30 Problems

Triton Kernels

Author Triton kernels - vector ops, fused reductions, tiled matmul, layernorm, and FlashAttention - and run them against torch references on real GPU hardware.

11 Easy15 Medium4 Hard

6 sections

Vector Addition Fused Multiply-Add ReLU GELU SiLU Triton Foundations and Block Model Quiz10 questions

Vector Sum Reduction Vector Max Reduction Single-Pass Mean and Variance L2 Vector Norm Fused Row-Wise Softmax Row-Wise LogSumExp Triton Reductions Quiz10 questions

Tiled Matrix Multiplication Autotuned Matrix Multiplication Tiled Transpose GEMV: Matrix Vector Product Fused Matmul + Bias + ReLU Triton Matrix Multiply and Autotuning Quiz10 questions

LayerNorm Forward RMSNorm Forward Dropout (Inverted Scaling)Cross Entropy Loss (Mean Reduction)LayerNorm Backward Triton Normalization and Training Ops Quiz10 questions

FlashAttention Forward Scaled Dot-Product Attention Rotary Position Embedding KV Cache Append Online Softmax Triton Attention and LLM Kernels Quiz10 questions

Grouped Program-ID Matmul Split-K Matmul Block-Pointer Matmul Vectorized Vector Add Triton Performance Patterns Quiz10 questions

Summary

Total30

Easy11

Medium15

Hard4

Progress0%

Quizzes0/6