diff --git a/cross-project-tests/CMakeLists.txt b/cross-project-tests/CMakeLists.txt
--- a/cross-project-tests/CMakeLists.txt
+++ b/cross-project-tests/CMakeLists.txt
@@ -94,3 +94,11 @@
 set_target_properties(check-cross-project PROPERTIES FOLDER "Tests")
 set_target_properties(check-debuginfo PROPERTIES FOLDER "Tests")
 set_target_properties(check-intrinsic-headers PROPERTIES FOLDER "Tests")
+if ("AMDGPU" IN_LIST LLVM_TARGETS_TO_BUILD)
+  add_lit_testsuite(check-amdgpu-cross "Running AMDGPU cross-project-tests"
+    ${CMAKE_CURRENT_BINARY_DIR}/amdgpu
+    EXCLUDE_FROM_CHECK_ALL
+    DEPENDS clang
+    )
+  set_target_properties(check-amdgpu-cross PROPERTIES FOLDER "Tests")
+endif()
diff --git a/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w32.cl b/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w32.cl
new file mode 100644
--- /dev/null
+++ b/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w32.cl
@@ -0,0 +1,69 @@
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1100 -target-feature +wavefrontsize32 -DWMMA_GFX1100_TESTS -S -o - %s | FileCheck %s --check-prefix=CHECK-GFX1100
+
+typedef float  v4f   __attribute__((ext_vector_type(4)));
+typedef float  v8f   __attribute__((ext_vector_type(8)));
+typedef half   v16h  __attribute__((ext_vector_type(16)));
+typedef int    v2i   __attribute__((ext_vector_type(2)));
+typedef int    v4i   __attribute__((ext_vector_type(4)));
+typedef int    v8i   __attribute__((ext_vector_type(8)));
+typedef short  v16s  __attribute__((ext_vector_type(16)));
+
+#ifdef WMMA_GFX1100_TESTS
+
+// Wave32
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f32_16x16x16_f16_w32:
+// CHECK-GFX1100: v_wmma_f32_16x16x16_f16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}]
+//
+void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v16h a, v16h b, v8f c)
+{
+  *out = __builtin_amdgcn_wmma_f32_16x16x16_f16_w32(a, b, c);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f32_16x16x16_bf16_w32:
+// CHECK-GFX1100: v_wmma_f32_16x16x16_bf16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}]
+//
+void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v16s a, v16s b, v8f c)
+{
+  *out = __builtin_amdgcn_wmma_f32_16x16x16_bf16_w32(a, b, c);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f16_16x16x16_f16_w32:
+// CHECK-GFX1100: v_wmma_f16_16x16x16_f16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}] op_sel:[0,0,1]
+//
+void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v16h* out, v16h a, v16h b, v16h c)
+{
+  *out = __builtin_amdgcn_wmma_f16_16x16x16_f16_w32(a, b, c, true);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_bf16_16x16x16_bf16_w32:
+// CHECK-GFX1100: v_wmma_bf16_16x16x16_bf16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}] op_sel:[0,0,1]
+//
+void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v16s* out, v16s a, v16s b, v16s c)
+{
+  *out = __builtin_amdgcn_wmma_bf16_16x16x16_bf16_w32(a, b, c, true);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_i32_16x16x16_iu8_w32:
+// CHECK-GFX1100: v_wmma_i32_16x16x16_iu8 v[{{.*}}], v[{{.*}} v[{{.*}} v[{{.*}}] neg_lo:[1,1,0]
+//
+void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v4i a, v4i b, v8i c)
+{
+  *out = __builtin_amdgcn_wmma_i32_16x16x16_iu8_w32(true, a, true, b, c, false);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_i32_16x16x16_iu4_w32:
+// CHECK-GFX1100: v_wmma_i32_16x16x16_iu4 v[{{.*}}, v[{{.*}} v[{{.*}} v[{{.*}} neg_lo:[1,1,0]
+void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, v2i a, v2i b, v8i c)
+{
+  *out = __builtin_amdgcn_wmma_i32_16x16x16_iu4_w32(true, a, true, b, c, false);
+}
+
+#endif
diff --git a/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w64.cl b/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w64.cl
new file mode 100644
--- /dev/null
+++ b/cross-project-tests/amdgpu/builtins-amdgcn-wmma-w64.cl
@@ -0,0 +1,73 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1100 -target-feature +wavefrontsize64 -DWMMA_GFX1100_TESTS -S -o - %s | FileCheck %s --check-prefix=CHECK-GFX1100
+
+typedef float  v4f   __attribute__((ext_vector_type(4)));
+typedef float  v8f   __attribute__((ext_vector_type(8)));
+typedef half   v8h   __attribute__((ext_vector_type(8)));
+typedef half   v16h  __attribute__((ext_vector_type(16)));
+typedef int    v2i   __attribute__((ext_vector_type(2)));
+typedef int    v4i   __attribute__((ext_vector_type(4)));
+typedef int    v8i   __attribute__((ext_vector_type(8)));
+typedef short  v8s   __attribute__((ext_vector_type(8)));
+typedef short  v16s  __attribute__((ext_vector_type(16)));
+
+#ifdef WMMA_GFX1100_TESTS
+
+// Wave64
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f32_16x16x16_f16_w64:
+// CHECK-GFX1100: v_wmma_f32_16x16x16_f16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}]
+//
+void test_amdgcn_wmma_f32_16x16x16_f16_w64(global v4f* out, v16h a, v16h b, v4f c)
+{
+  *out = __builtin_amdgcn_wmma_f32_16x16x16_f16_w64(a, b, c);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f32_16x16x16_bf16_w64:
+// CHECK-GFX1100: v_wmma_f32_16x16x16_bf16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}]
+//
+void test_amdgcn_wmma_f32_16x16x16_bf16_w64(global v4f* out, v16s a, v16s b, v4f c)
+{
+  *out = __builtin_amdgcn_wmma_f32_16x16x16_bf16_w64(a, b, c);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_f16_16x16x16_f16_w64:
+// CHECK-GFX1100: v_wmma_f16_16x16x16_f16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}] op_sel:[0,0,1]
+//
+void test_amdgcn_wmma_f16_16x16x16_f16_w64(global v8h* out, v16h a, v16h b, v8h c)
+{
+  *out = __builtin_amdgcn_wmma_f16_16x16x16_f16_w64(a, b, c, true);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_bf16_16x16x16_bf16_w64:
+// CHECK-GFX1100: v_wmma_bf16_16x16x16_bf16 v[{{.*}}], v[{{.*}} v[{{.*}}], v[{{.*}}] op_sel:[0,0,1]
+//
+void test_amdgcn_wmma_bf16_16x16x16_bf16_w64(global v8s* out, v16s a, v16s b, v8s c)
+{
+  *out = __builtin_amdgcn_wmma_bf16_16x16x16_bf16_w64(a, b, c, true);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_i32_16x16x16_iu8_w64:
+// CHECK-GFX1100: v_wmma_i32_16x16x16_iu8 v[{{.*}}], v[{{.*}} v[{{.*}} v[{{.*}}] neg_lo:[1,1,0]
+//
+void test_amdgcn_wmma_i32_16x16x16_iu8_w64(global v4i* out, v4i a, v4i b, v4i c)
+{
+  *out = __builtin_amdgcn_wmma_i32_16x16x16_iu8_w64(true, a, true, b, c, false);
+}
+
+
+// CHECK-GFX1100-LABEL: test_amdgcn_wmma_i32_16x16x16_iu4_w64:
+// CHECK-GFX1100: v_wmma_i32_16x16x16_iu4 v[{{.*}} v[{{.*}} v[{{.*}} v[{{.*}}neg_lo:[1,1,0]
+//
+void test_amdgcn_wmma_i32_16x16x16_iu4_w64(global v4i* out, v2i a, v2i b, v4i c)
+{
+  *out = __builtin_amdgcn_wmma_i32_16x16x16_iu4_w64(true, a, true, b, c, false);
+}
+
+#endif
diff --git a/cross-project-tests/amdgpu/lit.local.cfg b/cross-project-tests/amdgpu/lit.local.cfg
new file mode 100644
--- /dev/null
+++ b/cross-project-tests/amdgpu/lit.local.cfg
@@ -0,0 +1,2 @@
+if 'clang' not in config.available_features:
+    config.unsupported = True
diff --git a/cross-project-tests/lit.cfg.py b/cross-project-tests/lit.cfg.py
--- a/cross-project-tests/lit.cfg.py
+++ b/cross-project-tests/lit.cfg.py
@@ -22,7 +22,7 @@
 config.test_format = lit.formats.ShTest(not llvm_config.use_lit_shell)
 
 # suffixes: A list of file extensions to treat as test files.
-config.suffixes = ['.c', '.cpp', '.m']
+config.suffixes = ['.c', ".cl", '.cpp', '.m']
 
 # excludes: A list of directories to exclude from the testsuite. The 'Inputs'
 # subdirectories contain auxiliary inputs for various tests in their parent