🔥🔥[FFPA] FFPA: Yet another Faster Flash Prefill Attention with O(1) S…

…RAM complexity for headdim > 256, ~1.5x faster than SDPA EA(@DefTruth) (#111)

🔥🔥[FFPA] FFPA: Yet another Faster Flash Prefill Attention with O(1) SRAM complexity for headdim > 256, ~1.5x faster than SDPA EA(@DefTruth)

🔥🔥[HADACORE] HADACORE: TENSOR CORE ACCELERATED HADAMARD TRANSFORM KER…

…NEL (#108)

🔥[BatchLLM] BatchLLM: Optimizing Large Batched LLM Inference with Glo…

…bal Prefix Sharing and Throughput-oriented Token Batching (#104)

🔥[BatchLLM] BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching

🔥[KV Cache Recomputation] Efficient LLM Inference with I/O-Aware Part…

…ial KV Cache Recomputation (#102)

🔥[SparseInfer] SparseInfer: Training-free Prediction of Activation Sp…

…arsity for Fast LLM Inference (#100)

🔥🔥[TP: Comm Compression] Communication Compression for Tensor Paralle…

…l LLM Inference (#94)

🔥🔥[TP: Comm Compression] Communication Compression for Tensor Parallel LLM Inference

🔥[VL-CACHE] VL-CACHE: SPARSITY AND MODALITY-AWARE KV CACHE COMPRESSIO…

…N FOR VISION-LANGUAGE MODEL INFERENCE ACCELERATION

🔥[Tensor Product] Acceleration of Tensor-Product Operations with Tens…

…or Cores (#90)

🔥[FastAttention] FastAttention: Extend FlashAttention2 to NPUs and Lo…

…w-resource GPUs for Efficient Inference (#88)

🔥[PARALLELSPEC] PARALLELSPEC: PARALLEL DRAFTER FOR EFFICIENT SPECULAT…

…IVE DECODING (#84)