REVIEW 28 cited by
DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference
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
DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference
read the original abstract
The deployment and scaling of large language models (LLMs) have become critical as they permeate various applications, demanding high-throughput and low-latency serving systems. Existing frameworks struggle to balance these requirements, especially for workloads with long prompts. This paper introduces DeepSpeed-FastGen, a system that employs Dynamic SplitFuse, a novel prompt and generation composition strategy, to deliver up to 2.3x higher effective throughput, 2x lower latency on average, and up to 3.7x lower (token-level) tail latency, compared to state-of-the-art systems like vLLM. We leverage a synergistic combination of DeepSpeed-MII and DeepSpeed-Inference to provide an efficient and easy-to-use serving system for LLMs. DeepSpeed-FastGen's advanced implementation supports a range of models and offers both non-persistent and persistent deployment options, catering to diverse user scenarios from interactive sessions to long-running applications. We present a detailed benchmarking methodology, analyze the performance through latency-throughput curves, and investigate scalability via load balancing. Our evaluations demonstrate substantial improvements in throughput and latency across various models and hardware configurations. We discuss our roadmap for future enhancements, including broader model support and new hardware backends. The DeepSpeed-FastGen code is readily available for community engagement and contribution.
Forward citations
Cited by 28 Pith papers
-
SmoothAgent: Efficient Long-Horizon LLM-Based Agent Serving with Lookahead Context Engineering
SmoothAgent introduces lookahead context engineering to eliminate transformation overhead in LLM agents, reducing TTFT by up to 11.9x through proactive KV cache preparation.
-
Observation, Not Prediction: Conversation-Level Disaggregated Scheduling for Agentic Serving
Conversation-level scheduling in ConServe observes first-turn input length and KV occupancy to route prefill once and pin decoders, cutting p95 time-to-first-effective-token by 51% and improving energy efficiency by 7...
-
Characterizing Performance-Energy Trade-offs of Large Language Models in Multi-Request Workflows
This work delivers the first measurements of performance-energy trade-offs across four multi-request LLM workflow patterns on A100 GPUs using vLLM and Parrot.
-
PrefixWall: Mitigating Prefix Caching Side Channels in Shared LLM Systems
PrefixWall mitigates APC side channels in multi-tenant LLM systems via selective prefix isolation, delivering up to 70% higher cache reuse and 30% lower latency than full-isolation baselines.
-
MIST: A Co-Design Framework for Heterogeneous, Multi-Stage LLM Inference
MIST is a new simulator for heterogeneous multi-stage LLM inference that combines hardware traces with analytical models to explore configuration trade-offs in hybrid CPU-accelerator systems.
-
From Tensor Buffer to Distributed Memory Hierarchy: A Survey of KV Cache Management for LLM Serving
KV-cache serving systems concentrate into five archetypes under a four-axis taxonomy, with ownership explaining residual distributed design variance and seven measurement gaps blocking next steps.
-
Beyond Storage: State as a Runtime Control Problem in Parallel and Distributed Systems
State management across streaming, serving, retrieval, and learning systems is best modeled as one coupled runtime control loop spanning access, execution, and evolution.
-
BlockServe: Block-Grained Continuous Batching for High-Throughput Diffusion LLM Serving
BlockServe raises offline diffusion-LLM batch throughput 1.9–10.6× over Fast-dLLM by block-boundary eviction, mixed-state gather-scatter execution, and token-budget admission.
-
LiveServe: Interaction-Aware Serving for Real-Time Omni-Modal LLMs
LiveServe exposes audio playback and barge-in signals to the scheduler and KV manager, lowering P90 audio TTFP by 1.55x on average and raising completed-request throughput by 1.15x on two Omni-LMs.
-
ReMP: Low-Downtime Runtime Model-Parallelism Reconfiguration for LLM Serving
ReMP enables 1-7s runtime TP/PP topology switches in LLM inference with 10-100x speedups over restarts by decoupling parallelism from state and migrating KV cache, outperforming fixed configs on dynamic workloads.
-
Vortex: Efficient and Programmable Sparse Attention Serving for AI Agents
Vortex provides a programmable frontend and backend for sparse attention in LLM serving, delivering up to 3.46x throughput over full attention while preserving accuracy.
-
Beyond Greedy Chunking: SLO-Aware Sliding-Window Scheduling for LLM Inference
SlidingServe achieves up to 30% higher service capacity and 16-53% fewer SLO violations in LLM inference by using dynamic chunking and priority-based batch construction.
-
Federation of Experts: Communication Efficient Distributed Inference for Large Language Models
FoE restructures MoE blocks into per-KV-head clusters with sum-based synchronization, removing all-to-all communication in single-node settings and limiting it to intra-node in multi-node settings for up to 5.2x faste...
-
Scepsy: Serving Agentic Workflows Using Aggregate LLM Pipelines
Scepsy schedules arbitrary multi-LLM agentic workflows on GPU clusters by constructing Aggregate LLM Pipelines from stable per-LLM execution time shares, then searching fractional GPU allocations, tensor parallelism, ...
-
MARS: Efficient, Adaptive Co-Scheduling for Heterogeneous Agentic Systems
MARS coordinates heterogeneous GPU-CPU resources for agentic LLM workloads via decoupled admission control and agent-centric KV cache management, delivering up to 5.94x lower latency and 1.87x faster task completion.
-
FASTER: Rethinking Real-Time Flow VLAs
FASTER adds a Horizon-Aware Schedule to flow VLAs that compresses immediate-action denoising to one step while keeping long-horizon trajectory quality, lowering real-robot reaction latency.
-
FASTER: Rethinking Real-Time Flow VLAs
FASTER uses a horizon-aware flow sampling schedule to compress immediate-action denoising to one step, slashing effective reaction latency in real-robot VLA deployments.
-
WarmServe: Enabling One-for-Many GPU Prewarming for Multi-LLM Serving
WarmServe reduces tail TTFT by up to 50.8× versus autoscaling and supports 2.5× higher throughput than GPU-sharing by using one-for-many prewarming, model placement, KV cache reservation, and efficient tensor switching.
-
BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching
BatchLLM achieves 1.3x-10.8x higher throughput than vLLM and SGLang for batched LLM inference with prefix sharing via global prefix identification, decoding-first reordering, and memory-centric token batching.
-
HybridFlow: A Flexible and Efficient RLHF Framework
HybridFlow combines single- and multi-controller paradigms with a 3D-HybridEngine to deliver 1.53x to 20.57x higher throughput for various RLHF algorithms compared to prior systems.
-
Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving
AsymCache combines Multi-Segment Attention, position-aware eviction, and adaptive chunking to cut TTFT by up to 2.03x and TPOT by up to 1.71x versus recent baselines in LLM serving.
-
AlignedServe: Orchestrating Prefix-aware Batching to Build a High-throughput and Computing-efficient LLM Serving System
AlignedServe uses prefix-aware batching, large CPU in-flight request pools, batch scheduling, and GPU-to-GPU KV prefetching to raise decoding throughput up to 1.98x and cut latency up to 7.4x versus prior serving systems.
-
Focused Forcing: Content-Aware Per-Frame KV Selection for Efficient Autoregressive Video Diffusion
Focused Forcing is a training-free per-frame KV selection method that combines attention scores with diversity metrics and head-importance estimation to accelerate autoregressive video diffusion up to 1.48x while impr...
-
ServeGen: Workload Characterization and Generation of Large Language Model Serving in Production
ServeGen characterizes production LLM inference workloads across model types and generates realistic per-client composed workloads that reduce under-provisioning by 50% in a production validation.
-
MixLLM: LLM Quantization with Global Mixed-precision between Output-features and Highly-efficient System Design
MixLLM uses global output-feature importance to set mixed bit-widths for LLM quantization and adds two-step dequantization plus software pipelining for system efficiency.
-
Network Edge Inference for Large Language Models: Principles, Techniques, and Opportunities
A survey synthesizing challenges, system architectures, model optimizations, deployment methods, and resource management techniques for large language model inference at the network edge.
-
A Survey on Efficient Inference for Large Language Models
The paper surveys techniques to speed up and reduce the resource needs of LLM inference, organized by data-level, model-level, and system-level changes, with comparative experiments on representative methods.
-
Towards Efficient Large Vision-Language Models: A Comprehensive Survey on Inference Strategies
The paper surveys and taxonomizes inference optimization methods for large vision-language models across four categories while noting limitations and open problems.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.