All challenges

Optimize a Triton vector-addition kernel for 2^20 CUDA elements.

Optimize a Triton vector-addition kernel for 2^28 CUDA elements.

Optimize a Triton vector-addition kernel for 2^24 CUDA elements.

Optimize ragged CUDA attention with per-row length masks.

Optimize packed INT4 quantized dot products on CUDA.

Optimize QK RMSNorm on CUDA tensors.

Optimize mixed-precision linear, bias, and GELU CUDA computation.

Optimize a CUDA/Triton implementation of the Mamba2 sequential scan recurrence.

Optimize batched CUDA matrix multiplication across grouped shapes.

Optimize Triton GEMM with GELU for transformer-like CUDA shapes.

Optimize Triton GEMM with GELU for square CUDA matrix shapes.

Optimize a Triton GEMM with GELU for tall/skinny and short/wide matrices.