AsyncTLS: Efficient Generative LLM Inference with Asynchronous Two-level Sparse Attention

arXiv:2604.07815v1 Announce Type: new Abstract: Long-context inference in LLMs faces the dual challenges of quadratic attention complexity and prohibitive KV cache memory. While token-level sparse attention offers superior accuracy, its indexing overhead is costly; block-level methods improve efficiency but sacrifice precision. We propose AsyncTLS, a hierarchical sparse attention system that combines coarse-grained block filtering with fine-grained token selection to balance accuracy and efficiency, coupled with an asynchronous offloading engine that overlaps KV cache transfers with computation via temporal locality exploitation. Evaluated on Qwen3 and GLM-4.7-Flash across GQA, and MLA architectures, AsyncTLS achieves accuracy comparable to full attention while delivering 1.2x - 10.0x operator speedups and 1.3x - 4.7x end-to-end throughput improvements on 48k - 96k contexts.

Related

• Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference
• LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification
• DIVERSED: Relaxed Speculative Decoding via Dynamic Ensemble Verification
• DYCP: Dynamic Context Pruning for Long-Form Dialogue with LLMs
• SAT: Balancing Reasoning Accuracy and Efficiency with Stepwise Adaptive Thinking

Source: arXiv cs.CL | 2026-04-10