Port and run tests in python/test/unit/tools

.gitignore

@@ -32,6 +32,9 @@ python/triton/backends/
 # Language extras
 python/triton/language/extra
 
+# Tools extras
+python/triton/tools/extra
+
 # Proton
 python/triton/profiler
 

python/setup.py

@@ -34,9 +34,11 @@ class Backend:
     name: str
     package_data: List[str]
     language_package_data: List[str]
+    tools_package_data: List[str]
     src_dir: str
     backend_dir: str
     language_dir: Optional[str]
+    tools_dir: Optional[str]
     install_dir: str
     is_external: bool
 
@@ -68,6 +70,10 @@ def prepare(backend_name: str, backend_src_dir: str = None, is_external: bool =
         if not os.path.exists(language_dir):
             language_dir = None
 
+        tools_dir = os.path.abspath(os.path.join(backend_src_dir, "tools"))
+        if not os.path.exists(tools_dir):
+            tools_dir = None
+
         for file in ["compiler.py", "driver.py"]:
             assert os.path.exists(os.path.join(backend_path, file)), f"${file} does not exist in ${backend_path}"
 
@@ -78,9 +84,12 @@ def prepare(backend_name: str, backend_src_dir: str = None, is_external: bool =
         if language_dir is not None:
             language_package_data = [f"{os.path.relpath(p, language_dir)}/*" for p, _, _, in os.walk(language_dir)]
 
+        tools_package_data = []
+        if tools_dir is not None:
+            tools_package_data = [f"{os.path.relpath(p, tools_dir)}/*" for p, _, _, in os.walk(tools_dir)]
         return Backend(name=backend_name, package_data=package_data, language_package_data=language_package_data,
-                       src_dir=backend_src_dir, backend_dir=backend_path, language_dir=language_dir,
-                       install_dir=install_dir, is_external=is_external)
+                       tools_package_data=tools_package_data, src_dir=backend_src_dir, backend_dir=backend_path,
+                       language_dir=language_dir, tools_dir=tools_dir, install_dir=install_dir, is_external=is_external)
 
     # Copy all in-tree backends under triton/third_party.
     @staticmethod
@@ -600,6 +609,15 @@ def add_link_to_backends():
                 install_dir = os.path.join(extra_dir, x)
                 update_symlink(install_dir, src_dir)
 
+        if backend.tools_dir:
+            # Link the contents of each backend's `tools` directory into
+            # `triton.tools.extra`.
+            extra_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), "triton", "tools", "extra"))
+            for x in os.listdir(backend.tools_dir):
+                src_dir = os.path.join(backend.tools_dir, x)
+                install_dir = os.path.join(extra_dir, x)
+                update_symlink(install_dir, src_dir)
+
 
 def add_link_to_proton():
     proton_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), os.pardir, "third_party", "proton", "proton"))
@@ -642,9 +660,9 @@ def run(self):
 
 
 package_data = {
-    "triton/tools": ["compile.h", "compile.c"], **{f"triton/backends/{b.name}": b.package_data
-                                                   for b in backends}, "triton/language/extra": sum(
-        (b.language_package_data for b in backends), [])
+    "triton/tools/extra": sum((b.tools_package_data for b in backends), []),
+    **{f"triton/backends/{b.name}": b.package_data
+       for b in backends}, "triton/language/extra": sum((b.language_package_data for b in backends), [])
 }
 
 
@@ -669,6 +687,27 @@ def get_language_extra_packages():
     return list(packages)
 
 
+def get_tools_extra_packages():
+    packages = []
+    for backend in backends:
+        if backend.tools_dir is None:
+            continue
+
+        # Walk the `tools` directory of each backend to enumerate
+        # any subpackages, which will be added to `triton.tools.extra`.
+        for dir, dirs, files in os.walk(backend.tools_dir, followlinks=True):
+            if not any(f for f in files if f.endswith(('.c', '.h', '.cpp'))) or dir == backend.tools_dir:
+                # Ignore directories with no src and header files.
+                # Also ignore the root directory which corresponds to
+                # "triton/tools/extra".
+                continue
+            subpackage = os.path.relpath(dir, backend.tools_dir)
+            package = os.path.join("triton/tools/extra", subpackage)
+            packages.append(package)
+
+    return list(packages)
+
+
 def get_packages():
     packages = [
         "triton",
@@ -679,9 +718,11 @@ def get_packages():
         "triton/runtime",
         "triton/backends",
         "triton/tools",
+        "triton/tools/extra",
     ]
     packages += [f'triton/backends/{backend.name}' for backend in backends]
     packages += get_language_extra_packages()
+    packages += get_tools_extra_packages()
     if check_env_flag("TRITON_BUILD_PROTON", "ON"):  # Default ON
         packages += ["triton/profiler"]
 

python/test/unit/tools/test_aot.py

@@ -5,10 +5,13 @@
 import tempfile
 
 import numpy as np
+import pytest
 
 import triton
+from triton._internal_testing import is_cuda, is_xpu
 from triton.backends.compiler import GPUTarget
 from triton.backends.nvidia.driver import include_dir, library_dirs
+from triton.backends.intel.driver import compilation_helper
 
 kernel_utils_src = """
 import triton
@@ -97,21 +100,42 @@ def kernel(C, A, B, M, N, K,
 }"""
 
 
-def gen_kernel_library(dir, libname):
-    c_files = glob.glob(os.path.join(dir, "*.c"))
+def gen_kernel_library_xpu(dir, libname):
+    cpp_files = glob.glob(os.path.join(dir, "*.cpp"))
     subprocess.run(
-        ["gcc"] + c_files + ["-I", include_dir[0], "-c", "-fPIC"],
+        ["icpx"] + cpp_files + ["-I", compilation_helper.include_dir[0], "-c", "-fsycl", "-fPIC"],
         check=True,
         cwd=dir,
     )
     o_files = glob.glob(os.path.join(dir, "*.o"))
 
-    command = ["gcc", *o_files, "-shared", "-o", libname]
-    for lib_dir in library_dirs():
+    command = ["icpx", "-fsycl", "-lze_loader", *o_files, "-shared", "-o", libname]
+    for lib_dir in compilation_helper.library_dir:
         command.extend(["-L", lib_dir])
+    if compilation_helper.libsycl_dir:
+        for lib_dir in compilation_helper.libsycl_dir:
+            command.extend(["-L", lib_dir])
     subprocess.run(command, check=True, cwd=dir)
 
 
+def gen_kernel_library(dir, libname):
+    if is_xpu():
+        gen_kernel_library_xpu(dir, libname)
+    else:
+        c_files = glob.glob(os.path.join(dir, "*.c"))
+        subprocess.run(
+            ["gcc"] + c_files + ["-I", include_dir[0], "-c", "-fPIC"],
+            check=True,
+            cwd=dir,
+        )
+        o_files = glob.glob(os.path.join(dir, "*.o"))
+
+        command = ["gcc", *o_files, "-shared", "-o", libname]
+        for lib_dir in library_dirs():
+            command.extend(["-L", lib_dir])
+        subprocess.run(command, check=True, cwd=dir)
+
+
 def gen_test_bin(dir, M, N, K, exe="test", algo_id=0):
     test_src = f"""
 int main(int argc, char **argv) {{
@@ -171,15 +195,118 @@ def gen_test_bin(dir, M, N, K, exe="test", algo_id=0):
 }}
 """
     src = test_utils_src + test_src
-    with open(os.path.join(dir, "test.c"), "w") as file:
+    if is_xpu():
+        src = f"""
+#include "kernel.h"
+#include <assert.h>
+#include <cmath>
+#include <cstddef>
+#include <level_zero/ze_api.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <sycl/sycl.hpp>
+
+static void write_buffer_to_csv(char *filename, int32_t *buffer, int size) {{
+    FILE *file = fopen(filename, "w");
+    if (file == NULL) {{
+        printf("Could not open file %s\\n", filename);
+        return;
+    }}
+    for (int i = 0; i < size; i++) {{
+        fprintf(file, "%d", buffer[i]);
+        if (i < size - 1) {{
+            fprintf(file, ",");
+        }}
+    }}
+    fclose(file);
+}}
+
+static void read_csv_to_buffer(char *filename, int16_t *buffer, int size) {{
+    FILE *file = fopen(filename, "r");
+    if (file == NULL) {{
+        printf("Could not open file %s\\n", filename);
+        return;
+    }}
+    int index = 0;
+    while (fscanf(file, "%hd,", &buffer[index]) != EOF && index < size) {{
+        index++;
+    }}
+    fclose(file);
+}}
+int main(int argc, char ** argv) {{
+    int M = {M}, N = {N}, K = {K};
+
+    // initialize sycl handles
+    sycl::queue q{{sycl::gpu_selector_v}};
+    sycl::ext::intel::device_ptr<sycl::float16> A =
+        sycl::malloc_device<sycl::float16>(M * K * 2, q);
+    sycl::ext::intel::device_ptr<sycl::float16> B =
+        sycl::malloc_device<sycl::float16>(K * N * 2, q);
+    sycl::ext::intel::device_ptr<sycl::float16> C =
+        sycl::malloc_device<sycl::float16>(M * N * 4, q);
+
+    // initialize input data
+    int16_t hA[M * K];
+    int16_t hB[K * N];
+    memset(hA, 0, M * K * 2);
+    memset(hB, 0, K * N * 2);
+    read_csv_to_buffer(argv[1], hA, M * K);
+    read_csv_to_buffer(argv[2], hB, K * N);
+    q.memcpy(A, hA, M * K * 2).wait();
+    q.memcpy(B, hB, K * N * 2).wait();
+
+    // launch kernel
+    load_matmul_fp16();
+    int32_t ret;
+    int algo_id = {algo_id};
+    if (algo_id == 0) {{
+        ret = matmul_fp16_default(q, C, A, B, M, N, K, N, 1, K, 1, N, 1);
+    }} else {{
+        ret = matmul_fp16(q, C, A, B, M, N, K, N, 1, K, 1, N, 1, {algo_id});
+    }}
+    if (ret != 0) fprintf(stderr, "kernel launch failed\\n");
+    assert(ret == 0);
+
+    q.wait();
+
+    // read data
+    int32_t hC[M * N];
+    memset(hC, 0, M * N * 4);
+    q.memcpy(hC, C, M * N * 4).wait();
+    write_buffer_to_csv(argv[3], hC, M * N);
+
+    // free sycl resources
+    unload_matmul_fp16();
+    sycl::free(A, q);
+    sycl::free(B, q);
+    sycl::free(C, q);
+}}
+"""
+    src_name = "test.c"
+    if is_xpu():
+        src_name = "test.cpp"
+    with open(os.path.join(dir, src_name), "w") as file:
         file.write(src)
 
-    command = ["gcc", "test.c"]
-    for inc_dir in include_dir:
-        command.extend(["-I", inc_dir])
-    for lib_dir in library_dirs():
-        command.extend(["-L", lib_dir])
-    command.extend(["-l", "cuda", "-L", dir, "-l", "kernel", "-o", exe])
+    if is_cuda():
+        command = ["gcc", "test.c"]
+        for inc_dir in include_dir:
+            command.extend(["-I", inc_dir])
+        for lib_dir in library_dirs():
+            command.extend(["-L", lib_dir])
+        command.extend(["-l", "cuda", "-L", dir, "-l", "kernel", "-o", exe])
+
+    if is_xpu():
+        command = ["icpx", "test.cpp"]
+        for inc_dir in compilation_helper.include_dir:
+            command.extend(["-I", inc_dir])
+        for lib_dir in compilation_helper.library_dir:
+            command.extend(["-L", lib_dir])
+        if compilation_helper.libsycl_dir:
+            for lib_dir in compilation_helper.libsycl_dir:
+                command.extend(["-L", lib_dir])
+        command.extend(["-fsycl", "-lze_loader", "-L", dir, "-l", "kernel", "-o", exe])
     subprocess.run(command, check=True, cwd=dir)
 
 
@@ -283,6 +410,8 @@ def test_compile_link_matmul_no_specialization():
 
     with tempfile.TemporaryDirectory() as tmp_dir:
         dtype = "fp16"
+        if is_xpu():
+            pytest.skip("FIXME: AssertionError on XPU")
         BM, BN, BK = 16, 16, 16
 
         kernel_path = write_triton_kernels(tmp_dir, kernel_src, kernel_utils_src)
@@ -299,9 +428,8 @@ def test_compile_link_matmul_no_specialization():
 
         # run test case
         env = os.environ.copy()
-        env["LD_LIBRARY_PATH"] = tmp_dir
+        env["LD_LIBRARY_PATH"] = tmp_dir + ":" + env.get("LD_LIBRARY_PATH", "")
         subprocess.run(["./test", a_path, b_path, c_path], env=env, check=True, cwd=tmp_dir)
-
         # read data and compare against reference
         c = np.genfromtxt(c_path, delimiter=",", dtype=np.int32)
         c_tri = c.reshape((M, N)).view(np.float32)
@@ -330,7 +458,7 @@ def test_compile_link_matmul():
 
         # run test case
         env = os.environ.copy()
-        env["LD_LIBRARY_PATH"] = tmp_dir
+        env["LD_LIBRARY_PATH"] = tmp_dir + ":" + env.get("LD_LIBRARY_PATH", "")
         subprocess.run(["./test", a_path, b_path, c_path], env=env, check=True, cwd=tmp_dir)
 
         # read data and compare against reference
@@ -361,7 +489,7 @@ def test_launcher_has_no_available_kernel():
 
         # run test case
         env = os.environ.copy()
-        env["LD_LIBRARY_PATH"] = tmp_dir
+        env["LD_LIBRARY_PATH"] = tmp_dir + ":" + env.get("LD_LIBRARY_PATH", "")
         result = subprocess.run(
             ["./test", a_path, b_path, c_path],
             env=env,
@@ -410,7 +538,7 @@ def test_compile_link_autotune_matmul():
             gen_test_bin(tmp_dir, M, N, K, exe=test_name, algo_id=algo_id)
 
             env = os.environ.copy()
-            env["LD_LIBRARY_PATH"] = tmp_dir
+            env["LD_LIBRARY_PATH"] = tmp_dir + ":" + env.get("LD_LIBRARY_PATH", "")
             subprocess.run(
                 [f"./{test_name}", a_path, b_path, c_path],
                 check=True,
@@ -440,3 +568,21 @@ def test_ttgir_to_ptx():
         ptx = k.asm["ptx"]
         assert ".target sm_80" in ptx
         assert ".address_size 64" in ptx
+
+
+def test_ttgir_to_spv():
+    src = """
+module attributes {"ttg.num-warps" = 4 : i32, "ttg.threads-per-warp" = 32 : i32, "ttg.num-ctas" = 1 : i32} {
+  tt.func public @sum_kernel_0d1d(%arg0: !tt.ptr<i32>, %arg1: !tt.ptr<i32>) {
+    tt.return
+  }
+}
+"""
+    with tempfile.TemporaryDirectory() as tmp_dir:
+        kernel_path = os.path.join(tmp_dir, "empty_kernel.ttgir")
+        with open(kernel_path, "w") as fp:
+            fp.write(src)
+        k = triton.compile(kernel_path, target=triton.runtime.driver.active.get_current_target())
+        spv = k.asm['spvdis']
+        assert "OpCapability KernelAttributesINTEL" in spv
+        assert "SubgroupSize 32" in spv

python/test/unit/tools/test_disasm.py

@@ -19,3 +19,19 @@ def kernel(X, i: tl.constexpr):
     sass = h.asm["sass"]
     # check that the sass has a store instruction.
     assert "STG.E" in sass
+
+
+def test_disam_spvbin():
+    if not triton.runtime.driver.active.get_current_target().backend == "xpu":
+        pytest.skip("Test requires XPU.")
+
+    @triton.jit
+    def kernel(X, i: tl.constexpr):
+        tl.store(X, i)
+
+    x = torch.empty(1, dtype=torch.int32, device='xpu')
+    h = kernel[(1, )](x, i=12)
+    assert x[0] == 12
+    dis = h.asm["spvdis"]
+    # check that the spvdis has a store instruction.
+    assert "OpStore" in dis