arXiv AI
Jul 3
Towards Load-Aware Prefill Deflection for Disaggregated LLM Serving
arXiv:2607. 02043v1 Announce Type: cross Abstract: Disaggregated LLM serving runs prefill and decode on separate GPU pools to keep the two phases from interfering.
Fine-Grained Computation Offload for Off-the-Shelf Servers in Tens of Lines
arXiv:2607. 02630v1 Announce Type: cross Abstract: Hardware accelerators now sit on the critical path of online serving.
Related stories
arXiv AI
3d ago
Decode-Latency Feedback Prefill: A Model-Free Controller and Its Generalization Limits
The paper introduces Decode‑Latency Feedback Prefill (DLFP), a model‑free controller that adjusts prefill chunk sizes during concurrent autoregressive inference to reduce interference between new and ongoing requests. Implemented in vLLM, DLFP achieves significant reductions in P99 inter‑token latency on Qwen3‑0.6B while maintaining output correctness and SLO compliance, though it fails to generalize to larger models or multi‑GPU setups. The study highlights the limits of this approach and suggests the need for a completion‑timed controller for broader applicability.
arXiv Machine Learning
Jun 2
Design Space Exploration of DMA based Finer-Grain Compute Communication Overlap
arXiv:2512. 10236v2 Announce Type: replace-cross Abstract: Modern ML workloads demand distributing training and inference across multiple GPUs.
arXiv AI
4d ago
SPLASH: Switching Parallel Layouts of Attention with Seamless Handoff for LLM Serving
arXiv:2609.37626v1 Announce Type: cross Abstract: No single way of parallelizing attention serves large language models well under all loads. Low concurrency favors tensor parallelism, many independe...
Hugging Face Trending Papers
Aug 6
Operating Multi-Node Full Fine-Tuning on NVIDIA B300: A Field Report on Telemetry-Based Triage, Negative Results, and Operational Hardening
We report operational experience full-fine-tuning a 32. 76B-parameter dense model (Qwen3-32B) on 16 x NVIDIA B300 (two nodes, FSDP / ZeRO-3) -- among the first published field accounts on this accelerator.
arXiv Machine Learning
Aug 7
Operating Multi-Node Full Fine-Tuning on NVIDIA B300: A Field Report on Telemetry-Based Triage, Negative Results, and Operational Hardening
arXiv:2608. 05944v1 Announce Type: cross Abstract: We report operational experience full-fine-tuning a 32.
arXiv AI
Sep 15
X-Stage: Modeling Post-Issue Backpressure in GPU Communication--Computation Fusion
arXiv:2607.23264v2 Announce Type: replace-cross Abstract: Fine-grained, device-initiated communication allows fused GPU kernels to issue remote stores directly from their compute pipelines, a pattern...
arXiv AI
Aug 13
Ready Cohorts: Bounding GPU Opportunity and Avoiding Host Round Trips in LLM-Agent Control
arXiv:2608. 12123v1 Announce Type: cross Abstract: LLM-agent services repeatedly execute small deterministic transitions between model and tool calls: route an outcome, update state, and emit the next effect.
arXiv AI
Jul 7
Elastic Gang: Per-Token Membership Change for a Hard-Barriered LLM Inference Gang Co-Scheduled with OS Processes
arXiv:2607. 04668v1 Announce Type: cross Abstract: On-device LLM decoding is a hard-barriered CPU-SIMD computation that wants every core for milliseconds per token, while the rest of the OS wants those same cores continuously.
arXiv AI
Aug 26
Elastic KV Cache for LLM Serving:A Working Reclamation Mechanism, and Why Chunked Prefill Already Closes the Gap
The paper introduces an elastic key‑value (KV) cache for large language model (LLM) serving that dynamically reclaims a pre‑allocated reserve during decode‑heavy phases and restores it before prefill, using a userspace CUDA virtual‑memory trick that requires no driver changes. The authors implement this mechanism, test it under realistic workloads, and find that it offers only marginal benefits—about a 1 % difference in time‑to‑first‑token for large prefill chunks—and that simpler strategies such as lowering the maximum batch size can achieve similar results. The study also notes that the reserve’s impact diminishes with higher tensor‑parallelism levels. whyItMatters":"The work demonstrates that a dynamic KV cache reclamation strategy can be implemented without driver patches and that its practical benefits are limited, guiding future LLM serving optimizations toward simpler approaches."
arXiv AI
Aug 14
TEMPO: Makespan-Aware Expert-Parallel Load Balancing Across Memory- and Compute-Bound Regimes
arXiv:2608. 13057v1 Announce Type: cross Abstract: In expert-parallel (EP) MoE serving, every layer synchronizes at the slowest GPU.
arXiv AI
Jul 21
Talaria: Session-Aware Serverless Serving of Hundred-Billion-Parameter LLMs
arXiv:2607. 17181v1 Announce Type: cross Abstract: Serverless multi-model LLM systems multiplex popularity-skewed model catalogs over shared GPU pools, yet typically schedule each request independently.