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @combine_fmls(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmls(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+; TODO: Missing combine!
+; add(mul_u(b, c), a) -> mla_u(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_6(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_6(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @combine_fmls_u(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmls_u(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fmls.u.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+; TODO: Missing combine!
+; add_u(mul_u(b, c), a) -> mla_u(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_7(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_7(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
+}
+
+; TODO: Missing combine!
+; add_u(mul(b, c), a) -> mla_u(a, b, c)
+define <vscale x 16 x i8> @combine_muladd_8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_muladd_8(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
+;
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 16 x i8> @combine_mls_i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @combine_mls_i8(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[C:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]])
+; sub(a, mul(b, c)) -> mls(a, b, c)
+define <vscale x 16 x i8> @combine_mulsub_1(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %c, <vscale x 16 x i8> %1)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 16 x i8> @combine_mls_i8_u(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @combine_mls_i8_u(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[C:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]])
+; TODO: Missing combine!
+; sub(a, mul_u(b, c)) -> mls(a, b, c)
+define <vscale x 16 x i8> @combine_mulsub_2(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_2(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
+;
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
+  ret <vscale x 16 x i8> %2
+}
+
+; sub_u(a, mul_u(b, c)) -> mls_u(a, b, c)
+define <vscale x 16 x i8> @combine_mulsub_3(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_3(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %c, <vscale x 16 x i8> %1)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @combine_fnmsb(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fnmsb(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call fast <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+; TODO: Missing combine!
+; sub_u(a, mul(b, c)) -> mls_u(a, b, c)
+define <vscale x 16 x i8> @combine_mulsub_4(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_4(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %2, <vscale x 8 x half> %c)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %1)
+  ret <vscale x 16 x i8> %2
 }
 
-; No integer variant of fnmsb exists; Do not combine
-define <vscale x 16 x i8> @neg_combine_nmsb_i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
-; CHECK-LABEL: @neg_combine_nmsb_i8(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[A:%.*]], <vscale x 16 x i8> [[B:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[C:%.*]])
+; sub(mul(b, c), a) -> sub(mul(b, c), a)
+define <vscale x 16 x i8> @combine_mulsub_5(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_5(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b)
-  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %c)
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
   ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @combine_fmla_contract_flag_only(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @combine_fmla_contract_flag_only(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = call contract <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
+; TODO: Missing combine!
+; sub(mul_u(b, c), a) -> sub_u(mul_u(b, c), a)
+define <vscale x 16 x i8> @combine_mulsub_6(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_6(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call contract <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call contract <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @neg_combine_fmla_no_flags(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @neg_combine_fmla_no_flags(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[TMP2]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
+; sub_u(mul_u(b, c), a) -> sub_u(mul_u(b, c), a)
+define <vscale x 16 x i8> @combine_mulsub_7(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_7(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 ;
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
 }
 
-define <vscale x 8 x half> @neg_combine_fmla_neq_pred(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @neg_combine_fmla_neq_pred(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 5)
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[TMP2]])
-; CHECK-NEXT:    [[TMP4:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    [[TMP5:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP3]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[TMP4]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP5]]
+; TODO: Missing combine!
+; sub_u(mul(b, c), a) -> sub_u(mul_u(b, c), a)
+define <vscale x 16 x i8> @combine_mulsub_8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
+; CHECK-LABEL: @combine_mulsub_8(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[P:%.*]], <vscale x 16 x i8> [[B:%.*]], <vscale x 16 x i8> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> [[P]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
+;
+  %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
+  %2 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1> %p, <vscale x 16 x i8> %1, <vscale x 16 x i8> %a)
+  ret <vscale x 16 x i8> %2
+}
+
+define <vscale x 8 x half> @neg_combine_fmuladd_neq_pred(<vscale x 8 x i1> %p1, <vscale x 8 x i1> %p2, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @neg_combine_fmuladd_neq_pred(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[P1:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[P2:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 ;
 ; ret <vscale x 8 x half> %9
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 5)
-  %3 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %2)
-  %4 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %5 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %3, <vscale x 8 x half> %c, <vscale x 8 x half> %4)
-  ret <vscale x 8 x half> %5
-}
-
-define <vscale x 8 x half> @neg_combine_fmla_two_fmul_uses(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @neg_combine_fmla_two_fmul_uses(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[B:%.*]])
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[C:%.*]], <vscale x 8 x half> [[TMP2]])
-; CHECK-NEXT:    [[TMP4:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP2]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP4]]
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p1, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p2, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
+}
+
+define <vscale x 8 x half> @neg_combine_fmuladd_two_fmul_uses(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @neg_combine_fmuladd_two_fmul_uses(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
 ;
 ; ret <vscale x 8 x half> %8
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %b)
-  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %c, <vscale x 8 x half> %2)
-  %4 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %3, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %4
-}
-
-define <vscale x 8 x half> @neg_combine_fmla_neq_flags(<vscale x 16 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
-; CHECK-LABEL: @neg_combine_fmla_neq_flags(
-; CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[P:%.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = tail call reassoc nnan contract <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
-; CHECK-NEXT:    [[TMP3:%.*]] = tail call reassoc contract <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP1]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP2]])
-; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
+  %1 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  %3 = tail call fast <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %2, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %3
+}
+
+define <vscale x 8 x half> @neg_combine_fmuladd_neq_flags(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %b, <vscale x 8 x half> %c) #0 {
+; CHECK-LABEL: @neg_combine_fmuladd_neq_flags(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call reassoc nnan contract <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[P:%.*]], <vscale x 8 x half> [[B:%.*]], <vscale x 8 x half> [[C:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = tail call reassoc contract <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[P]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 ;
 ; ret <vscale x 8 x half> %7
-  %1 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %p)
-  %2 = tail call reassoc nnan contract <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
-  %3 = tail call reassoc contract <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %1, <vscale x 8 x half> %a, <vscale x 8 x half> %2)
-  ret <vscale x 8 x half> %3
+  %1 = tail call reassoc nnan contract <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %b, <vscale x 8 x half> %c)
+  %2 = tail call reassoc contract <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> %p, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
+  ret <vscale x 8 x half> %2
 }
 
-declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32)
-declare <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1>)
-declare <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1>)
-declare <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
+
 declare <vscale x 8 x half> @llvm.aarch64.sve.fmul.u.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.fadd.u.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.fsub.u.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
+
 declare <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
 declare <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
 declare <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
+
 declare <vscale x 16 x i8> @llvm.aarch64.sve.mul.u.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
 declare <vscale x 16 x i8> @llvm.aarch64.sve.add.u.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
 declare <vscale x 16 x i8> @llvm.aarch64.sve.sub.u.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)