{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/5bb95731f8294bb39084adae3ca1571b\" frameborder=\"0\" width=\"1730\" height=\"1297\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1297,"width":1730,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1297,"thumbnail_width":1730,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/5bb95731f8294bb39084adae3ca1571b-4c6d7c5458cda4f9.gif","duration":123.659,"title":"[Feature Tutorial] AI Agent Features for Kernel Analysis and Comparison","description":"In this video, I demonstrate our AI agent features that integrate with NC Rep, allowing developers to analyze and compare their kernels. I show how to add an NC Rep file to the chat context, which automatically extracts the SAS, PTX, and session details for analysis. I specifically compare two kernels, checking if they are persistent and the number of pipeline stages, revealing that my kernel has a four-stage pipeline while the Nvidia kernel's complexity requires further analysis. This is a work in progress, and I encourage you to explore the next steps or guide the chat in a different direction if needed."}