Pith. sign in

REVIEW 4 cited by

Seesaw: High-throughput LLM Inference via Model Re-sharding

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2503.06433 v1 pith:WN6AD3D3 submitted 2025-03-09 cs.DC cs.AI

classification cs.DCcs.AI
keywords inferenceseesawefficiencymodelparallelizationre-shardingstagescomputational
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

To improve the efficiency of distributed large language model (LLM) inference, various parallelization strategies, such as tensor and pipeline parallelism, have been proposed. However, the distinct computational characteristics inherent in the two stages of LLM inference-prefilling and decoding-render a single static parallelization strategy insufficient for the effective optimization of both stages. In this work, we present Seesaw, an LLM inference engine optimized for throughput-oriented tasks. The key idea behind Seesaw is dynamic model re-sharding, a technique that facilitates the dynamic reconfiguration of parallelization strategies across stages, thereby maximizing throughput at both phases. To mitigate re-sharding overhead and optimize computational efficiency, we employ tiered KV cache buffering and transition-minimizing scheduling. These approaches work synergistically to reduce the overhead caused by frequent stage transitions while ensuring maximum batching efficiency. Our evaluation demonstrates that Seesaw achieves a throughput increase of up to 1.78x (1.36x on average) compared to vLLM, the most widely used state-of-the-art LLM inference engine.

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Attention to Detail: Evaluating Energy, Performance, and Accuracy Trade-offs Across vLLM Configurations

    cs.SE 2026-07 conditional novelty 6.0 of 10

    vLLM attention kernel and prefix caching drive model- and task-dependent energy and latency effects, with no universal best config, and can unexpectedly shift measured accuracy.

  2. Learning to Shard: RL for Co-optimizing the Parallelism Degrees and Per-operator Sharding Dimensions in Distributed LLM Inference

    cs.LG 2025-08 conditional novelty 6.0 of 10

    An RL agent that co-optimizes parallelism degrees and per-operator sharding dimensions finds distributed inference strategies that beat random search and simulated annealing, and slightly outperform Megatron-LM heuris...

  3. Military AI Cyber Agents (MAICAs) Constitute a Global Threat to Critical Infrastructure

    cs.CY 2025-06 conditional novelty 6.0 of 10

    Autonomous AI cyber agents could credibly cause catastrophic damage to critical infrastructure by self-replicating and operating across global networks, according to this risk analysis.

  4. Hardware-Efficient Attention for Fast Decoding

    cs.LG 2025-05 conditional novelty 6.0 of 10

    Grouped-Tied Attention and Grouped Latent Attention reduce KV-cache memory and speed up LLM decoding by up to 2x while matching the quality of GQA and MLA at up to 1.47B parameters.

Pith tools