BatchGen: An Architecture for Scalable and Efficient Batch Inference

• URL: http://arxiv.org/abs/2606.21712v1
• Date: Fri, 19 Jun 2026 19:56:21 GMT
• Title: BatchGen: An Architecture for Scalable and Efficient Batch Inference
• Abstract summary: Batch inference has become a central mode of AI computation, yet existing inference engines still rely on execution models designed for interactive serving.<n>We introduce a new architectural foundation for batch inference: the sequence compute model, which represents each sequence as a fine-grained, event-driven sequence.<n>This model exposes expressive primitives that allow the runtime to reorganize work dynamically, enabling larger expert-level batches, mitigating stragglers, real work across devices, and maintaining utilization even on cost-effective or memory-constrained GPU.
• Score: 7.794394498151309
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Batch inference has become a central mode of AI computation, yet existing inference engines still rely on execution models designed for interactive serving. When scaled to millions of sequences, batch workloads reveal two fundamental requirements: the ability to handle extreme inter- and intra-sequence load variation that emerges only at runtime, and the ability to sustain high utilization across large fleets of GPUs. Existing systems fail to meet these requirements, losing substantial fractions of achievable throughput. We introduce a new architectural foundation for batch inference: the sequence coroutine compute model, which represents each sequence as a fine-grained, event-driven coroutine. This model exposes expressive primitives that allow the runtime to reorganize work dynamically, enabling larger expert-level batches, mitigating stragglers, reallocating work across devices, and maintaining utilization even on cost-effective or memory-constrained GPUs. Building on this abstraction, we implement BatchGen, a production-ready system that uses the coroutine model at cluster scale. On a 128-GPU cluster, BatchGen reduces batch completion time by up to $2.3\times$, and on memory-constrained accelerators it outperforms the strongest offloading baseline by up to $9.6\times$. We will open-source BatchGen at https://github.com/batchgen-project/batchgen

Related papers

• Label-Free Cross-Task LoRA Merging with Null-Space Compression [50.63908869296697]
  We introduce Null-Space Compression (NSC) Merging, a label-free, output-agnostic method that sets merge weights from adapter geometry.<n>NSC achieves state-of-the-art performance across twenty heterogeneous vision tasks with balanced gains where prior methods overfit subsets of tasks.
  arXiv  Detail & Related papers  (2026-03-27T11:34:41Z)
• Scalable Training of Mixture-of-Experts Models with Megatron Core [26.9162079065285]
  Scaling Mixture-of-Experts (MoE) training introduces systems challenges absent in dense models.<n>Because each token activates only a subset of experts, this sparsity allows total parameters to grow much faster than per-token computation.<n>We address these challenges for MoE training through integrated optimizations spanning memory (fine-dimensional recomputation, offloading), communication, and computation.
  arXiv  Detail & Related papers  (2026-03-08T15:42:43Z)
• MELINOE: Fine-Tuning Enables Memory-Efficient Inference for Mixture-of-Experts Models [13.907916161242794]
  Mixture-of-Experts (MoE) model architectures can significantly reduce the number of activated parameters per token.<n>Their large overall parameter counts and model sizes have precluded their widespread usage in resource-constrained settings.<n>We present MELINOE, a method that fine-tunes an MoE model to more strongly prefer activating a smaller number of experts per sequence.
  arXiv  Detail & Related papers  (2026-01-30T14:40:18Z)
• Eliminating Multi-GPU Performance Taxes: A Systems Approach to Efficient Distributed LLMs [61.953548065938385]
  We introduce the ''Three Taxes'' (Bulk Synchronous, Inter- Kernel Data Locality, and Kernel Launch Overhead) as an analytical framework.<n>We propose moving beyond the rigid BSP model to address key inefficiencies in distributed GPU execution.<n>We observe a 10-20% speedup in end-to-end latency over BSP-based approaches.
  arXiv  Detail & Related papers  (2025-11-04T01:15:44Z)
• Visual Perception Engine: Fast and Flexible Multi-Head Inference for Robotic Vision Tasks [6.943057640797408]
  Visual Perception Engine (VPEngine) is a modular framework designed to enable efficient GPU usage for visual multitasking while maintaining developer accessibility.<n>Our framework architecture leverages a shared foundation model backbone that extracts image representations, which are efficiently shared, across multiple specialized task-specific model heads running in parallel.<n>Our example implementation demonstrates end-to-end real-time performance at $geq$50 Hz on NVIDIA Jetson Orin AGX forRT optimized models.
  arXiv  Detail & Related papers  (2025-08-15T16:42:23Z)
• Recurrent Diffusion for Large-Scale Parameter Generation [52.98888368644455]
  We introduce Recurrent Diffusion for Large Scale Generation (RPG), a novel framework that generates full neural network parameters up to hundreds of millions on a single GPU.<n>RPG serves as a critical advance in AI generating AI, potentially enabling efficient weight generation at scales previously deemed infeasible.
  arXiv  Detail & Related papers  (2025-01-20T16:46:26Z)
• The Streaming Batch Model for Efficient and Fault-Tolerant Heterogeneous Execution [28.768566833298365]
  We present Ray Data, a streaming batch system that improves throughput on heterogeneous batch inference pipelines by 2.5-12$times$.<n>Ray Data improves training throughput for multimodal models such as Stable Diffusion by 31% compared to single-node ML data loaders.
  arXiv  Detail & Related papers  (2025-01-16T19:54:01Z)
• Retentive Network: A Successor to Transformer for Large Language Models [91.6652200825638]
  We propose Retentive Network (RetNet) as a foundation architecture for large language models. We theoretically derive the connection between recurrence and attention. Experimental results on language modeling show that RetNet achieves favorable scaling results, parallel training, low-cost deployment, and efficient inference.
  arXiv  Detail & Related papers  (2023-07-17T16:40:01Z)
• SqueezeLLM: Dense-and-Sparse Quantization [80.32162537942138]
  Main bottleneck for generative inference with LLMs is memory bandwidth, rather than compute, for single batch inference. We introduce SqueezeLLM, a post-training quantization framework that enables lossless compression to ultra-low precisions of up to 3-bit. Our framework incorporates two novel ideas: (i) sensitivity-based non-uniform quantization, which searches for the optimal bit precision assignment based on second-order information; and (ii) the Dense-and-Sparse decomposition that stores outliers and sensitive weight values in an efficient sparse format.
  arXiv  Detail & Related papers  (2023-06-13T08:57:54Z)
• FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU [89.2451963569343]
  FlexGen is a generation engine for running large language model (LLM) inference on a single commodity GPU. When running OPT-175B on a single 16GB GPU, FlexGen achieves significantly higher throughput compared to state-of-the-art offloading systems. On the HELM benchmark, FlexGen can benchmark a 30B model with a 16GB GPU on 7 representative sub-scenarios in 21 hours.
  arXiv  Detail & Related papers  (2023-03-13T05:19:28Z)
• Efficient Strong Scaling Through Burst Parallel Training [13.656104138147967]
  Using large GPU clusters to train deep neural network (DNN) models is becoming an essential requirement. We present DeepPool, a system that addresses this efficiency challenge through two key ideas.
  arXiv  Detail & Related papers  (2021-12-19T05:18:39Z)
• Heterogeneous CPU+GPU Stochastic Gradient Descent Algorithms [1.3249453757295084]
  We study training algorithms for deep learning on heterogeneous CPU+GPU architectures. Our two-fold objective -- maximize convergence rate and resource utilization simultaneously -- makes the problem challenging. We show that the implementation of these algorithms achieves both faster convergence and higher resource utilization than on several real datasets.
  arXiv  Detail & Related papers  (2020-04-19T05:21:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.