arXiv AI By Tim Beringer (Technical University of Darmstadt), Patrick Diem (Technical University of Darmstadt), Felix Wolf (Technical University of Darmstadt), Arya Mazaheri (Technical University of Darmstadt, PanocularAI)

ShardMeter: Sharded and Geo-Distributed Training Without the Guesswork

Read the original on arXiv AI →

ShardMeter is a lightweight analytical performance model that predicts end-to-end runtime for transformer-based workloads across sharded, distributed, and decentralized training setups. By taking a model’s characteristics and a target hardware topology as input, it estimates per-GPU and per-island throughput, training cost, total wall-clock time, and pinpoints performance bottlenecks. The model reveals diminishing-return regimes with increasing island size, quantifies compute- versus communication-bound scaling, evaluates hyperparameter trade-offs, and models cost-throughput for large-scale decentralized training, enabling rapid exploration of configuration space and near-optimal deployment plans.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

arXiv AI
Aug 20

Pre-Compiled Pipeline Shards for Distributed LLM Inference on Intel AI PC Fleets

The paper presents a method for distributing large language model inference across multiple Intel AI PCs by splitting the model into pipeline shards, each pre‑compiled into an OpenVINO graph. Three key techniques—beam_idx Gather to enable GPU optimizations, speculative decoding on stateful models, and interleaved micro‑batching—allow a two‑node Llama 3.1 8B INT4 pipeline to serve two users at 1.79× the throughput of a single‑node model, while a four‑node deployment can run a 70B model that no single PC can hold. The authors provide code, benchmark logs, and reproduction scripts on GitHub.

By Tate Berenbaum, Muthaiah Venkatachalam
arXiv AI
Jul 7

Data Driven Optimization of GPU efficiency for Distributed LLM-Adapter Serving

arXiv:2602. 24044v2 Announce Type: replace-cross Abstract: Large Language Model (LLM) adapters enable low-cost model specialization, but introduce complex caching and scheduling challenges in distributed serving systems where hundreds of adapters must be hosted concurrently.

By Ferran Agullo, Joan Oliveras, Chen Wang, Alberto Gutierrez-Torre, Olivier Tardieu, Alaa Youssef, Jordi Torres, Josep Ll. Berral
Hugging Face Trending Papers
5d ago

Accelerating Sharded Data Parallelism at Scale with Federated Learning

The paper proposes two hybrid algorithms, FL+FSDP and FL+HSDP, that combine sharded data parallelism with federated learning-style aggregations to reduce communication overhead in large-scale AI training. By forming loosely‑coupled federation groups, the methods keep inter‑group traffic minimal while maintaining a bounded global batch size. Experiments on a Llama3.1 8B model trained on 512 A100 GPUs show up to 8.04× faster data processing and 4.48 lower evaluation perplexity compared to traditional sharded DP approaches.

arXiv AI
5d ago

Accelerating Sharded Data Parallelism at Scale with Federated Learning

The paper proposes two hybrid algorithms, FL+FSDP and FL+HSDP, that combine sharded data parallelism with federated learning-style aggregations to reduce communication overhead in large-scale AI training. By partitioning GPUs into loosely‑coupled federation groups, the methods keep inter‑group traffic minimal while maintaining a bounded global batch size. Experiments on a Llama3.1 8B model trained on 512 A100 GPUs show up to 8.04× faster data processing and 4.48 lower evaluation perplexity compared to conventional sharded DP.

By Gianluca Mittone, Marco Aldinucci
arXiv AI
Sep 12

Characterizing Job Power Elasticity for Power-Flexible AI Training

The paper introduces the Power Flexibility Index (PFI) to measure how large language model (LLM) training performance changes when GPU power is reduced. Using 131 training runs on H200 and H100 GPUs, the study finds that LLM jobs have significant but variable power elasticity and identifies telemetry signals that can predict PFI during runtime. The authors demonstrate that allocating power based on PFI maximizes overall token throughput, recovering about 1.5k tokens/s per job under a 30% power reduction, which represents 63% of the gap between equal-weight and perfect-information allocations.

By Philip Colangelo, Charles Dawson, Shayan Sengupta, Ayse Coskun, Varun Sivaram