arXiv subjects

Cong Guo

Publications and source records attributed to Cong Guo.

At least 55 records · Page 3Linked to original sources

arXiv2024

Vortex: Efficient Sample-Free Dynamic Tensor Program Optimization via Hardware-aware Strategy Space Hierarchization

Dynamic-shape deep neural networks (DNNs) are rapidly evolving, attracting attention for their ability to handle variable input sizes in real-time applications. However, existing compilation optimization methods for such networks often rely heavily on predefined samples to guide the compilation process, which restricts their adaptability and efficiency. These sample-driven methods struggle to efficiently manage the diverse and unpredictable shapes encountered in real-world scenarios, often resulting in suboptimal performance. To tackle these issues, we introduce Vortex, a hardware-driven and sample-free compiler tailored for dynamic-shape tensor programs. Vortex capitalizes on detailed hardware information and hierarchizes the strategy space to facilitate high-performance code generation without relying on runtime shape samples. It features a unique bidirectional compilation workflow, combining top-down abstraction for aligning tensor program execution with hardware hierarchies and bottom-up kernel construction to narrow the search space, enabling Vortex to achieve remarkable efficiency. Comprehensive evaluations confirm that Vortex reduces compilation time by $176\times$ compared to the existing dynamic-shape compiler. Additionally, it substantially outperforms existing vendor-provided libraries and dynamic-shape compilers on both CPU and GPU platforms, delivering speedups of $2.53\times$ and $3.01\times$, respectively.

cs.DC↗

arXiv2024

vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving

Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. This surge in demand poses significant challenges in optimizing throughput and latency while keeping costs manageable. The Key-Value (KV) cache, a standard method for retaining previous computations, makes LLM inference highly bounded by memory. While batching strategies can enhance performance, they frequently lead to significant memory fragmentation. Even though cutting-edge systems like vLLM mitigate KV cache fragmentation using paged Attention mechanisms, they still suffer from inefficient memory and computational operations due to the tightly coupled page management and computation kernels. This study introduces the vTensor, an innovative tensor structure for LLM inference based on GPU virtual memory management (VMM). vTensor addresses existing limitations by decoupling computation from memory defragmentation and offering dynamic extensibility. Our framework employs a CPU-GPU heterogeneous approach, ensuring efficient, fragmentation-free memory management while accommodating various computation kernels across different LLM architectures. Experimental results indicate that vTensor achieves an average speedup of 1.86x across different models, with up to 2.42x in multi-turn chat scenarios. Additionally, vTensor provides average speedups of 2.12x and 3.15x in kernel evaluation, reaching up to 3.92x and 3.27x compared to SGLang Triton prefix-prefilling kernels and vLLM paged Attention kernel, respectively. Furthermore, it frees approximately 71.25% (57GB) of memory on the NVIDIA A100 GPU compared to vLLM, enabling more memory-intensive workloads.

cs.DC↗

arXiv2024

A novel measurement method for SiPM external crosstalk probability at low temperature

Silicon photomultipliers (SiPMs) are being considered as potential replacements for conventional photomultiplier tubes (PMTs). However, a significant disadvantage of SiPMs is crosstalk (CT), wherein photons propagate through other pixels, resulting in secondary avalanches. CT can be categorized into internal crosstalk and external crosstalk based on whether the secondary avalanche occurs within the same SiPM or a different one. Numerous methods exist for quantitatively estimating the percentage of internal crosstalk (iCT). However, external crosstalk (eCT) has not been extensively studied. This article presents a novel measurement method for the probability of emitting an external crosstalk photon during a single pixel avalanche, using a setup involving two identical SiPMs facing each other, and without the need for complex optical designs. The entire apparatus is enclosed within a stainless steel chamber, functioning as a light-tight enclosure, and maintained at liquid nitrogen temperature. The experimental setup incorporates two Sensl J-60035 SiPM chips along with two 0.5-inch Hamamatsu Photonics (HPK) VUV4 S13370-6050CN SiPM arrays. The findings show a linear relationship between the probability of emitting an external crosstalk photon and the SiPM overvoltage for both SiPM samples. Surprisingly, this novel measurement method also rovides measurements of the SiPM photon detection efficiency (PDE) for eCT photons at low temperature.

physics.ins-det↗

arXiv2024

Accelerating Sparse DNNs Based on Tiled GEMM

Network pruning can reduce the computation cost of deep neural network (DNN) models. However, sparse models often produce randomly-distributed weights to maintain accuracy, leading to irregular computations. Consequently, unstructured sparse models cannot achieve meaningful speedup on commodity hardware built for dense matrix computations. Accelerators are usually modified or designed with structured sparsity-optimized architectures for exploiting sparsity. For example, the Ampere architecture introduces a sparse tensor core, which adopts the 2:4 sparsity pattern. We propose a pruning method that builds upon the insight that matrix multiplication generally breaks the large matrix into multiple smaller tiles for parallel execution. We present the tile-wise sparsity pattern, which maintains a structured sparsity pattern at the tile level for efficient execution but allows for irregular pruning at the global scale to maintain high accuracy. In addition, the tile-wise sparsity is implemented at the global memory level, and the 2:4 sparsity executes at the register level inside the sparse tensor core. We can combine these two patterns into a tile-vector-wise (TVW) sparsity pattern to explore more fine-grained sparsity and further accelerate the sparse DNN models. We evaluate the TVW on the GPU, achieving averages of $1.85\times$, $2.75\times$, and $22.18\times$ speedups over the dense model, block sparsity, and unstructured sparsity.

cs.DC↗

arXiv2024

GMLake: Efficient and Transparent GPU Memory Defragmentation for Large-scale DNN Training with Virtual Memory Stitching

Large-scale deep neural networks (DNNs), such as large language models (LLMs), have revolutionized the artificial intelligence (AI) field and become increasingly popular. However, training or fine-tuning such models requires substantial computational power and resources, where the memory capacity of a single acceleration device like a GPU is one of the most important bottlenecks. Owing to the prohibitively large overhead (e.g., $10 \times$) of GPUs' native memory allocator, DNN frameworks like PyTorch and TensorFlow adopt a caching allocator that maintains a memory pool with a splitting mechanism for fast memory (de)allocation. Unfortunately, the caching allocator's efficiency degrades quickly for popular memory reduction techniques such as recomputation, offloading, distributed training, and low-rank adaptation. The primary reason is that those memory reduction techniques introduce frequent and irregular memory (de)allocation requests, leading to severe fragmentation problems for the splitting-based caching allocator. To mitigate this fragmentation problem, we propose a novel memory allocation framework based on low-level GPU virtual memory management called GPU memory lake (GMLake). GMLake employs a novel virtual memory stitching (VMS) mechanism, which can fuse or combine non-contiguous memory blocks with a virtual memory address mapping. GMLake can reduce an average of 9.2 GB (up to 25 GB) GPU memory usage and 15% (up to 33% ) fragmentation among eight LLM models on GPU A100 with 80 GB memory. GMLake is completely transparent to the DNN models and memory reduction techniques and ensures the seamless execution of resource-intensive deep-learning tasks. We have open-sourced GMLake at https://github.com/intelligent-machine-learning/glake/tree/main/GMLake.

cs.DC↗

arXiv2023

A novel feature selection framework for incomplete data

Feature selection on incomplete datasets is an exceptionally challenging task. Existing methods address this challenge by first employing imputation methods to complete the incomplete data and then conducting feature selection based on the imputed data. Since imputation and feature selection are entirely independent steps, the importance of features cannot be considered during imputation. However, in real-world scenarios or datasets, different features have varying degrees of importance. To address this, we propose a novel incomplete data feature selection framework that considers feature importance. The framework mainly consists of two alternating iterative stages: the M-stage and the W-stage. In the M-stage, missing values are imputed based on a given feature importance vector and multiple initial imputation results. In the W-stage, an improved reliefF algorithm is employed to learn the feature importance vector based on the imputed data. Specifically, the feature importance vector obtained in the current iteration of the W-stage serves as input for the next iteration of the M-stage. Experimental results on both artificially generated and real incomplete datasets demonstrate that the proposed method outperforms other approaches significantly.

cs.LG↗

arXiv2023

JUNO: Optimizing High-Dimensional Approximate Nearest Neighbour Search with Sparsity-Aware Algorithm and Ray-Tracing Core Mapping

Approximate nearest neighbor (ANN) search is a widely applied technique in modern intelligent applications, such as recommendation systems and vector databases. Therefore, efficient and high-throughput execution of ANN search has become increasingly important. In this paper, we first characterize the state-of-the-art product quantization-based method of ANN search and identify a significant source of inefficiency in the form of unnecessary pairwise distance calculations and accumulations. To improve efficiency, we propose JUNO, an end-to-end ANN search system that adopts a carefully designed sparsity- and locality-aware search algorithm. We also present an efficient hardware mapping that utilizes ray tracing cores in modern GPUs with pipelined execution on tensor cores to execute our sparsity-aware ANN search algorithm. Our evaluations on four datasets ranging in size from 1 to 100 million search points demonstrate 2.2x-8.5x improvements in search throughput. Moreover, our algorithmic enhancements alone achieve a maximal 2.6x improvement on the hardware without the acceleration of the RT core.

cs.DC↗

arXiv2023

Iterative missing value imputation based on feature importance

Many datasets suffer from missing values due to various reasons,which not only increases the processing difficulty of related tasks but also reduces the accuracy of classification. To address this problem, the mainstream approach is to use missing value imputation to complete the dataset. Existing imputation methods estimate the missing parts based on the observed values in the original feature space, and they treat all features as equally important during data completion, while in fact different features have different importance. Therefore, we have designed an imputation method that considers feature importance. This algorithm iteratively performs matrix completion and feature importance learning, and specifically, matrix completion is based on a filling loss that incorporates feature importance. Our experimental analysis involves three types of datasets: synthetic datasets with different noisy features and missing values, real-world datasets with artificially generated missing values, and real-world datasets originally containing missing values. The results on these datasets consistently show that the proposed method outperforms the existing five imputation algorithms.To the best of our knowledge, this is the first work that considers feature importance in the imputation model.

cs.LG↗

arXiv2023

Study on the radon adsorption capability of low-background activated carbon

Radon is a significant background source in rare event detection experiments. Activated Carbon (AC) adsorption is widely used for effective radon removal. The selection of AC considers its adsorption capacity and radioactive background. In this study, using self-developed devices, we screened and identified a new kind of low-background AC from Qingdao Inaf Technology Company that has very high Radon adsorption capacity. By adjusting the average pore size to 2.3 nm, this AC demonstrates a radon adsorption capacity of 2.6 or 4.7 times higher than Saratech or Carboact activated carbon under the same conditions.

hep-ex↗

arXiv2023

Environmental radon control in the 700-m underground laboratory at JUNO

The Jiangmen Underground Neutrino Observatory is building the world's largest liquid scintillator detector with a 20 kt target mass and about 700 m overburden. The total underground space of civil construction is about 300,000 m$^3$ with the main hall volume of about 120,000 m$^3$, which is the biggest laboratory in the world. Radon concentration in the underground air is quite important for not only human beings' health but also the background of experiments with rare decay detection, such as neutrino and dark matter experiments. The radon concentration is the main hall is required to be around 100 Bq/m$^3$. Optimization of the ventilation with fresh air is effective to control the radon underground. To find the radon sources in the underground laboratory, we made a benchmark experiment in the refuge room near the main hall. The result shows that the radon emanating from underground water is one of the main radon sources in the underground air. The total underground ventilation rate is about 160,000 m$^3$/h fresh air with about 30 Bq/m$^3$ $^{222}$Rn from the bottom of the vertical tunnel after optimization, and 55,000 m$^3$/h is used for the ventilation in the main hall. Finally, the radon concentration inside the main hall decreased from 1600 Bq/m$^3$ to around 100 Bq/m$^3$. The suggested strategies for controlling radon concentration in the underground air are described in this paper.

physics.ins-det↗

arXiv2023

Efficient Adaptive Activation Rounding for Post-Training Quantization

Post-training quantization attracts increasing attention due to its convenience in deploying quantized neural networks. Although rounding-to-nearest remains the prevailing method for DNN quantization, prior research has demonstrated its suboptimal nature when applied to weight quantization. They propose optimizing weight rounding schemes by leveraging output error rather than the traditional weight quantization error. Our study reveals that similar rounding challenges also extend to activation quantization. Despite the easy generalization, the challenges lie in the dynamic nature of activation. Adaptive rounding is expected for varying activations and the method is subjected to runtime overhead. To tackle this, we propose the AQuant quantization framework with a novel perspective to reduce output error by adjusting rounding schemes of activations. Instead of using the constant rounding border 0.5 of the rounding-to-nearest operation, we make the border become a function w.r.t. the activation value to change the activation rounding by the adaptive border. To deal with the runtime overhead, we use a coarse-grained version of the border function. Finally, we introduce our framework to optimize the border function. Extensive experiments show that AQuant achieves notable improvements compared to state-of-the-art works and pushes the accuracy of ResNet-18 up to 60.31% under the 2-bit weight and activation quantization.

cs.LG↗

arXiv2023

Accelerating Generic Graph Neural Networks via Architecture, Compiler, Partition Method Co-Design

Graph neural networks (GNNs) have shown significant accuracy improvements in a variety of graph learning domains, sparking considerable research interest. To translate these accuracy improvements into practical applications, it is essential to develop high-performance and efficient hardware acceleration for GNN models. However, designing GNN accelerators faces two fundamental challenges: the high bandwidth requirement of GNN models and the diversity of GNN models. Previous works have addressed the first challenge by using more expensive memory interfaces to achieve higher bandwidth. For the second challenge, existing works either support specific GNN models or have generic designs with poor hardware utilization. In this work, we tackle both challenges simultaneously. First, we identify a new type of partition-level operator fusion, which we utilize to internally reduce the high bandwidth requirement of GNNs. Next, we introduce partition-level multi-threading to schedule the concurrent processing of graph partitions, utilizing different hardware resources. To further reduce the extra on-chip memory required by multi-threading, we propose fine-grained graph partitioning to generate denser graph partitions. Importantly, these three methods make no assumptions about the targeted GNN models, addressing the challenge of model variety. We implement these methods in a framework called SwitchBlade, consisting of a compiler, a graph partitioner, and a hardware accelerator. Our evaluation demonstrates that SwitchBlade achieves an average speedup of $1.85\times$ and energy savings of $19.03\times$ compared to the NVIDIA V100 GPU. Additionally, SwitchBlade delivers performance comparable to state-of-the-art specialized accelerators.

cs.AR↗

arXiv2023

A Class of Smoothing Modulus-Based Iterative Method for Solving Implicit Complementarity Problems

In this paper, a class of smoothing modulus-based iterative method was presented for solving implicit complementarity problems. The main idea was to transform the implicit complementarity problem into an equivalent implicit fixed-point equation, then introduces a smoothing function to obtain its approximation solutions. The convergence analysis of the algorithm was given, and the efficiency of the algorithms was verified by numerical experiment

math.NA↗

arXiv2023

AdaptGear: Accelerating GNN Training via Adaptive Subgraph-Level Kernels on GPUs

Graph neural networks (GNNs) are powerful tools for exploring and learning from graph structures and features. As such, achieving high-performance execution for GNNs becomes crucially important. Prior works have proposed to explore the sparsity (i.e., low density) in the input graph to accelerate GNNs, which uses the full-graph-level or block-level sparsity format. We show that they fail to balance the sparsity benefit and kernel execution efficiency. In this paper, we propose a novel system, referred to as AdaptGear, that addresses the challenge of optimizing GNNs performance by leveraging kernels tailored to the density characteristics at the subgraph level. Meanwhile, we also propose a method that dynamically chooses the optimal set of kernels for a given input graph. Our evaluation shows that AdaptGear can achieve a significant performance improvement, up to $6.49 \times$ ($1.87 \times$ on average), over the state-of-the-art works on two mainstream NVIDIA GPUs across various datasets.

cs.DC↗

arXiv2023

Pure Quantum Gradient Descent Algorithm and Full Quantum Variational Eigensolver

Optimization problems are prevalent in various fields, and the gradient-based gradient descent algorithm is a widely adopted optimization method. However, in classical computing, computing the numerical gradient for a function with $d$ variables necessitates at least $d+1$ function evaluations, resulting in a computational complexity of $O(d)$. As the number of variables increases, the classical gradient estimation methods require substantial resources, ultimately surpassing the capabilities of classical computers. Fortunately, leveraging the principles of superposition and entanglement in quantum mechanics, quantum computers can achieve genuine parallel computing, leading to exponential acceleration over classical algorithms in some cases.In this paper, we propose a novel quantum-based gradient calculation method that requires only a single oracle calculation to obtain the numerical gradient result for a multivariate function. The complexity of this algorithm is just $O(1)$. Building upon this approach, we successfully implemented the quantum gradient descent algorithm and applied it to the Variational Quantum Eigensolver (VQE), creating a pure quantum variational optimization algorithm. Compared with classical gradient-based optimization algorithm, this quantum optimization algorithm has remarkable complexity advantages, providing an efficient solution to optimization problems.The proposed quantum-based method shows promise in enhancing the performance of optimization algorithms, highlighting the potential of quantum computing in this field.

quant-ph↗

arXiv2023

OliVe: Accelerating Large Language Models via Hardware-friendly Outlier-Victim Pair Quantization

Transformer-based large language models (LLMs) have achieved great success with the growing model size. LLMs' size grows by $240\times$ every two years, which outpaces the hardware progress and makes model inference increasingly costly. Model quantization is a promising approach to mitigate the widening gap between LLM size and hardware capacity. However, the existence of outliers, values with significant magnitudes, in LLMs makes existing quantization methods less effective. Prior outlier-aware quantization schemes adopt sparsity encoding techniques to separate outliers from normal values where the process requires global coordination (e.g., a global sparsity coordination list). This incurs complex encoding/decoding hardware logics and an extra orchestration controller for the computation between outlier and normal values. As such, it is not hardware-efficient and hence only achieves sub-optimal quantization benefits. We propose OliVe, an algorithm/architecture co-designed solution that adopts an outlier-victim pair (OVP) quantization and handles outlier values locally with low hardware overheads and high performance gains. The key insight of OliVe is that outliers are important while the normal values next to them are not. Thus those normal values (called victims) can be sacrificed to accommodate outliers. This enables a memory-aligned OVP encoding scheme, which can be efficiently integrated to the existing hardware accelerators like systolic array and tensor core. As a result, OliVe-based accelerator surpasses the existing outlier-aware accelerator, GOBO, by 4.5$\times$ speedup and 4.0$\times$ energy reduction, respectively, with a superior model accuracy.

cs.AR↗

arXiv2022

Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training

An activation function is an element-wise mathematical function and plays a crucial role in deep neural networks (DNN). Many novel and sophisticated activation functions have been proposed to improve the DNN accuracy but also consume massive memory in the training process with back-propagation. In this study, we propose the nested forward automatic differentiation (Forward-AD), specifically for the element-wise activation function for memory-efficient DNN training. We deploy nested Forward-AD in two widely-used deep learning frameworks, TensorFlow and PyTorch, which support the static and dynamic computation graph, respectively. Our evaluation shows that nested Forward-AD reduces the memory footprint by up to 1.97x than the baseline model and outperforms the recomputation by 20% under the same memory reduction ratio.

cs.LG↗

arXiv2022

ANT: Exploiting Adaptive Numerical Data Type for Low-bit Deep Neural Network Quantization

Quantization is a technique to reduce the computation and memory cost of DNN models, which are getting increasingly large. Existing quantization solutions use fixed-point integer or floating-point types, which have limited benefits, as both require more bits to maintain the accuracy of original models. On the other hand, variable-length quantization uses low-bit quantization for normal values and high-precision for a fraction of outlier values. Even though this line of work brings algorithmic benefits, it also introduces significant hardware overheads due to variable-length encoding and decoding. In this work, we propose a fixed-length adaptive numerical data type called ANT to achieve low-bit quantization with tiny hardware overheads. Our data type ANT leverages two key innovations to exploit the intra-tensor and inter-tensor adaptive opportunities in DNN models. First, we propose a particular data type, flint, that combines the advantages of float and int for adapting to the importance of different values within a tensor. Second, we propose an adaptive framework that selects the best type for each tensor according to its distribution characteristics. We design a unified processing element architecture for ANT and show its ease of integration with existing DNN accelerators. Our design results in 2.8$\times$ speedup and 2.5$\times$ energy efficiency improvement over the state-of-the-art quantization accelerators.

cs.LG↗