diff --git a/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll b/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll
new file mode 100644
index 0000000000000..1de8ab5d3e590
--- /dev/null
+++ b/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll
@@ -0,0 +1,30 @@
+; RUN: opt %s -aa-pipeline=basic-aa -passes=aa-eval -print-all-alias-modref-info -disable-output 2>&1 | FileCheck %s
+
+; BasicAA should prove that loads from sufficiently large static offsets
+; don't overlap with matrix loads with a statically known size.
+
+define <8 x double> @non_overlapping_strided_load(ptr %src) {
+; CHECK-LABEL: Function: non_overlapping_strided_load:
+; Just Ref: %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2) <-> call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+; Just Mod: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2) <-> %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 12
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+  ret <8 x double> %l
+}
+
+define <8 x double> @overlapping_strided_load(ptr %src) {
+; CHECK-LABEL: Function: overlapping_strided_load:
+; CHECK: Just Ref: %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2) <-> call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+; CHECK: Just Mod: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2) <-> %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+;
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 11
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+  ret <8 x double> %l
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
diff --git a/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll b/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll
new file mode 100644
index 0000000000000..ae3c7464656df
--- /dev/null
+++ b/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll
@@ -0,0 +1,338 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
+; RUN: opt -passes=dse -S %s | FileCheck %s
+
+define void @dead_unstrided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_unstrided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = load double, ptr %src
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_unstrided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_unstrided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l.2 = load double, ptr %dst
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_strided_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_strided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 200, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 200, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_strided_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_strided_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 200, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 100, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 200, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 100, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  %l.2 = load double, ptr %src
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  %l.2 = load double, ptr %dst
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_dynamically_strided_store(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @dead_dynamically_strided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_dynamically_strided_store(ptr %ptr, i32 %stride) {
+; CHECK-LABEL: define void @live_dynamically_strided_store(
+; CHECK-SAME: ptr [[PTR:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_dynamically_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @dead_dynamically_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  %l.2 = load double, ptr %src
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_dynamically_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @live_dynamically_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  %l.2 = load double, ptr %dst
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_unstrided_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_unstrided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_unstrided_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_unstrided_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_non_matrix_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_non_matrix_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[DST_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 6
+; CHECK-NEXT:    store double 4.200000e+01, ptr [[DST_OFFSET]], align 8
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %dst.offset = getelementptr inbounds double, ptr %src, i32 6
+  store double 42.0, ptr %dst.offset
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @live_non_matrix_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_non_matrix_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[PTR_OFFSET:%.*]] = getelementptr inbounds double, ptr [[PTR]], i32 6
+; CHECK-NEXT:    store double 4.200000e+01, ptr [[PTR_OFFSET]], align 8
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %ptr.offset = getelementptr inbounds double, ptr %ptr, i32 6
+  store double 42.0, ptr %ptr.offset
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+define void @dead_matrix_store_non_matrix_overwrite_unstrided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_non_matrix_overwrite_unstrided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store <8 x double> zeroinitializer, ptr [[DST]], align 64
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  store <8 x double> zeroinitializer, ptr %dst
+  ret void
+}
+
+define void @dead_matrix_store_non_matrix_overwrite_strided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_non_matrix_overwrite_strided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store <16 x double> zeroinitializer, ptr [[DST]], align 128
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+  store <16 x double> zeroinitializer, ptr %dst
+  ret void
+}
+
+define void @live_matrix_store_non_matrix_overwrite_unstrided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_non_matrix_overwrite_unstrided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store <4 x double> zeroinitializer, ptr [[DST]], align 32
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  store <4 x double> zeroinitializer, ptr %dst
+  ret void
+}
+
+define void @live_matrix_store_non_matrix_overwrite_strided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_non_matrix_overwrite_strided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store <8 x double> zeroinitializer, ptr [[DST]], align 64
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+  store <8 x double> zeroinitializer, ptr %dst
+  ret void
+}
+
+define void @dead_matrix_store_dimension_change(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_dimension_change(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr [[DST]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr %dst, i32 3, i1 false, i32 3, i32 3)
+  ret void
+}
+
+define void @live_matrix_store_dimension_change(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_dimension_change(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr [[DST]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr %dst, i32 3, i1 false, i32 3, i32 3)
+  call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+  ret void
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
diff --git a/llvm/test/Transforms/GVN/matrix-intrinsics.ll b/llvm/test/Transforms/GVN/matrix-intrinsics.ll
new file mode 100644
index 0000000000000..78dbfe1ef6bd8
--- /dev/null
+++ b/llvm/test/Transforms/GVN/matrix-intrinsics.ll
@@ -0,0 +1,136 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
+; RUN: opt -passes=gvn -S %s | FileCheck %s
+
+define void @redundant_unstrided_load(ptr %src) {
+; CHECK-LABEL: define void @redundant_unstrided_load(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 8
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_2]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 8
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64(<8 x double> %l, ptr %src, i32 4, i1 false, i32 4, i32 2)
+  %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.2)
+  ret void
+}
+
+define void @redundant_unstrided_load_non_matrix_store(ptr %src) {
+; CHECK-LABEL: define void @redundant_unstrided_load_non_matrix_store(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 1
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store double 4.200000e+01, ptr [[SRC]], align 8
+; CHECK-NEXT:    [[L_2:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_2]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 1
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+  store double 42.0, ptr %src
+  %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.2)
+  ret void
+}
+
+define void @redundant_strided_load(ptr %src) {
+; CHECK-LABEL: define void @redundant_strided_load(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 16
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_2]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 16
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  call void @llvm.matrix.column.major.store.v8f64(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+  %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.2)
+  ret void
+
+}
+
+define void @redundant_strided_load_non_matrix_store(ptr %src) {
+; CHECK-LABEL: define void @redundant_strided_load_non_matrix_store(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 16
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    store double 4.200000e+01, ptr [[SRC]], align 8
+; CHECK-NEXT:    [[L_2:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_2]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %src.offset = getelementptr inbounds double, ptr %src, i32 16
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  store double 42.0, ptr %src
+  %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.2)
+  ret void
+}
+
+define void @repeat_load_dimension_change_project(ptr %src) {
+; CHECK-LABEL: define void @repeat_load_dimension_change_project(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr [[SRC]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT:    [[L_3:%.*]] = shufflevector <9 x double> [[L_2]], <9 x double> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_3]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  %l.2 = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr %src, i32 3, i1 false, i32 3, i32 3)
+  %l.3 = shufflevector <9 x double> %l.2, <9 x double> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.3)
+  ret void
+}
+
+define void @repeat_load_dimension_change_shuffle(ptr %src) {
+; CHECK-LABEL: define void @repeat_load_dimension_change_shuffle(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT:    [[L_2:%.*]] = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr [[SRC]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT:    [[L_3:%.*]] = shufflevector <9 x double> [[L_2]], <9 x double> zeroinitializer, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    call void @use(<8 x double> [[L]])
+; CHECK-NEXT:    call void @use(<8 x double> [[L_3]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+  %l.2 = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr %src, i32 3, i1 false, i32 3, i32 3)
+  %l.3 = shufflevector <9 x double> %l.2, <9 x double> zeroinitializer, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  call void @use(<8 x double> %l)
+  call void @use(<8 x double> %l.3)
+  ret void
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
+declare void @use(<8 x double>)