v4.3 update. (#2709)

* v4.3 update.

* Update the cute_dsl_api changelog's doc link

* Update version to 4.3.0

* Update the example link

* Update doc to encourage user to install DSL from requirements.txt

---------

Co-authored-by: Larry Wu <larwu@nvidia.com>

examples/python/CuTeDSL/cute/ffi/CMakeLists.txt

@@ -30,11 +30,12 @@ cmake_minimum_required(VERSION 3.15)
 project(tensor)
 
 # Find Python
+find_package(Python COMPONENTS Interpreter Development REQUIRED)
 find_package(Python3 COMPONENTS Interpreter Development REQUIRED)
 
 # Get Python site-packages directory using Python
 execute_process(
-    COMMAND ${Python_EXECUTABLE} -c "import site; print(site.getsitepackages()[0])"
+    COMMAND ${Python3_EXECUTABLE} -c "import site; print(site.getsitepackages()[0])"
     OUTPUT_VARIABLE Python_SITE_PACKAGES
     OUTPUT_STRIP_TRAILING_WHITESPACE
 )
@@ -45,7 +46,13 @@ message(STATUS "Python site-packages directory: ${Python_SITE_PACKAGES}")
 list(APPEND CMAKE_PREFIX_PATH ${Python_SITE_PACKAGES}/nanobind/cmake)
 
 # Find nanobind
-find_package(nanobind REQUIRED)
+find_package(nanobind)
+if(NOT nanobind_FOUND)
+    message(FATAL_ERROR 
+        "nanobind not found!\n"
+        "Please install nanobind with: pip install nanobind\n"
+    )
+endif()
 
 # Add the module
 nanobind_add_module(tensor tensor.cpp)

examples/python/CuTeDSL/cute/ffi/jit_argument.py

@@ -54,7 +54,6 @@ import cutlass.cute as cute
 
 from cutlass._mlir import ir
 from cutlass._mlir.dialects import llvm
-import cutlass._mlir.extras.types as T
 
 
 class ExampleTensorValue(ir.Value):
@@ -244,7 +243,7 @@ import tempfile
 import torch
 
 
-def run_test(tmpdir=None):
+def run_test(tmpdir=None, cmake_args=""):
     # Skip cleanup if user provides tmpdir
     cleanup = tmpdir is None
     # Initialize temporary build directory
@@ -253,7 +252,8 @@ def run_test(tmpdir=None):
     try:
         current_dir = os.path.dirname(os.path.abspath(__file__))
 
-        subprocess.run(["cmake", "-B", tmpdir, current_dir], check=True)
+        cmake_args = cmake_args.split()
+        subprocess.run(["cmake", "-B", tmpdir, current_dir] + cmake_args, check=True)
         subprocess.run(["cmake", "--build", tmpdir], check=True)
 
         sys.path.append(tmpdir)
@@ -284,7 +284,10 @@ def run_test(tmpdir=None):
         # Execute compiled function
         compiled_func(tensor)
     except Exception as e:
-        print(e)
+        import traceback
+
+        traceback.print_exception(type(e), e, e.__traceback__)
+        raise e
     finally:
         if cleanup:
             # Clean up the temporary directory
@@ -298,8 +301,17 @@ if __name__ == "__main__":
         description="Set temporary directory for building C modules"
     )
     parser.add_argument(
-        "--tmp-dir", type=str, help="Temporary directory path for building C modules"
+        "--tmp-dir",
+        type=str,
+        default=None,
+        help="Temporary directory path for building C modules",
+    )
+    parser.add_argument(
+        "--cmake-args",
+        type=str,
+        default="",
+        help="Extra CMake arguments for building C modules",
     )
     args = parser.parse_args()
 
-    run_test(args.tmp_dir)
+    run_test(tmpdir=args.tmp_dir, cmake_args=args.cmake_args)