arXiv subjects

Guohao Dai

Publications and source records attributed to Guohao Dai.

At least 73 records · Page 4Linked to original sources

arXiv2024

A Survey on Efficient Inference for Large Language Models

Large Language Models (LLMs) have attracted extensive attention due to their remarkable performance across various tasks. However, the substantial computational and memory requirements of LLM inference pose challenges for deployment in resource-constrained scenarios. Efforts within the field have been directed towards developing techniques aimed at enhancing the efficiency of LLM inference. This paper presents a comprehensive survey of the existing literature on efficient LLM inference. We start by analyzing the primary causes of the inefficient LLM inference, i.e., the large model size, the quadratic-complexity attention operation, and the auto-regressive decoding approach. Then, we introduce a comprehensive taxonomy that organizes the current literature into data-level, model-level, and system-level optimization. Moreover, the paper includes comparative experiments on representative methods within critical sub-fields to provide quantitative insights. Last but not least, we provide some knowledge summary and discuss future research directions.

cs.CL↗

arXiv2024

DiM: Diffusion Mamba for Efficient High-Resolution Image Synthesis

Diffusion models have achieved great success in image generation, with the backbone evolving from U-Net to Vision Transformers. However, the computational cost of Transformers is quadratic to the number of tokens, leading to significant challenges when dealing with high-resolution images. In this work, we propose Diffusion Mamba (DiM), which combines the efficiency of Mamba, a sequence model based on State Space Models (SSM), with the expressive power of diffusion models for efficient high-resolution image synthesis. To address the challenge that Mamba cannot generalize to 2D signals, we make several architecture designs including multi-directional scans, learnable padding tokens at the end of each row and column, and lightweight local feature enhancement. Our DiM architecture achieves inference-time efficiency for high-resolution images. In addition, to further improve training efficiency for high-resolution image generation with DiM, we investigate "weak-to-strong" training strategy that pretrains DiM on low-resolution images ($256\times 256$) and then finetune it on high-resolution images ($512 \times 512$). We further explore training-free upsampling strategies to enable the model to generate higher-resolution images (e.g., $1024\times 1024$ and $1536\times 1536$) without further fine-tuning. Experiments demonstrate the effectiveness and efficiency of our DiM. The code of our work is available here: {\url{https://github.com/tyshiwo1/DiM-DiffusionMamba/}}.

cs.CV↗

arXiv2024

Efficient Expert Pruning for Sparse Mixture-of-Experts Language Models: Enhancing Performance and Reducing Inference Costs

The rapid advancement of large language models (LLMs) has led to architectures with billions to trillions of parameters, posing significant deployment challenges due to their substantial demands on memory, processing power, and energy consumption. Sparse Mixture-of-Experts (SMoE) architectures have emerged as a solution, activating only a subset of parameters per token, thereby achieving faster inference while maintaining performance. However, SMoE models still face limitations in broader deployment due to their large parameter counts and significant GPU memory requirements. In this work, we introduce a gradient-free evolutionary strategy named EEP (Efficient Expert P}runing) to enhance the pruning of experts in SMoE models. EEP relies solely on model inference (i.e., no gradient computation) and achieves greater sparsity while maintaining or even improving performance on downstream tasks. EEP can be used to reduce both the total number of experts (thus saving GPU memory) and the number of active experts (thus accelerating inference). For example, we demonstrate that pruning up to 75% of experts in Mixtral $8\times7$B-Instruct results in a substantial reduction in parameters with minimal performance loss. Remarkably, we observe improved performance on certain tasks, such as a significant increase in accuracy on the SQuAD dataset (from 53.4% to 75.4%), when pruning half of the experts. With these results, EEP not only lowers the barrier to deploying SMoE models,but also challenges the conventional understanding of model pruning by showing that fewer experts can lead to better task-specific performance without any fine-tuning. Code is available at https://github.com/imagination-research/EEP.

cs.LG↗

arXiv2024

Evaluating Quantized Large Language Models

Post-training quantization (PTQ) has emerged as a promising technique to reduce the cost of large language models (LLMs). Specifically, PTQ can effectively mitigate memory consumption and reduce computational overhead in LLMs. To meet the requirements of both high efficiency and performance across diverse scenarios, a comprehensive evaluation of quantized LLMs is essential to guide the selection of quantization methods. This paper presents a thorough evaluation of these factors by evaluating the effect of PTQ on Weight, Activation, and KV Cache on 11 model families, including OPT, LLaMA2, Falcon, Bloomz, Mistral, ChatGLM, Vicuna, LongChat, StableLM, Gemma, and Mamba, with parameters ranging from 125M to 180B. The evaluation encompasses five types of tasks: basic NLP, emergent ability, trustworthiness, dialogue, and long-context tasks. Moreover, we also evaluate the state-of-the-art (SOTA) quantization methods to demonstrate their applicability. Based on the extensive experiments, we systematically summarize the effect of quantization, provide recommendations to apply quantization techniques, and point out future directions. The code can be found in https://github.com/thu-nics/qllm-eval.

cs.CL↗

arXiv2024

MixDQ: Memory-Efficient Few-Step Text-to-Image Diffusion Models with Metric-Decoupled Mixed Precision Quantization

Diffusion models have achieved significant visual generation quality. However, their significant computational and memory costs pose challenge for their application on resource-constrained mobile devices or even desktop GPUs. Recent few-step diffusion models reduces the inference time by reducing the denoising steps. However, their memory consumptions are still excessive. The Post Training Quantization (PTQ) replaces high bit-width FP representation with low-bit integer values (INT4/8) , which is an effective and efficient technique to reduce the memory cost. However, when applying to few-step diffusion models, existing quantization methods face challenges in preserving both the image quality and text alignment. To address this issue, we propose an mixed-precision quantization framework - MixDQ. Firstly, We design specialized BOS-aware quantization method for highly sensitive text embedding quantization. Then, we conduct metric-decoupled sensitivity analysis to measure the sensitivity of each layer. Finally, we develop an integer-programming-based method to conduct bit-width allocation. While existing quantization methods fall short at W8A8, MixDQ could achieve W8A8 without performance loss, and W4A8 with negligible visual degradation. Compared with FP16, we achieve 3-4x reduction in model size and memory cost, and 1.45x latency speedup.

cs.CV↗

arXiv2024

FlashEval: Towards Fast and Accurate Evaluation of Text-to-image Diffusion Generative Models

In recent years, there has been significant progress in the development of text-to-image generative models. Evaluating the quality of the generative models is one essential step in the development process. Unfortunately, the evaluation process could consume a significant amount of computational resources, making the required periodic evaluation of model performance (e.g., monitoring training progress) impractical. Therefore, we seek to improve the evaluation efficiency by selecting the representative subset of the text-image dataset. We systematically investigate the design choices, including the selection criteria (textural features or image-based metrics) and the selection granularity (prompt-level or set-level). We find that the insights from prior work on subset selection for training data do not generalize to this problem, and we propose FlashEval, an iterative search algorithm tailored to evaluation data selection. We demonstrate the effectiveness of FlashEval on ranking diffusion models with various configurations, including architectures, quantization levels, and sampler schedules on COCO and DiffusionDB datasets. Our searched 50-item subset could achieve comparable evaluation quality to the randomly sampled 500-item subset for COCO annotations on unseen models, achieving a 10x evaluation speedup. We release the condensed subset of these commonly used datasets to help facilitate diffusion algorithm design and evaluation, and open-source FlashEval as a tool for condensing future datasets, accessible at https://github.com/thu-nics/FlashEval.

cs.CV↗

arXiv2024

FlightLLM: Efficient Large Language Model Inference with a Complete Mapping Flow on FPGAs

Transformer-based Large Language Models (LLMs) have made a significant impact on various domains. However, LLMs' efficiency suffers from both heavy computation and memory overheads. Compression techniques like sparsification and quantization are commonly used to mitigate the gap between LLM's computation/memory overheads and hardware capacity. However, existing GPU and transformer-based accelerators cannot efficiently process compressed LLMs, due to the following unresolved challenges: low computational efficiency, underutilized memory bandwidth, and large compilation overheads. This paper proposes FlightLLM, enabling efficient LLMs inference with a complete mapping flow on FPGAs. In FlightLLM, we highlight an innovative solution that the computation and memory overhead of LLMs can be solved by utilizing FPGA-specific resources (e.g., DSP48 and heterogeneous memory hierarchy). We propose a configurable sparse DSP chain to support different sparsity patterns with high computation efficiency. Second, we propose an always-on-chip decode scheme to boost memory bandwidth with mixed-precision support. Finally, to make FlightLLM available for real-world LLMs, we propose a length adaptive compilation method to reduce the compilation overhead. Implemented on the Xilinx Alveo U280 FPGA, FlightLLM achieves 6.0$\times$ higher energy efficiency and 1.8$\times$ better cost efficiency against commercial GPUs (e.g., NVIDIA V100S) on modern LLMs (e.g., LLaMA2-7B) using vLLM and SmoothQuant under the batch size of one. FlightLLM beats NVIDIA A100 GPU with 1.2$\times$ higher throughput using the latest Versal VHK158 FPGA.

cs.AR↗

arXiv2024

FlashDecoding++: Faster Large Language Model Inference on GPUs

As the Large Language Model (LLM) becomes increasingly important in various domains. However, the following challenges still remain unsolved in accelerating LLM inference: (1) Synchronized partial softmax update. The softmax operation requires a synchronized update operation among each partial softmax result, leading to ~20% overheads for the attention computation in LLMs. (2) Under-utilized computation of flat GEMM. The shape of matrices performing GEMM in LLM inference is flat, leading to under-utilized computation and >50% performance loss after padding zeros in previous designs. (3) Performance loss due to static dataflow. Kernel performance in LLM depends on varied input data features, hardware configurations, etc. A single and static dataflow may lead to a 50.25% performance loss for GEMMs of different shapes in LLM inference. We present FlashDecoding++, a fast LLM inference engine supporting mainstream LLMs and hardware back-ends. To tackle the above challenges, FlashDecoding++ creatively proposes: (1) Asynchronized softmax with unified max value. FlashDecoding++ introduces a unified max value technique for different partial softmax computations to avoid synchronization. (2) Flat GEMM optimization with double buffering. FlashDecoding++ points out that flat GEMMs with different shapes face varied bottlenecks. Then, techniques like double buffering are introduced. (3) Heuristic dataflow with hardware resource adaptation. FlashDecoding++ heuristically optimizes dataflow using different hardware resource considering input dynamics. Due to the versatility of optimizations in FlashDecoding++, FlashDecoding++ can achieve up to 4.86x and 2.18x speedup on both NVIDIA and AMD GPUs compared to Hugging Face implementations. FlashDecoding++ also achieves an average speedup of 1.37x compared to state-of-the-art LLM inference engines on mainstream LLMs.

cs.LG↗

arXiv2023

TorchSparse++: Efficient Training and Inference Framework for Sparse Convolution on GPUs

Sparse convolution plays a pivotal role in emerging workloads, including point cloud processing in AR/VR, autonomous driving, and graph understanding in recommendation systems. Since the computation pattern is sparse and irregular, specialized high-performance kernels are required. Existing GPU libraries offer two dataflow types for sparse convolution. The gather-GEMM-scatter dataflow is easy to implement but not optimal in performance, while the dataflows with overlapped computation and memory access (e.g.implicit GEMM) are highly performant but have very high engineering costs. In this paper, we introduce TorchSparse++, a new GPU library that achieves the best of both worlds. We create a highly efficient Sparse Kernel Generator that generates performant sparse convolution kernels at less than one-tenth of the engineering cost of the current state-of-the-art system. On top of this, we design the Sparse Autotuner, which extends the design space of existing sparse convolution libraries and searches for the best dataflow configurations for training and inference workloads. Consequently, TorchSparse++ achieves 2.9x, 3.3x, 2.2x and 1.7x measured end-to-end speedup on an NVIDIA A100 GPU over state-of-the-art MinkowskiEngine, SpConv 1.2, TorchSparse and SpConv v2 in inference; and is 1.2-1.3x faster than SpConv v2 in mixed precision training across seven representative autonomous driving benchmarks. It also seamlessly supports graph convolutions, achieving 2.6-7.6x faster inference speed compared with state-of-the-art graph deep learning libraries.

cs.DC↗

arXiv2023

Ada3D : Exploiting the Spatial Redundancy with Adaptive Inference for Efficient 3D Object Detection

Voxel-based methods have achieved state-of-the-art performance for 3D object detection in autonomous driving. However, their significant computational and memory costs pose a challenge for their application to resource-constrained vehicles. One reason for this high resource consumption is the presence of a large number of redundant background points in Lidar point clouds, resulting in spatial redundancy in both 3D voxel and dense BEV map representations. To address this issue, we propose an adaptive inference framework called Ada3D, which focuses on exploiting the input-level spatial redundancy. Ada3D adaptively filters the redundant input, guided by a lightweight importance predictor and the unique properties of the Lidar point cloud. Additionally, we utilize the BEV features' intrinsic sparsity by introducing the Sparsity Preserving Batch Normalization. With Ada3D, we achieve 40% reduction for 3D voxels and decrease the density of 2D BEV feature maps from 100% to 20% without sacrificing accuracy. Ada3D reduces the model computational and memory cost by 5x, and achieves 1.52x/1.45x end-to-end GPU latency and 1.5x/4.5x GPU peak memory optimization for the 3D and 2D backbone respectively.

cs.CV↗

arXiv2023

Seeking the Yield Barrier: High-Dimensional SRAM Evaluation Through Optimal Manifold

Being able to efficiently obtain an accurate estimate of the failure probability of SRAM components has become a central issue as model circuits shrink their scale to submicrometer with advanced technology nodes. In this work, we revisit the classic norm minimization method. We then generalize it with infinite components and derive the novel optimal manifold concept, which bridges the surrogate-based and importance sampling (IS) yield estimation methods. We then derive a sub-optimal manifold, optimal hypersphere, which leads to an efficient sampling method being aware of the failure boundary called onion sampling. Finally, we use a neural coupling flow (which learns from samples like a surrogate model) as the IS proposal distribution. These combinations give rise to a novel yield estimation method, named Optimal Manifold Important Sampling (OPTIMIS), which keeps the advantages of the surrogate and IS methods to deliver state-of-the-art performance with robustness and consistency, with up to 3.5x in efficiency and 3x in accuracy over the best of SOTA methods in High-dimensional SRAM evaluation.

cs.LG↗

arXiv2023

Adam Accumulation to Reduce Memory Footprints of both Activations and Gradients for Large-scale DNN Training

Running out of GPU memory has become a main bottleneck for large-scale DNN training. How to reduce the memory footprint during training has received intensive research attention. We find that previous gradient accumulation reduces activation memory but fails to be compatible with gradient memory reduction due to a contradiction between preserving gradients and releasing gradients. To address this issue, we propose a novel optimizer accumulation method for Adam, named Adam Accumulation (AdamA), which enables reducing both activation and gradient memory. Specifically, AdamA directly integrates gradients into optimizer states and accumulates optimizer states over micro-batches, so that gradients can be released immediately after use. We mathematically and experimentally demonstrate AdamA yields the same convergence properties as Adam. Evaluated on transformer-based models, AdamA achieves up to 23% memory reduction compared to gradient accumulation with less than 2% degradation in training throughput. Notably, AdamA can work together with memory reduction methods for optimizer states to fit 1.26x~3.14x larger models over PyTorch and DeepSpeed baseline on GPUs with different memory capacities.

cs.LG↗

arXiv2023

CogDL: A Comprehensive Library for Graph Deep Learning

Graph neural networks (GNNs) have attracted tremendous attention from the graph learning community in recent years. It has been widely adopted in various real-world applications from diverse domains, such as social networks and biological graphs. The research and applications of graph deep learning present new challenges, including the sparse nature of graph data, complicated training of GNNs, and non-standard evaluation of graph tasks. To tackle the issues, we present CogDL, a comprehensive library for graph deep learning that allows researchers and practitioners to conduct experiments, compare methods, and build applications with ease and efficiency. In CogDL, we propose a unified design for the training and evaluation of GNN models for various graph tasks, making it unique among existing graph learning libraries. By utilizing this unified trainer, CogDL can optimize the GNN training loop with several training techniques, such as mixed precision training. Moreover, we develop efficient sparse operators for CogDL, enabling it to become the most competitive graph library for efficiency. Another important CogDL feature is its focus on ease of use with the aim of facilitating open and reproducible research of graph learning. We leverage CogDL to report and maintain benchmark results on fundamental graph tasks, which can be reproduced and directly used by the community.

cs.SI↗

arXiv2023

Sgap: Towards Efficient Sparse Tensor Algebra Compilation for GPU

Sparse compiler is a promising solution for sparse tensor algebra optimization. In compiler implementation, reduction in sparse-dense hybrid algebra plays a key role in performance. Though GPU provides various reduction semantics that can better utilize the parallel computing and memory bandwidth capacity, the central question is: how to elevate the flexible reduction semantics to sparse compilation theory that assumes serial execution. Specifically, we have to tackle two main challenges: (1) there are wasted parallelism by adopting static synchronization granularity (2) static reduction strategy limits optimization space exploration. We propose Sgap: segment group and atomic parallelism to solve these problems. Atomic parallelism captures the flexible reduction semantics to systematically analyze the optimization space of sparse-dense hybrid algebra on GPU. It is a new optimization technique beyond current compiler-based and open-source runtime libraries. Segment group elevates the flexible reduction semantics to suitable levels of abstraction in the sparse compilation theory. It adopts changeable group size and user-defined reduction strategy to solve challenge (1) and (2), respectively. Finally, we use GPU sparse matrix-matrix multiplication (SpMM) on the TACO compiler as a use case to demonstrate the effectiveness of segment group in reduction semantics elevation. We achieve up to 1.2x speedup over the original TACO's SpMM kernels. We also apply new optimization techniques found by atomic parallelism to an open-source state-of-the-art SpMM library dgSPARSE. We achieve 1.6x - 2.3x speedup on the algorithm tuned with atomic parallelism.

cs.DC↗

arXiv2022

High-Dimensional Yield Estimation using Shrinkage Deep Features and Maximization of Integral Entropy Reduction

Despite the fast advances in high-sigma yield analysis with the help of machine learning techniques in the past decade, one of the main challenges, the curse of dimensionality, which is inevitable when dealing with modern large-scale circuits, remains unsolved. To resolve this challenge, we propose an absolute shrinkage deep kernel learning, ASDK, which automatically identifies the dominant process variation parameters in a nonlinear-correlated deep kernel and acts as a surrogate model to emulate the expensive SPICE simulation. To further improve the yield estimation efficiency, we propose a novel maximization of approximated entropy reduction for an efficient model update, which is also enhanced with parallel batch sampling for parallel computing, making it ready for practical deployment. Experiments on SRAM column circuits demonstrate the superiority of ASDK over the state-of-the-art (SOTA) approaches in terms of accuracy and efficiency with up to 10.3x speedup over SOTA methods.

cs.CE↗

arXiv2022

GRAPHIC: GatheR-And-Process in Highly parallel with In-SSD Compression Architecture in Very Large-Scale Graph

Graph convolutional network (GCN), an emerging algorithm for graph computing, has achieved promising performance in graphstructure tasks. To achieve acceleration for data-intensive and sparse graph computing, ASICs such as GCNAX have been proposed for efficient execution of aggregation and combination in GCN. GCNAX reducing 8x DRAM accesses compared with previous efforts. However, as graphs have reached terabytes in size, off-chip data movement from SSD to DRAM becomes a serious latency bottleneck. This paper proposes Compressive Graph Transmission (CGTrans), which performs the aggregation in SSD to dramatically relieves the transfer latency bottleneck due to SSD loading compared to CMOS-based graph accelerator ASICs. InSSD computing technique is required for CGTrans. Recently, Insider was proposed as a near-SSD processing system computing by integrating FPGA in SSD. However, the Insider still suffers low area efficiency, which will limit the performance of CGTrans. The recently proposed Fully Concurrent Access Technique (FAST) is utilized. FAST-GAS, as an in-SSD graph computing accelerator, is proposed to provide high-concurrent gather-andscatter operations to overcome the area efficiency problem. We proposed the GRAPHIC system containing CGTrans dataflow deployed on FAST-GAS. Experiments show CGTrans reduces SSD loading by a factor of 50x, while GRAPHIC achieves 3.6x, and 2.4x speedup on average over GCNAX and CGTrans on Insider, respectively.

cs.AR↗

arXiv2022

Heuristic Adaptability to Input Dynamics for SpMM on GPUs

Sparse Matrix-Matrix Multiplication (SpMM) has served as fundamental components in various domains. Many previous studies exploit GPUs for SpMM acceleration because GPUs provide high bandwidth and parallelism. We point out that a static design does not always improve the performance of SpMM on different input data (e.g., >85\% performance loss with a single algorithm). In this paper, we consider the challenge of input dynamics from a novel auto-tuning perspective, while following issues remain to be solved: (1) Orthogonal design principles considering sparsity. Orthogonal design principles for such a sparse problem should be extracted to form different algorithms, and further used for performance tuning. (2) Nontrivial implementations in the algorithm space. Combining orthogonal design principles to create new algorithms needs to tackle with new challenges like thread race handling. (3) Heuristic adaptability to input dynamics. The heuristic adaptability is required to dynamically optimize code for input dynamics. To tackle these challenges, we first propose a novel three-loop model to extract orthogonal design principles for SpMM on GPUs. The model not only covers previous SpMM designs, but also comes up with new designs absent from previous studies. We propose techniques like conditional reduction to implement algorithms missing in previous studies. We further propose DA-SpMM, a Data-Aware heuristic GPU kernel for SpMM. DA-SpMM adaptively optimizes code considering input dynamics. Extensive experimental results show that, DA-SpMM achieves 1.26x~1.37x speedup compared with the best NVIDIA cuSPARSE algorithm on average, and brings up to 5.59x end-to-end speedup to applications like Graph Neural Networks.

cs.AR↗

arXiv2021

Understanding GNN Computational Graph: A Coordinated Computation, IO, and Memory Perspective

Graph Neural Networks (GNNs) have been widely used in various domains, and GNNs with sophisticated computational graph lead to higher latency and larger memory consumption. Optimizing the GNN computational graph suffers from: (1) Redundant neural operator computation. The same data are propagated through the graph structure to perform the same neural operation multiple times in GNNs, leading to redundant computation which accounts for 92.4% of total operators. (2) Inconsistent thread mapping. Efficient thread mapping schemes for vertex-centric and edge-centric operators are different. This inconsistency prohibits operator fusion to reduce memory IO. (3) Excessive intermediate data. For GNN training which is usually performed concurrently with inference, intermediate data must be stored for the backward pass, consuming 91.9% of the total memory requirement. To tackle these challenges, we propose following designs to optimize the GNN computational graph from a novel coordinated computation, IO, and memory perspective: (1) Propagation-postponed operator reorganization. We reorganize operators to perform neural operations before the propagation, thus the redundant computation is eliminated. (2) Unified thread mapping for fusion. We propose a unified thread mapping scheme for both vertex- and edge-centric operators to enable fusion and reduce IO. (3) Intermediate data recomputation. Intermediate data are recomputed during the backward pass to reduce the total memory consumption. Extensive experimental results on three typical GNN models show that, we achieve up to 2.75x end-to-end speedup, 6.89x less memory IO, and 7.73x less memory consumption over state-of-the-art frameworks.

cs.LG↗