Kernel Looping: Eliminating Synchronization Boundaries for Peak Inference Performance

• URL: http://arxiv.org/abs/2410.23668v1
• Date: Thu, 31 Oct 2024 06:32:47 GMT
• Title: Kernel Looping: Eliminating Synchronization Boundaries for Peak Inference Performance
• Abstract summary: Kernel looping exploits the unique layer-level fusion possible in modern dataflow architectures with the repeated layer structure found in language models. We evaluate kernel looping on the SambaNova SN40L Reconfigurable Dataflow Unit (RDU), a commercial dataflow accelerator for AI.
• Score: 2.747566636546902
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Token generation speed is critical to power the next wave of AI inference applications. GPUs significantly underperform during token generation due to synchronization overheads at kernel boundaries, utilizing only 21% of their peak memory bandwidth. While recent dataflow architectures mitigate these overheads by enabling aggressive fusion of decoder layers into a single kernel, they too leave performance on the table due to synchronization penalties at layer boundaries. This paper presents kernel looping, a specialized global optimization technique which exploits an optimization opportunity brought by combining the unique layer-level fusion possible in modern dataflow architectures with the repeated layer structure found in language models. Kernel looping eliminates synchronization costs between consecutive calls to the same kernel by transforming these calls into a single call to a modified kernel containing a pipelined outer loop. We evaluate kernel looping on the SambaNova SN40L Reconfigurable Dataflow Unit (RDU), a commercial dataflow accelerator for AI. Experiments demonstrate that kernel looping speeds up the decode phase of a wide array of powerful open-source models by up to 2.2$\times$ on SN40L. Kernel looping allows scaling of decode performance over multiple SN40L sockets, achieving speedups of up to 2.5$\times$. Finally, kernel looping enables SN40L to achieve over 90% of peak performance on 8 and 16 sockets and achieve a speedup of up to 3.7$\times$ over DGX H100. Kernel looping, as well as the models evaluated in this paper, are deployed in production in a commercial AI inference cloud.

Related papers

• mKernel: Fast Multi-GPU, Multi-Node Fused Kernels [37.86216303017995]
  mKernel is a library of multi-GPU, multi-node fused kernels that overlap computation, intra-node NVLink communication, and inter-node RDMA at tile granularity.<n>It achieves speedups of up to 1.72x on GEMM+AllReduce and $1.88times$ on Ring Attention.
  arXiv  Detail & Related papers  (2026-09-11T22:50:04Z)
• TurboT2VA: Fast Large-Scale Text-to-Video-Audio Generation via Score-Regularized Consistency Distillation [105.14745366152725]
  TurboT2VA is a distillation and inference framework for joint text-to-video-audio generation.<n>On LTX-2, four-step distillation reduces generator latency from 50.52s to 2.51s at the standard evaluation resolution of 512$times$768.<n>The complete stack reduces generator latency from 318.74s to 5.83s on one NVIDIA H20, achieving a 54.67$times$ generator-only speedup.
  arXiv  Detail & Related papers  (2026-08-25T15:09:02Z)
• FluxBin: Flexible LUT-based Ultra-low-bit LLM Inference by Algorithm-Kernel Synergy [32.21687694159134]
  FluxBin is an algorithm that synerizes binary quantization with a highly optimized kernel.<n>FluxBin achieves up to $5.92times speedup and $10.19times energy savings.<n>This effectively enables the deployment of 70B-scale models on one single A100 bases with a $4times memory reduction.
  arXiv  Detail & Related papers  (2026-08-16T08:01:06Z)
• KForge: LLM-Driven Cross-Platform Kernel Generation for AI Accelerators [5.703659894138959]
  Production inference increasingly targets a heterogeneous mix of accelerators.<n>This creates a systems challenge: each pipeline now requires high-performance kernels across a growing set of hardware backends and programming models.<n>We present KForge, a cross-platform framework built around an iterative refinement loop driven by two collaborating LLM-based agents.
  arXiv  Detail & Related papers  (2026-06-01T23:48:55Z)
• Xe-Forge: Multi-Stage LLM-Powered Kernel Optimization for Intel GPU [38.81440160993858]
  Porting deep learning algorithms to new hardware accelerators requires developers to apply the same low-level optimizations to every Triton kernel in their code-base.<n>We present Xe-Forge, a multi-stage LLM-powered pipeline that automates this process for Intel GPU.<n>A curated knowledge base encodes Intel GPU constraints that are absent from LLM training data, keeping the model within architecturally valid bounds.
  arXiv  Detail & Related papers  (2026-04-16T20:21:01Z)
• Towards Cold-Start Drafting and Continual Refining: A Value-Driven Memory Approach with Application to NPU Kernel Synthesis [68.7701048879757]
  EvoKernel is a self-evolving agentic framework that automates the lifecycle of kernel synthesis.<n>It learns stage-specific Q-values that prioritize experiences based on their contribution to the current objective.<n>It improves models' correctness from 11.0% to 83.0% and a median speedup of 3.60x over initial drafts.
  arXiv  Detail & Related papers  (2026-03-11T14:57:06Z)
• AscendCraft: Automatic Ascend NPU Kernel Generation via DSL-Guided Transcompilation [8.878393510726008]
  We present AscendCraft, a DSL-guided approach for automatic AscendC kernel generation.<n>AscendingCraft achieves 98.1% compilation success and 90.4% functional correctness.<n>We also demonstrate that DSL-guided transcompilation can enable LLMs to generate both correct and competitive NPU kernels.
  arXiv  Detail & Related papers  (2026-01-30T09:34:59Z)
• 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)
• CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization [36.794824560677064]
  CudaForge is a training-free multi-agent workflow for kernel generation and optimization.<n>By leveraging base models like OpenAI-o3, CudaForge achieves 97.6% correctness generated kernels and an average 1.68$times$ speedup.
  arXiv  Detail & Related papers  (2025-10-23T22:52:00Z)
• dParallel: Learnable Parallel Decoding for dLLMs [77.24184219948337]
  Diffusion large language models (dLLMs) offer parallel token prediction and lower inference latency.<n>Existing open-source models still require nearly token-length decoding steps to ensure performance.<n>We introduce dParallel, a simple and effective method that unlocks the inherent parallelism of dLLMs for fast sampling.
  arXiv  Detail & Related papers  (2025-09-30T16:32:52Z)
• MARLIN: Mixed-Precision Auto-Regressive Parallel Inference on Large Language Models [58.3342517278868]
  This paper describes the design of Mixed-precision AutoRegressive LINear kernels. It shows that batchsizes up to 16-32 can be supported with close to maximum ($4times$) quantization speedup. MarLIN accomplishes this via a combination of techniques, such as asynchronous memory access, complex task scheduling and pipelining.
  arXiv  Detail & Related papers  (2024-08-21T16:10:41Z)
• AsyncDiff: Parallelizing Diffusion Models by Asynchronous Denoising [49.785626309848276]
  AsyncDiff is a universal and plug-and-play acceleration scheme that enables model parallelism across multiple devices. For the Stable Diffusion v2.1, AsyncDiff achieves a 2.7x speedup with negligible degradation and a 4.0x speedup with only a slight reduction of 0.38 in CLIP Score. Our experiments also demonstrate that AsyncDiff can be readily applied to video diffusion models with encouraging performances.
  arXiv  Detail & Related papers  (2024-06-11T03:09:37Z)
• FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion [9.5114389643299]
  This paper proposes a novel method, Flux, to significantly hide communication latencies with dependent computations for GPUs. Flux can potentially overlap up to 96% of communication given a fused kernel. Overall, it can achieve up to 1.24x speedups for training over Megatron-LM on a cluster of 128 GPU with various GPU generations and interconnects.
  arXiv  Detail & Related papers  (2024-06-11T00:17:39Z)
• Kernel-U-Net: Multivariate Time Series Forecasting using Custom Kernels [1.8816077341295625]
  We introduce Kernel-U-Net, a flexible and kernel-customizable U-shape neural network architecture. Specifically, Kernel-U-Net separates the procedure of partitioning input time series into patches from kernel manipulation. Our method offers two primary advantages: 1) Flexibility in kernel customization to adapt to specific datasets; and 2) Enhanced computational efficiency, with the complexity of the Transformer layer reduced to linear.
  arXiv  Detail & Related papers  (2024-01-03T00:49:51Z)
• Efficient Dataset Distillation Using Random Feature Approximation [109.07737733329019]
  We propose a novel algorithm that uses a random feature approximation (RFA) of the Neural Network Gaussian Process (NNGP) kernel. Our algorithm provides at least a 100-fold speedup over KIP and can run on a single GPU. Our new method, termed an RFA Distillation (RFAD), performs competitively with KIP and other dataset condensation algorithms in accuracy over a range of large-scale datasets.
  arXiv  Detail & Related papers  (2022-10-21T15:56:13Z)
• Multiple Kernel Clustering with Dual Noise Minimization [56.009011016367744]
  Multiple kernel clustering (MKC) aims to group data by integrating complementary information from base kernels. In this paper, we rigorously define dual noise and propose a novel parameter-free MKC algorithm by minimizing them. We observe that dual noise will pollute the block diagonal structures and incur the degeneration of clustering performance, and C-noise exhibits stronger destruction than N-noise.
  arXiv  Detail & Related papers  (2022-07-13T08:37:42Z)
• FlexConv: Continuous Kernel Convolutions with Differentiable Kernel Sizes [34.90912459206022]
  Recent works show CNNs benefit from different kernel sizes at different layers, but exploring all possible combinations is unfeasible in practice. We propose FlexConv, a novel convolutional operation with which high bandwidth convolutional kernels of learnable kernel size can be learned at a fixed parameter cost.
  arXiv  Detail & Related papers  (2021-10-15T12:35:49Z)
• Kernel Identification Through Transformers [54.3795894579111]
  Kernel selection plays a central role in determining the performance of Gaussian Process (GP) models. This work addresses the challenge of constructing custom kernel functions for high-dimensional GP regression models. We introduce a novel approach named KITT: Kernel Identification Through Transformers.
  arXiv  Detail & Related papers  (2021-06-15T14:32:38Z)
• PolyScientist: Automatic Loop Transformations Combined with Microkernels for Optimization of Deep Learning Primitives [55.79741270235602]
  We develop a hybrid solution to the development of deep learning kernels. We use the advanced polyhedral technology to automatically tune the outer loops for performance.
  arXiv  Detail & Related papers  (2020-02-06T08:02:34Z)

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.