Faizan Haider@faizan_haider_690

PortForge is an AI-powered CUDA-to-HIP migration platform that automatically converts NVIDIA GPU code to AMD ROCm, compiles it on real AMD MI300X hardware, and delivers validated benchmark results — reducing weeks of manual migration work to minutes. The Problem NVIDIA's CUDA ecosystem has dominated GPU computing for 15+ years, creating massive developer lock-in. When developers evaluate AMD hardware, they hit an immediate wall: migrating existing CUDA code to AMD's ROCm/HIP platform is slow and painful. AMD's official tool hipify-clang only does mechanical syntax substitution — it doesn't understand semantics, compile output, or benchmark performance. Developers are left to do the hard part manually. What PortForge Does PortForge automates the entire migration pipeline. A developer uploads any CUDA kernel file and PortForge handles everything automatically: our AST Parser scans the code and maps every CUDA-specific call that needs conversion. Our LLM Migration Agent uses DeepSeek via Fireworks AI to semantically understand and rewrite the code as valid HIP. The migrated code is then compiled using hipcc on a real AMD MI300X via ROCm 7.2.4, and benchmarked using rocprof — returning actual hardware timing measurements. Real Results on AMD MI300X We validated on 5 CUDA kernels from NVIDIA's official cuda-samples repository. vectorAdd compiled and ran in 7,778 ns. warpAggregatedAtomicsCG ran in 1,485,581 ns. 3 of 5 kernels achieved full compilation with zero manual intervention. AMD Technology Used AMD MI300X, ROCm 7.2.4, hipcc, rocprof, AMD Developer Cloud, Fireworks AI API. Market Potential Every enterprise evaluating AMD GPUs faces this migration challenge. PortForge has a clear SaaS path: pay-per-kernel API for organizations migrating CUDA workloads to AMD infrastructure.

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.