Performant Automatic BLAS Offloading on Unified Memory Architecture with OpenMP First-Touch Style Data Movement

• URL: http://arxiv.org/abs/2501.00279v3
• Date: Tue, 15 Apr 2025 15:40:25 GMT
• Title: Performant Automatic BLAS Offloading on Unified Memory Architecture with OpenMP First-Touch Style Data Movement
• Authors: Junjie Li,
• Abstract summary: This paper introduces SCILIB-Accel, a novel tool for automatic BLAS offload.<n>The tool intercepts BLAS symbols directly from a CPU binary, requiring no code modifications or recompilation.<n> SCILIB-Accel has been evaluated using multiple quantum physics codes on up to a few hundred GPU nodes, yielding promising speedups.
• Score: 16.464496913614315
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: BLAS is a fundamental building block of advanced linear algebra libraries and many modern scientific computing applications. GPUs are known for their strong arithmetic computing capabilities and are highly suited for BLAS operations. However, porting code to GPUs often requires significant effort, especially for large, complex codes or legacy codes, even for BLAS-heavy applications. While various tools exist to automatically offload BLAS to GPUs, they are often impractical due to the high costs associated with mandatory data transfers. The advent of unified memory architectures in recent GPU designs, such as the NVIDIA Grace-Hopper, allows cache-coherent memory access across all types of memory for both CPU and GPU, potentially eliminating the bottlenecks faced in conventional architectures. This breakthrough paves the way for innovative application developments and porting strategies. Building on our preliminary work demonstrating the potential of automatic *gemm offload, this paper extends the framework to all level-3 BLAS operations and introduces SCILIB-Accel, a novel tool for automatic BLAS offload. SCILIB-Accel leverages the memory coherency in Grace-Hopper and introduces a Device First-Use data movement policy inspired by the OpenMP First-Touch approach in multi-socket CPU programming, minimizing CPU-GPU data transfers for typical scientific computing codes. Additionally, utilizing dynamic binary instrumentation, the tool intercepts BLAS symbols directly from a CPU binary, requiring no code modifications or recompilation. SCILIB-Accel has been evaluated using multiple quantum physics codes on up to a few hundred GPU nodes, yielding promising speedups. Notably, for the LSMS method in the MuST suite, a 3x speedup was achieved on Grace-Hopper compared to Grace-Grace.

Related papers

• AIvailable: A Software-Defined Architecture for LLM-as-a-Service on Heterogeneous and Legacy GPUs [0.5863360388454261]
  We introduce AIvailable, a low-cost, highly available LLM-as-a-Service (LLM) platform.<n>It uses a software-defined approach for running LLMs across heterogeneous and legacy GPU nodes.<n>It features a unified client interface that allows seamless interaction with all deployed LLMs.
  arXiv  Detail & Related papers  (2025-11-06T14:19:57Z)
• LightMem: Lightweight and Efficient Memory-Augmented Generation [72.21680105265824]
  We introduce a new memory system called LightMem, which strikes a balance between the performance and efficiency of memory systems.<n>Inspired by the Atkinson-Shiffrin model of human memory, LightMem organizes memory into three complementary stages.<n>Experiments on LongMemEval with GPT and Qwen backbones show that LightMem outperforms strong baselines in accuracy (up to 10.9% gains) while reducing token usage by up to 117x.
  arXiv  Detail & Related papers  (2025-10-21T17:58:17Z)
• AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM [60.583864162256525]
  We propose a scalable, effective, and general way to augment large language models with billion-scale knowledge graphs.<n> KG2KV and HiKVP integrate KG triples into LLMs at scale with sub-linear time and memory complexity.<n>It maintains strong knowledge grounding and generalization performance using the LLMs' inherent attention mechanism.
  arXiv  Detail & Related papers  (2025-10-20T15:40:14Z)
• Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference [4.497936996651617]
  Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. In this paper, through an in-depth GPU-level analysis, we reveal that large-batch inference remains memory-bound, with most GPU compute capabilities underutilized due to DRAM bandwidth saturation as the primary bottleneck.
  arXiv  Detail & Related papers  (2025-03-11T11:21:35Z)
• A Universal Framework for Compressing Embeddings in CTR Prediction [68.27582084015044]
  We introduce a Model-agnostic Embedding Compression (MEC) framework that compresses embedding tables by quantizing pre-trained embeddings. Our approach consists of two stages: first, we apply popularity-weighted regularization to balance code distribution between high- and low-frequency features. Experiments on three datasets reveal that our method reduces memory usage by over 50x while maintaining or improving recommendation performance.
  arXiv  Detail & Related papers  (2025-02-21T10:12:34Z)
• MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs [55.95879347182669]
  MoE architecture is renowned for its ability to increase model capacity without a proportional increase in inference cost. MoE-Lightning introduces a novel CPU-GPU-I/O pipelining schedule, CGOPipe, with paged weights to achieve high resource utilization. MoE-Lightning can achieve up to 10.3x higher throughput than state-of-the-art offloading-enabled LLM inference systems for Mixtral 8x7B on a single T4 GPU (16GB)
  arXiv  Detail & Related papers  (2024-11-18T01:06:12Z)
• Deep Optimizer States: Towards Scalable Training of Transformer Models Using Interleaved Offloading [2.8231000588510757]
  Transformers and large language models(LLMs) have seen rapid adoption in all domains. Training of transformers is very expensive and often hits a memory wall'' We propose a novel technique to split the LLM into subgroups, whose update phase is scheduled on either the CPU or the GPU.
  arXiv  Detail & Related papers  (2024-10-26T00:43:59Z)
• vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving [53.972175896814505]
  Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests.
  arXiv  Detail & Related papers  (2024-07-22T14:37:58Z)
• Machine-Learning-Driven Runtime Optimization of BLAS Level 3 on Modern Multi-Core Systems [0.0]
  We present an extension to the Architecture and Data-Structure Aware Linear Algebra library that uses machine learning to optimize the runtime of all BLAS Level 3 operations. We test our method on two HPC platforms with Intel and AMD processors, using MKL and BLIS as baseline BLAS implementations. We achieve speedups of 1.5 to 3.0 for all operations, compared to using the maximum number of threads.
  arXiv  Detail & Related papers  (2024-06-28T03:07:53Z)
• AI and Memory Wall [81.06494558184049]
  We show how memory bandwidth can become the dominant bottleneck for decoder models. We argue for a redesign in model architecture, training, and deployment strategies to overcome this memory limitation.
  arXiv  Detail & Related papers  (2024-03-21T04:31:59Z)
• L2MAC: Large Language Model Automatic Computer for Extensive Code Generation [52.81694565226513]
  Transformer-based large language models (LLMs) are constrained by the fixed context window of the underlying transformer architecture. This paper presents L2MAC, the first practical LLM-based general-purpose stored-program automatic computer (von Neumann architecture) framework, for long and consistent output generation.
  arXiv  Detail & Related papers  (2023-10-02T16:55:19Z)
• FusionAI: Decentralized Training and Deploying LLMs with Massive Consumer-Level GPUs [57.12856172329322]
  We envision a decentralized system unlocking the potential vast untapped consumer-level GPU. This system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity.
  arXiv  Detail & Related papers  (2023-09-03T13:27:56Z)
• An Experimental Evaluation of Machine Learning Training on a Real Processing-in-Memory System [9.429605859159023]
  Training machine learning (ML) algorithms is a computationally intensive process, which is frequently memory-bound. Memory-centric computing systems, with processing-in-memory capabilities, can alleviate this data movement bottleneck. We implement several representative classic ML algorithms on a real-world general-purpose PIM architecture.
  arXiv  Detail & Related papers  (2022-07-16T09:39:53Z)
• ASH: A Modern Framework for Parallel Spatial Hashing in 3D Perception [91.24236600199542]
  ASH is a modern and high-performance framework for parallel spatial hashing on GPU. ASH achieves higher performance, supports richer functionality, and requires fewer lines of code. ASH and its example applications are open sourced in Open3D.
  arXiv  Detail & Related papers  (2021-10-01T16:25:40Z)
• In-memory Implementation of On-chip Trainable and Scalable ANN for AI/ML Applications [0.0]
  This paper presents an in-memory computing architecture for ANN enabling artificial intelligence (AI) and machine learning (ML) applications. Our novel on-chip training and inference in-memory architecture reduces energy cost and enhances throughput by simultaneously accessing the multiple rows of array per precharge cycle. The proposed architecture was trained and tested on the IRIS dataset which exhibits $46times$ more energy efficient per MAC (multiply and accumulate) operation compared to earlier classifiers.
  arXiv  Detail & Related papers  (2020-05-19T15:36:39Z)

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.