All activity
Coding agents like Claude Code are great at generating GPU kernels if they’re told what your bottlenecks are. But they struggle to diagnose these performance bottlenecks themselves. The nCompass agent fills this gap by identifying key performance bottlenecks in the code and providing strategies to resolve them. By supercharging Claude Code with our agent, we generated matrix multiplication kernels that were 3% faster than NVIDIA's CUTLASS kernels in a single session. Now it’s your turn.
nCompass AI AssistantEnabling everyone to write GPU kernels
Aditya Rajagopalleft a comment
Hey PH! Aditya here, co-founder of nCompass. My team and I have spent years accelerating GPU kernels. It’s complex and time consuming, primarily because we spent most our time identifying which kernel to optimize and then identifying what the bottlenecks were in our new kernel. Using the dev tool we’re launching today, we implemented a Hopper GEMM kernel that outperformed NVIDIA's CUTLASS GEMMs...
nCompass AI AssistantEnabling everyone to write GPU kernels
Aditya Rajagopalstarted a discussion
We're building the tool to allow anyone to try their hand at building GPU kernels
Hey everyone! I think writing GPU code is going to be something that's going to become ubiquitous even though today it's a tough and time consuming process. Claude Code and Cursor agents have gotten great at writing GPU kernels once they know what to target. What's missing is the tooling to get them there. We're building that tooling at https://ncompass.tech. Our agent analyzes the profile and...
Aditya Rajagopalleft a comment
Congratulations on the launch! Looks really exciting!
CodeWordsTurn ideas into automations by chatting with AI