Hi, I'm Sung-Woo Choi from South Korea.
I'm a Computer Science student at UNIST.

I'm interested in compilers, systems, and GPUs.
My work focuses on LLVM-based analysis, low-level performance, and correctness issues in systems software.

https://swc0317.tistory.com/

swc0317@unist.ac.kr / swchoi@nshc.net

• Opened a pull request in the exaloop/codon repository (#826) for implementing a Python GPU DSL decorator. This implementation separates the GPU path from the existing CPU target JIT decorator structure, adding a codon.gpu decorator and GPU-specific source rewriting. The pull request is currently open and has not been merged yet. View PR.

• Merged a pull request in the exaloop/codon repository (#821) that fixed an invalid behavior when using an ordered dictionary in GPU round-trip usage. This fix addressed a segmentation fault that occurred when copying an ordered dictionary with list entries from host to device and back. View PR.

• Opened a pull request (#819) to address the limited global variable initialization problem in the exaloop/codon repository. This update to the constant propagation pass allows global constants initialized through scalar cast/constructor patterns to be folded, expanding the conditions under which global constants are recognized. The pull request is currently open and has not been merged yet. View PR.

• Raised an issue (#818) regarding a complex sin operation resulting in nan values when executed on the GPU. This issue is believed to be related to the uninitialized global variable problem previously identified. View Issue.

• Commented on issue #826, confirming that test cases in codon/test/transform/kernels.codon pass when all kernels of Codon native are rewritten in

1. 🗣 Commented on #826 in exaloop/codon
2. 💪 Opened PR #826 in exaloop/codon
3. 🎉 Merged PR #821 in exaloop/codon
4. 💪 Opened PR #821 in exaloop/codon
5. 🗣 Commented on #819 in exaloop/codon
6. 💪 Opened PR #819 in exaloop/codon
7. ❗ Opened issue #818 in exaloop/codon
8. 🎉 Merged PR #804 in exaloop/codon
9. 🗣 Commented on #804 in exaloop/codon
10. ❗ Opened issue #811 in exaloop/codon