-
Notifications
You must be signed in to change notification settings - Fork 34
Home
Welcome to the PIMeval-PIMbench wiki! This wiki provides comprehensive documentation, including setup instructions, usage guidelines, architecture overview, and more.
PIMeval-PIMbench is a PIM simulator and benchmark suite designed for evaluating Processing in Memory (PIM). The simulator supports multiple PIM architectures and provides a performance and energy modeling.
- Ubuntu 22.04 or newer
- Pyhton3
- g++ 11.04 or newer
- CUDA Toolkit 11.8 or newer
Please note that the repository has not been tested on Windows or OSX, so we cannot guarantee compatibility with those systems. However, you're welcome to give it a try. CUDA and Python are required only if you intend to run the benchmarks. The PIMeval simulator itself does not require Python or CUDA compatibility.
git clone <url_to_your_fork>
cd PIMeval-PIMbench/
make -j<n_proc>
cd /PIMbench/<application_dir>/PIM
./<application_executable_name>.outReplace <url_to_your_fork>, <n_proc>, and <application_executable_name> with the appropriate values for your setup.
- Before submitting a PR, ensure you have fetched the latest changes from the main branch and run
make cleanfollowed bymaketo build the entire project and verify that no existing functionality is broken. - A small, focused PR is highly recommended. While it's understood that certain key features may require larger changes, try to keep each PR as concise as possible for easier review.
- Clear and descriptive commit messages explaining the purpose and scope of the changes is welcomed. And if you are wondering, we don't mind a dash of humor!
- If your PR addresses a specific issue, be sure to reference it in the description (e.g., “Fixes #123”).
- Make sure to test your changes and, if applicable, add or update any relevant tests to ensure coverage.
- Review your code for style and consistency with the coding standards before submitting the PR.