modal-cuda 0.1.0__tar.gz

modal_cuda-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 ExpressGradient
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

modal_cuda-0.1.0/PKG-INFO

@@ -0,0 +1,100 @@
+Metadata-Version: 2.4
+Name: modal-cuda
+Version: 0.1.0
+Summary: CLI that compiles and runs CUDA C programs on Modal GPUs
+Author: ExpressGradient
+License: MIT License
+        
+        Copyright (c) 2025 ExpressGradient
+        
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+        
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+        
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE.
+        
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: modal>=1.2.2
+Dynamic: license-file
+
+## mcc
+
+`mcc` is a tiny CLI that compiles and runs a local CUDA C (`.cu`) program on a cloud GPU using [Modal](https://modal.com). Install the package published as `modal-cuda`, keep using the `mcc` command, and the tool will ship your source to a disposable container, invoke `nvcc` inside the GPU runtime you pick, stream stdout/stderr back to your terminal, and clean up the artifacts once the run finishes.
+
+### Highlights
+
+- Runs CUDA samples on-demand without managing local GPU drivers.
+- Lets you choose GPU tier (`T4`, `A100`, `H100`, etc.) and container image per run.
+- Accepts extra `nvcc` arguments for fine-grained builds.
+- Wraps the entire workflow (upload → compile → execute → teardown) in a single command.
+
+## Requirements
+
+- Python 3.12+
+- A Modal account with CLI access (`modal token new`).
+- Network access that allows Modal to pull the chosen CUDA container image.
+
+## Installation
+
+```bash
+# install from PyPI/TestPyPI
+pip install modal-cuda          # exposes the `mcc` command
+
+# from source (editable)
+uv sync && uv pip install -e .
+# or: pip install -e .
+```
+
+## Usage
+
+```bash
+mcc path/to/program.cu [--gpu GPU] [--image IMAGE] [--timeout SECONDS] [--nvcc-arg FLAG ...]
+```
+
+| Flag | Description |
+| --- | --- |
+| `input` | Path to a `.cu` file. The file must exist and be non-empty. |
+| `--app` | Name of the Modal app; defaults to the source filename. Useful to group runs in the Modal dashboard. |
+| `--gpu` | GPU type to request (`T4`, `L4`, `A10`, `A100`, `A100-40GB`, `A100-80GB`, `L40S`, `H100`, `H200`, `B200`). Defaults to `T4`. |
+| `--image` | Container image reference passed to `modal.Image.from_registry`. Defaults to `nvidia/cuda:13.0.2-cudnn-devel-ubuntu22.04`. |
+| `--timeout` | Execution timeout (seconds) enforced by Modal. Defaults to 600. |
+| `--nvcc-arg` | Additional flag forwarded to `nvcc`. Repeat for multiple flags (e.g., `--nvcc-arg -arch=sm_90`). |
+
+Example session:
+
+```bash
+# Run the provided sample kernel on an A100 GPU with verbose PTX output
+mcc sample.cu --gpu A100 --nvcc-arg -Xptxas --nvcc-arg -v
+```
+
+Modal streams stdout/stderr as the compiler and executable run. If either command exits non‑zero, `mcc` surfaces the failure message and returns a non-zero status to your shell.
+
+## Development
+
+1. Install dependencies with `uv sync` (or `pip install -r requirements.txt` if you export them).
+2. Run `python -m mcc.main path/to/file.cu` to skip the console entry point while iterating.
+3. Use `uv run python -m pip install -e .` to get an editable install for local testing.
+
+## Troubleshooting
+
+- **`modal.ClientException: Unauthorized`** → Run `modal token new` and try again.
+- **`nvcc failed to compile...`** → Fix compiler errors shown in the streamed stderr or adjust `--nvcc-arg` flags.
+- **Long image pull times** → Prefer the default CUDA image or build a custom Modal image that layers your dependencies once and reuse it via `--image`.
+
+## License
+
+MIT © ExpressGradient

modal_cuda-0.1.0/README.md

@@ -0,0 +1,67 @@
+## mcc
+
+`mcc` is a tiny CLI that compiles and runs a local CUDA C (`.cu`) program on a cloud GPU using [Modal](https://modal.com). Install the package published as `modal-cuda`, keep using the `mcc` command, and the tool will ship your source to a disposable container, invoke `nvcc` inside the GPU runtime you pick, stream stdout/stderr back to your terminal, and clean up the artifacts once the run finishes.
+
+### Highlights
+
+- Runs CUDA samples on-demand without managing local GPU drivers.
+- Lets you choose GPU tier (`T4`, `A100`, `H100`, etc.) and container image per run.
+- Accepts extra `nvcc` arguments for fine-grained builds.
+- Wraps the entire workflow (upload → compile → execute → teardown) in a single command.
+
+## Requirements
+
+- Python 3.12+
+- A Modal account with CLI access (`modal token new`).
+- Network access that allows Modal to pull the chosen CUDA container image.
+
+## Installation
+
+```bash
+# install from PyPI/TestPyPI
+pip install modal-cuda          # exposes the `mcc` command
+
+# from source (editable)
+uv sync && uv pip install -e .
+# or: pip install -e .
+```
+
+## Usage
+
+```bash
+mcc path/to/program.cu [--gpu GPU] [--image IMAGE] [--timeout SECONDS] [--nvcc-arg FLAG ...]
+```
+
+| Flag | Description |
+| --- | --- |
+| `input` | Path to a `.cu` file. The file must exist and be non-empty. |
+| `--app` | Name of the Modal app; defaults to the source filename. Useful to group runs in the Modal dashboard. |
+| `--gpu` | GPU type to request (`T4`, `L4`, `A10`, `A100`, `A100-40GB`, `A100-80GB`, `L40S`, `H100`, `H200`, `B200`). Defaults to `T4`. |
+| `--image` | Container image reference passed to `modal.Image.from_registry`. Defaults to `nvidia/cuda:13.0.2-cudnn-devel-ubuntu22.04`. |
+| `--timeout` | Execution timeout (seconds) enforced by Modal. Defaults to 600. |
+| `--nvcc-arg` | Additional flag forwarded to `nvcc`. Repeat for multiple flags (e.g., `--nvcc-arg -arch=sm_90`). |
+
+Example session:
+
+```bash
+# Run the provided sample kernel on an A100 GPU with verbose PTX output
+mcc sample.cu --gpu A100 --nvcc-arg -Xptxas --nvcc-arg -v
+```
+
+Modal streams stdout/stderr as the compiler and executable run. If either command exits non‑zero, `mcc` surfaces the failure message and returns a non-zero status to your shell.
+
+## Development
+
+1. Install dependencies with `uv sync` (or `pip install -r requirements.txt` if you export them).
+2. Run `python -m mcc.main path/to/file.cu` to skip the console entry point while iterating.
+3. Use `uv run python -m pip install -e .` to get an editable install for local testing.
+
+## Troubleshooting
+
+- **`modal.ClientException: Unauthorized`** → Run `modal token new` and try again.
+- **`nvcc failed to compile...`** → Fix compiler errors shown in the streamed stderr or adjust `--nvcc-arg` flags.
+- **Long image pull times** → Prefer the default CUDA image or build a custom Modal image that layers your dependencies once and reuse it via `--image`.
+
+## License
+
+MIT © ExpressGradient

modal_cuda-0.1.0/main.py

@@ -0,0 +1,126 @@
+import argparse
+from pathlib import Path
+
+import modal
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        prog="mcc", description="Compile and Run CUDA C scripts with Modal"
+    )
+    parser.add_argument("input", help="Input CUDA C file", type=Path)
+    parser.add_argument(
+        "--app",
+        help="App name to run the CUDA script; defaults to the input file name",
+        type=str,
+    )
+    parser.add_argument(
+        "--gpu",
+        help="Choose the GPU to run the script",
+        type=str,
+        default="T4",
+        choices=[
+            "T4",
+            "L4",
+            "A10",
+            "A100",
+            "A100-40GB",
+            "A100-80GB",
+            "L40S",
+            "H100",
+            "H200",
+            "B200",
+        ],
+    )
+    parser.add_argument(
+        "--image",
+        type=str,
+        default="nvidia/cuda:13.0.2-cudnn-devel-ubuntu22.04",
+        help="Container image to use for Modal (registry reference)",
+    )
+    parser.add_argument(
+        "--timeout",
+        type=int,
+        default=60 * 10,
+        help="Execution timeout in seconds for the CUDA run (default: 600)",
+    )
+    parser.add_argument(
+        "--nvcc-arg",
+        dest="nvcc_args",
+        action="append",
+        default=None,
+        metavar="FLAG",
+        help="Extra argument to pass to nvcc; repeat for multiple flags",
+    )
+
+    args = parser.parse_args()
+
+    if not args.input.exists():
+        parser.error(f"File {args.input} not found")
+
+    if args.input.suffix.lower() != ".cu":
+        parser.error("Input file must have a .cu extension")
+
+    code = args.input.read_text()
+    if not code.strip():
+        parser.error(f"File {args.input} is empty")
+
+    app_name = args.app or args.input.name
+    cuda_image = modal.Image.from_registry(args.image, add_python="3.12")
+    app = modal.App(app_name, image=cuda_image)
+
+    @app.function(gpu=args.gpu, timeout=args.timeout, serialized=True)
+    def compile_and_run_cuda(cuda_source: str, nvcc_args: list[str] | None = None):
+        import subprocess
+        import uuid
+        from pathlib import Path
+
+        run_id = uuid.uuid4().hex
+        source_path = Path(f"/tmp/{run_id}.cu")
+        binary_path = Path(f"/tmp/{run_id}")
+        source_path.write_text(cuda_source)
+
+        def _run_command(cmd: list[str]) -> subprocess.CompletedProcess[str]:
+            print(f"$ {' '.join(cmd)}")
+
+            completed = subprocess.run(cmd, capture_output=True, text=True, check=False)
+
+            if completed.stdout:
+                print(completed.stdout)
+
+            if completed.stderr:
+                print(completed.stderr)
+
+            return completed
+
+        try:
+            extra_args = nvcc_args or []
+
+            compile_result = _run_command(
+                ["nvcc", str(source_path), "-o", str(binary_path), *extra_args]
+            )
+
+            if compile_result.returncode != 0:
+                raise RuntimeError("nvcc failed to compile the CUDA source file")
+
+            execution_result = _run_command([str(binary_path)])
+
+            if execution_result.returncode != 0:
+                raise RuntimeError(
+                    "Executable exited with a non-zero status while running the CUDA program"
+                )
+
+            return execution_result.stdout
+
+        finally:
+            for path in (source_path, binary_path):
+                if path.exists():
+                    path.unlink()
+
+    with modal.enable_output():
+        with app.run():
+            compile_and_run_cuda.remote(code, args.nvcc_args or [])
+
+
+if __name__ == "__main__":
+    main()

modal_cuda-0.1.0/modal_cuda.egg-info/PKG-INFO

@@ -0,0 +1,100 @@
+Metadata-Version: 2.4
+Name: modal-cuda
+Version: 0.1.0
+Summary: CLI that compiles and runs CUDA C programs on Modal GPUs
+Author: ExpressGradient
+License: MIT License
+        
+        Copyright (c) 2025 ExpressGradient
+        
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+        
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+        
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE.
+        
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: modal>=1.2.2
+Dynamic: license-file
+
+## mcc
+
+`mcc` is a tiny CLI that compiles and runs a local CUDA C (`.cu`) program on a cloud GPU using [Modal](https://modal.com). Install the package published as `modal-cuda`, keep using the `mcc` command, and the tool will ship your source to a disposable container, invoke `nvcc` inside the GPU runtime you pick, stream stdout/stderr back to your terminal, and clean up the artifacts once the run finishes.
+
+### Highlights
+
+- Runs CUDA samples on-demand without managing local GPU drivers.
+- Lets you choose GPU tier (`T4`, `A100`, `H100`, etc.) and container image per run.
+- Accepts extra `nvcc` arguments for fine-grained builds.
+- Wraps the entire workflow (upload → compile → execute → teardown) in a single command.
+
+## Requirements
+
+- Python 3.12+
+- A Modal account with CLI access (`modal token new`).
+- Network access that allows Modal to pull the chosen CUDA container image.
+
+## Installation
+
+```bash
+# install from PyPI/TestPyPI
+pip install modal-cuda          # exposes the `mcc` command
+
+# from source (editable)
+uv sync && uv pip install -e .
+# or: pip install -e .
+```
+
+## Usage
+
+```bash
+mcc path/to/program.cu [--gpu GPU] [--image IMAGE] [--timeout SECONDS] [--nvcc-arg FLAG ...]
+```
+
+| Flag | Description |
+| --- | --- |
+| `input` | Path to a `.cu` file. The file must exist and be non-empty. |
+| `--app` | Name of the Modal app; defaults to the source filename. Useful to group runs in the Modal dashboard. |
+| `--gpu` | GPU type to request (`T4`, `L4`, `A10`, `A100`, `A100-40GB`, `A100-80GB`, `L40S`, `H100`, `H200`, `B200`). Defaults to `T4`. |
+| `--image` | Container image reference passed to `modal.Image.from_registry`. Defaults to `nvidia/cuda:13.0.2-cudnn-devel-ubuntu22.04`. |
+| `--timeout` | Execution timeout (seconds) enforced by Modal. Defaults to 600. |
+| `--nvcc-arg` | Additional flag forwarded to `nvcc`. Repeat for multiple flags (e.g., `--nvcc-arg -arch=sm_90`). |
+
+Example session:
+
+```bash
+# Run the provided sample kernel on an A100 GPU with verbose PTX output
+mcc sample.cu --gpu A100 --nvcc-arg -Xptxas --nvcc-arg -v
+```
+
+Modal streams stdout/stderr as the compiler and executable run. If either command exits non‑zero, `mcc` surfaces the failure message and returns a non-zero status to your shell.
+
+## Development
+
+1. Install dependencies with `uv sync` (or `pip install -r requirements.txt` if you export them).
+2. Run `python -m mcc.main path/to/file.cu` to skip the console entry point while iterating.
+3. Use `uv run python -m pip install -e .` to get an editable install for local testing.
+
+## Troubleshooting
+
+- **`modal.ClientException: Unauthorized`** → Run `modal token new` and try again.
+- **`nvcc failed to compile...`** → Fix compiler errors shown in the streamed stderr or adjust `--nvcc-arg` flags.
+- **Long image pull times** → Prefer the default CUDA image or build a custom Modal image that layers your dependencies once and reuse it via `--image`.
+
+## License
+
+MIT © ExpressGradient

modal_cuda-0.1.0/modal_cuda.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+LICENSE
+README.md
+main.py
+pyproject.toml
+modal_cuda.egg-info/PKG-INFO
+modal_cuda.egg-info/SOURCES.txt
+modal_cuda.egg-info/dependency_links.txt
+modal_cuda.egg-info/entry_points.txt
+modal_cuda.egg-info/requires.txt
+modal_cuda.egg-info/top_level.txt

modal_cuda-0.1.0/modal_cuda.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

modal_cuda-0.1.0/modal_cuda.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+mcc = main:main

modal_cuda-0.1.0/modal_cuda.egg-info/requires.txt

@@ -0,0 +1 @@
+modal>=1.2.2

modal_cuda-0.1.0/modal_cuda.egg-info/top_level.txt

@@ -0,0 +1 @@
+main

modal_cuda-0.1.0/pyproject.toml

@@ -0,0 +1,14 @@
+[project]
+name = "modal-cuda"
+version = "0.1.0"
+description = "CLI that compiles and runs CUDA C programs on Modal GPUs"
+readme = "README.md"
+requires-python = ">=3.12"
+license = { file = "LICENSE" }
+authors = [{ name = "ExpressGradient" }]
+dependencies = [
+    "modal>=1.2.2",
+]
+
+[project.scripts]
+mcc = "main:main"

modal_cuda-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+