Pipeline parallelism is a key technique for training large language models within GPU clusters. However, it often leads to a memory imbalance problem, where certain GPUs face high memory pressure while others underutilize their capacity. This imbalance results in suboptimal training performance, even when the overall GPU memory capacity is sufficient for more efficient setups. To address this inefficiency, we propose BPipe, a novel approach for achieving memory balance in pipeline parallelism. BPipe employs an activation balancing method to transfer intermediate activations between GPUs during training, enabling all GPUs to utilize comparable amounts of memory. With balanced memory utilization, BPipe enhances the training efficiency of large language models like GPT-3 by eliminating redundant recomputations or increasing the micro-batch size. Our evaluation conducted on 48 A100 GPUs across six nodes interconnected with HDR InfiniBand shows that BPipe accelerates the training of GPT-3 96B and GPT-3 134B models by 1.25x-2.17x compared to Megatron-LM, a state-of-the-art framework for training large language models.

Imperative programming allows users to implement their deep neural networks (DNNs) easily and has become an essential part of recent deep learning (DL) frameworks. Recently, several systems have been proposed to combine the usability of imperative programming with the optimized performance of symbolic graph execution. Such systems convert imperative Python DL programs to optimized symbolic graphs and execute them. However, they cannot fully support the usability of imperative programming. For example, if an imperative DL program contains a Python feature with no corresponding symbolic representation (e.g., third-party library calls or unsupported dynamic control flows) they fail to execute the program. To overcome this limitation, we propose Terra, an imperative-symbolic co-execution system that can handle any imperative DL programs while achieving the optimized performance of symbolic graph execution. To achieve this, Terra builds a symbolic graph by decoupling DL operations from Python features. Then, Terra conducts the imperative execution to support all Python features, while delegating the decoupled operations to the symbolic execution. We evaluated Terra’s performance improvement and coverage with ten imperative DL programs for several DNN architectures. The results show that Terra can speed up the execution of all ten imperative DL programs, whereas AutoGraph, one of the state-of-the-art systems, fails to execute five of them.

While deep neural networks (DNNs) have shown to be successful in several domains like computer vision, non-DNN models such as linear models and gradient boosting trees are still considered state-of-the-art over tabular data. When using these models, data scientists often author machine learning (ML) pipelines: DAG of ML operators comprising data transforms and ML models, whereby each operator is sequentially trained one-at-a-time. Conversely, when training DNNs, layers composing the neural networks are simultaneously trained using backpropagation. In this paper, we argue that the training scheme of ML pipelines is sub-optimal because it tries to optimize a single operator at a time thus losing the chance of global optimization. We therefore propose WindTunnel: a system that translates a trained ML pipeline into a pipeline of neural network modules and jointly optimizes the modules using backpropagation. We also suggest translation methodologies for several non-differentiable operators such as gradient boosting trees and categorical feature encoders. Our experiments show that fine-tuning of the translated WindTunnel pipelines is a promising technique able to increase the final accuracy.

Deep learning (DL) frameworks take advantage of GPUs to improve the speed of DL inference and training. Ideally, DL frameworks should be able to fully utilize the computation power of GPUs such that the running time depends on the amount of computation assigned to GPUs. Yet, we observe that in scheduling GPU tasks, existing DL frameworks suffer from inefficiencies such as large scheduling overhead and unnecessary serial execution. To this end, we propose Nimble, a DL execution engine that runs GPU tasks in parallel with minimal scheduling overhead. Nimble introduces a novel technique called ahead-of-time (AoT) scheduling. Here, the scheduling procedure finishes before executing the GPU kernel, thereby removing most of the scheduling overhead during run time. Furthermore, Nimble automatically parallelizes the execution of GPU tasks by exploiting multiple GPU streams in a single GPU. Evaluation on a variety of neural networks shows that compared to PyTorch, Nimble speeds up inference and training by up to 22.34× and 3.61×, respectively. Moreover, Nimble outperforms state-of-the-art inference systems, TensorRT and TVM, by up to 2.81× and 1.70×, respectively.

The employment of high-performance servers and GPU accelerators for training deep neural network models have greatly accelerated recent advances in deep learning (DL). DL frameworks, such as TensorFlow, MXNet, and Caffe2, have emerged to assist DL researchers to train their models in a distributed manner. Although current DL frameworks scale well for image classification models, there remain opportunities for scalable distributed training on natural language processing (NLP) models. We found that current frameworks show relatively low scalability on training NLP models due to the lack of consideration to the difference in sparsity of model parameters. In this paper, we propose Parallax, a framework that optimizes data parallel training by utilizing the sparsity of model parameters. Parallax introduces a hybrid approach that combines Parameter Server and AllReduce architectures to optimize the amount of data transfer according to the sparsity. Experiments show that Parallax built atop Tensor-Flow achieves scalable training throughput on both dense and sparse models while requiring little effort from its users. Parallax achieves up to 2.8x, 6.02x speedup for NLP models than TensorFlow and Horovod with 48 GPUs, respectively. The training speed for the image classification models is equal to Horovod and 1.53x faster than TensorFlow.

The performance of distributed machine learning systems is dependent on their system configuration. However, configuring the system for optimal performance is challenging and time consuming even for experts due to the diverse runtime factors such as workloads or the system environment. We present cost-based optimization to automatically find a good system configuration for parameter server (PS) machine learning (ML) frameworks. We design and implement Cruise that applies the optimization technique to tune distributed PS ML execution automatically. Evaluation results on three ML applications verify that Cruise automates the system configuration of the applications to achieve good performance with minor reconfiguration costs.

Recursive neural networks have widely been used by researchers to handle applications with recursively or hierarchically structured data. However, embedded control flow deep learning frameworks such as TensorFlow, Theano, Caffe2, and MXNet fail to efficiently represent and execute such neural networks, due to lack of support for recursion. In this paper, we add recursion to the programming model of existing frameworks by complementing their design with recursive execution of dataflow graphs as well as additional APIs for recursive definitions. Unlike iterative implementations, which can only understand the topological index of each node in recursive data structures, our recursive implementation is able to exploit the recursive relationships between nodes for efficient execution based on parallel computation. We present an implementation on TensorFlow and evaluation results with various recursive neural network models, showing that our recursive implementation not only conveys the recursive nature of recursive neural networks better than other implementations, but also uses given resources more effectively to reduce training and inference time.