CuLifter: Lifting GPU Binaries to Typed IR
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914520139563008 |
|---|---|
| author | Zhao, Jisheng Pu, Huanzhi Jeong, Shinnung Ahn, Chihyo Kim, Hyesoon |
| author_facet | Zhao, Jisheng Pu, Huanzhi Jeong, Shinnung Ahn, Chihyo Kim, Hyesoon |
| contents | GPU compilers merge all data types into a single unified register file, erasing the type information that binary-analysis tools rely on. We show that type recovery from this untyped register file is the central challenge of GPU binary lifting. We present CuLifter, a SASS-to-LLVM IR lifting framework that recovers register types via constraint propagation with conflict detection, reconstructs explicit control flow, and aggregates multi-instruction patterns. Across eight benchmark suites (24,437 GPU functions in 919 cubins) spanning open-source applications, vendor libraries, and optimized ML runtimes, CuLifter successfully lifts 99.98% of functions to valid LLVM IR. An ablation study confirms that type recovery is the only step required to produce semantically correct IR: disabling it drops the x86 pass rate from 73.8% to 0%, a 73.8 percentage-point drop. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_27486 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CuLifter: Lifting GPU Binaries to Typed IR Zhao, Jisheng Pu, Huanzhi Jeong, Shinnung Ahn, Chihyo Kim, Hyesoon Hardware Architecture GPU compilers merge all data types into a single unified register file, erasing the type information that binary-analysis tools rely on. We show that type recovery from this untyped register file is the central challenge of GPU binary lifting. We present CuLifter, a SASS-to-LLVM IR lifting framework that recovers register types via constraint propagation with conflict detection, reconstructs explicit control flow, and aggregates multi-instruction patterns. Across eight benchmark suites (24,437 GPU functions in 919 cubins) spanning open-source applications, vendor libraries, and optimized ML runtimes, CuLifter successfully lifts 99.98% of functions to valid LLVM IR. An ablation study confirms that type recovery is the only step required to produce semantically correct IR: disabling it drops the x86 pass rate from 73.8% to 0%, a 73.8 percentage-point drop. |
| title | CuLifter: Lifting GPU Binaries to Typed IR |
| topic | Hardware Architecture |
| url | https://arxiv.org/abs/2604.27486 |