implementation detail · filed under inference & serving
Split-K CuTe GEMM for low-batch Hyper-Connection Mix
A low-latency split-K CuTe GEMM used for Hyper-Connection Mix at small M.
Also called low-latency split-K CuTe GEMM.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- used
How sources treat it
One count per evidence span, weakest treatment to strongest.
used 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
For M ≤ 16, we use the low-latency split-K CuTe GEMM from FlashInfer PR #4266.
usedunclearin SGLangSGLang
Filed alongside
Other methods under inference & serving :: inference kernel.
Batch-invariant deterministic kernelsCUDA GraphCUDA Graph capture size reductionFlashAttention 3Kernel fusionMoE-side chunkingSynchronization-free static-shape MoE executionTensorRT-LLM multi-head attention backendAvoiding split-KDeepGEMM-based batch-invariant matrix multiplicationDistributed shared memory for cross-SM attention data exchangeDual-kernel batch-invariant attention decodingDynamic load balancingExp-free TopK kernelExpert-optimized Triton kernelsFA4 sheared-bias attention kernelFlashAttentionFlashAttention 4FlashKDAFP8 GEMMFused AttnRes merge and RMSNorm kernelFused latent down-projection and MoE-router GEMMFused mHC kernelsFused QSA kernel