CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Heakl, Ahmed, Stahl, Gustavo Bertolo, Hashmi, Sarim, Han, Seung Hun Eddie, Ranjan, Mukul, Kharlamova, Arina, Khan, Salman, Mahmoud, Abdulrahman
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917423802744832
author Heakl, Ahmed
Stahl, Gustavo Bertolo
Hashmi, Sarim
Han, Seung Hun Eddie
Ranjan, Mukul
Kharlamova, Arina
Khan, Salman
Mahmoud, Abdulrahman
author_facet Heakl, Ahmed
Stahl, Gustavo Bertolo
Hashmi, Sarim
Han, Seung Hun Eddie
Ranjan, Mukul
Kharlamova, Arina
Khan, Salman
Mahmoud, Abdulrahman
contents Cross-architecture GPU code transpilation is essential for unlocking low-level hardware portability, yet no scalable solution exists. We introduce CASS, the first dataset and model suite for source- and assembly-level GPU translation (CUDA <--> HIP, SASS <--> RDNA3). CASS contains 60k verified host-device code pairs, enabling learning-based translation across both ISA and runtime boundaries. We generate each sample using our automated pipeline that scrapes, translates, compiles, and aligns GPU programs across vendor stacks. Leveraging CASS, we train a suite of domain-specific translation models that achieve 88.2% accuracy on CUDA -> HIP and 69.1% on SASS -> RDNA3, outperforming commercial baselines including GPT-5.1, Claude-4.5, and Hipify by wide margins. Generated code matches native performance in 85% of cases, preserving both runtime and memory behavior. To support rigorous evaluation, we introduce CASS-Bench, a curated benchmark spanning 18 GPU domains with ground-truth execution. All data, models, and evaluation tools will be released as open source to support progress in GPU compiler tooling, binary compatibility, and LLM-guided code translation.
format Preprint
id arxiv_https___arxiv_org_abs_2505_16968
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark
Heakl, Ahmed
Stahl, Gustavo Bertolo
Hashmi, Sarim
Han, Seung Hun Eddie
Ranjan, Mukul
Kharlamova, Arina
Khan, Salman
Mahmoud, Abdulrahman
Hardware Architecture
Artificial Intelligence
Computation and Language
Machine Learning
Programming Languages
Cross-architecture GPU code transpilation is essential for unlocking low-level hardware portability, yet no scalable solution exists. We introduce CASS, the first dataset and model suite for source- and assembly-level GPU translation (CUDA <--> HIP, SASS <--> RDNA3). CASS contains 60k verified host-device code pairs, enabling learning-based translation across both ISA and runtime boundaries. We generate each sample using our automated pipeline that scrapes, translates, compiles, and aligns GPU programs across vendor stacks. Leveraging CASS, we train a suite of domain-specific translation models that achieve 88.2% accuracy on CUDA -> HIP and 69.1% on SASS -> RDNA3, outperforming commercial baselines including GPT-5.1, Claude-4.5, and Hipify by wide margins. Generated code matches native performance in 85% of cases, preserving both runtime and memory behavior. To support rigorous evaluation, we introduce CASS-Bench, a curated benchmark spanning 18 GPU domains with ground-truth execution. All data, models, and evaluation tools will be released as open source to support progress in GPU compiler tooling, binary compatibility, and LLM-guided code translation.
title CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark
topic Hardware Architecture
Artificial Intelligence
Computation and Language
Machine Learning
Programming Languages
url https://arxiv.org/abs/2505.16968