Round two just dropped — the stakes are higher, the kernels need to be tighter.

Teaming up with a crew known for pushing GPU performance to absurd levels, we're throwing down a challenge that'll separate the optimization wizards from the weekend coders. This time? Low-precision matrix multiplication using NVFP4.

If you've been itching to prove your chops in kernel engineering, here's your arena. We're talking real-world performance constraints, hardcore throughput demands, and the kind of problem that makes your compiler sweat.

📌 Task: NVFP4 GEMM optimization
⏰ Submissions close: December 19th

No hand-holding. No shortcuts. Just you, your code, and the clock.