Deep learning has been shown as a successful machine learning method for a variety of tasks, and its popularity results in numerous open-source deep learning software tools. Training a deep network is usually a very time-consuming process. To address the computational challenge in deep learning, many tools exploit hardware features such as multi-core CPUs and many-core GPUs to shorten the training time. However, different tools exhibit different features and running performance when training different types of deep networks on different hardware platforms, which makes it difficult for end users to select an appropriate pair of software and hardware. In this paper, we aim to make a comparative study of the state-of-the-art GPU-accelerated deep learning software tools, including Caffe, CNTK, MXNet, TensorFlow, and Torch. We first benchmark the running performance of these tools with three popular types of neural networks on two CPU platforms and three GPU platforms. We then benchmark some distributed versions on multiple GPUs. Our contribution is two-fold. First, for end users of deep learning tools, our benchmarking results can serve as a guide to selecting appropriate hardware platforms and software tools. Second, for software developers of deep learning tools, our in-depth analysis points out possible future directions to further optimize the running performance.

Deep learning frameworks have been widely deployed on GPU servers for deep learning applications in both academia and industry. In training deep neural networks (DNNs), there are many standard processes or algorithms, such as convolution and stochastic gradient descent (SGD), but the running performance of different frameworks might be different even running the same deep model on the same GPU hardware. In this study, we evaluate the running performance of four stateof-the-art distributed deep learning frameworks (i.e., Caffe-MPI, CNTK, MXNet, and TensorFlow) over single-GPU, multi-GPU, and multi-node environments. We first build performance models of standard processes in training DNNs with SGD, and then we benchmark the running performance of these frameworks with three popular convolutional neural networks (i.e., AlexNet, GoogleNet and ResNet-50), after that, we analyze what factors that result in the performance gap among these four frameworks. Through both analytical and experimental analysis, we identify bottlenecks and overheads which could be further optimized. The main contribution is that the proposed performance models and the analysis provide further optimization directions in both algorithmic design and system configuration.

No abstract

As new data and updates are constantly arriving, the results of data mining applications become stale and obsolete over time. Incremental processing is a promising approach to refreshing mining results. It utilizes previously saved states to avoid the expense of re-computation from scratch.In this paper, we propose i 2 MapReduce, a novel incremental processing extension to MapReduce, the most widely used framework for mining big data. Compared with the state-of-the-art work on Incoop, i 2 MapReduce (i) performs key-value pair level incremental processing rather than task level re-computation, (ii) supports not only one-step computation but also more sophisticated iterative computation, which is widely used in data mining applications, and (iii) incorporates a set of novel techniques to reduce I/O overhead for accessing preserved fine-grain computation states. We evaluate i 2 MapReduce using a one-step algorithm and three iterative algorithms with diverse computation characteristics. Experimental results on Amazon EC2 show significant performance improvements of i 2 MapReduce compared to both plain and iterative MapReduce performing re-computation.

Energy efficiency has become one of the top design criteria for current computing systems. The dynamic voltage and frequency scaling (DVFS) has been widely adopted by laptop computers, servers, and mobile devices to conserve energy, while the GPU DVFS is still at a certain early age. This paper aims at exploring the impact of GPU DVFS on the application performance and power consumption, and furthermore, on energy conservation. We survey the state-of-the-art GPU DVFS characterizations, and then summarize recent research works on GPU power and performance models. We also conduct real GPU DVFS experiments on NVIDIA Fermi and Maxwell GPUs. According to our experimental results, GPU DVFS has significant potential for energy saving. The effect of scaling core voltage/frequency and memory voltage/frequency depends on not only the GPU architectures, but also the characteristic of GPU applications.

Gradient sparsification is a promising technique to significantly reduce the communication overhead in decentralized synchronous stochastic gradient descent (S-SGD) algorithms. Yet, many existing gradient sparsification schemes (e.g., Top-k sparsification) have a communication complexity of O(kP), where k is the number of selected gradients by each worker and P is the number of workers. Recently, the gTop-k sparsification scheme has been proposed to reduce the communication complexity from O(kP) to O(k logP), which significantly boosts the system scalability. However, it remains unclear whether the gTop-k sparsification scheme can converge in theory. In this paper, we first provide theoretical proofs on the convergence of the gTop-k scheme for non-convex objective functions under certain analytic assumptions. We then derive the convergence rate of gTop-k S-SGD, which is at the same order as the vanilla mini-batch SGD. Finally, we conduct extensive experiments on different machine learning models and data sets to verify the soundness of the assumptions and theoretical results, and discuss the impact of the compression ratio on the convergence performance.

No abstract

No abstract