Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userTENSILE: A Tensor granularity dynamic GPU memory scheduling method towards multiple dynamic workloads system
74
0
0.0
(
0
)
Added by
Kaixin Zhang
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Recently, deep learning has been an area of intense researching. However, as a kind of computing intensive task, deep learning highly relies on the scale of GPU memory, which is usually prohibitive and scarce. Although there are some extensive works have been proposed for dynamic GPU memory management, they are hard to be applied to systems with multiple dynamic workloads, such as in-database machine learning system. In this paper, we demonstrated TENSILE, a method of managing GPU memory in tensor granularity to reduce the GPU memory peak, with taking the multiple dynamic workloads into consideration. As far as we know, TENSILE is the first method which is designed to manage multiple workloads GPU memory using. We implement TENSILE on a deep learning framework built by ourselves, and evaluated its performance. The experiment results show that TENSILE can save more GPU memory with less extra time overhead than prior works in both single and multiple dynamic workloads scenarios.
rate research
Read More
Efficient GPU resource scheduling is essential to maximize resource utilization and save training costs for the increasing amount of deep learning workloads in shared GPU clusters. Existing GPU schedulers largely rely on static policies to leverage the performance characteristics of deep learning jobs. However, they can hardly reach optimal efficiency due to the lack of elasticity. To address the problem, we propose ONES, an ONline Evolutionary Scheduler for elastic batch size orchestration. ONES automatically manages the elasticity of each job based on the training batch size, so as to maximize GPU utilization and improve scheduling efficiency. It determines the batch size for each job through an online evolutionary search that can continuously optimize the scheduling decisions. We evaluate the effectiveness of ONES with 64 GPUs on TACCs Longhorn supercomputers. The results show that ONES can outperform the prior deep learning schedulers with a significantly shorter average job completion time.
The recent proposal of learned index structures opens up a new perspective on how traditional range indexes can be optimized. However, the current learned indexes assume the data distribution is relatively static and the access pattern is uniform, while real-world scenarios consist of skew query distribution and evolving data. In this paper, we demonstrate that the missing consideration of access patterns and dynamic data distribution notably hinders the applicability of learned indexes. To this end, we propose solutions for learned indexes for dynamic workloads (called Doraemon). To improve the latency for skew queries, Doraemon augments the training data with access frequencies. To address the slow model re-training when data distribution shifts, Doraemon caches the previously-trained models and incrementally fine-tunes them for similar access patterns and data distribution. Our preliminary result shows that, Doraemon improves the query latency by 45.1% and reduces the model re-training time to 1/20.
ML workloads are becoming increasingly popular in the cloud. Good cloud training performance is contingent on efficient parameter exchange among VMs. We find that Collectives, the widely used distributed communication algorithms, cannot perform optimally out of the box due to the hierarchical topology of datacenter networks and multi-tenancy nature of the cloudenvironment.In this paper, we present Cloud Collectives , a prototype that accelerates collectives by reordering theranks of participating VMs such that the communication pattern dictated by the selected collectives operation best exploits the locality in the network.Collectives is non-intrusive, requires no code changes nor rebuild of an existing application, and runs without support from cloud providers. Our preliminary application of Cloud Collectives on allreduce operations in public clouds results in a speedup of up to 3.7x in multiple microbenchmarks and 1.3x in real-world workloads of distributed training of deep neural networks and gradient boosted decision trees using state-of-the-art frameworks.
We present RDMAbox, a set of low level RDMA optimizations that provide better performance than previous approaches. The optimizations are packaged in easy-to-use kernel and user space libraries for applications and systems in data center. We demonstrate the flexibility and effectiveness of RDMAbox by implementing a kernel remote paging system and a user space file system using RDMAbox. RDMAbox employs two optimization techniques. First, we suggest RDMA request merging and chaining to further reduce the total number of I/O operations to the RDMA NIC. The I/O merge queue at the same time functions as a traffic regulator to enforce admission control and avoid overloading the NIC. Second, we propose Adaptive Polling to achieve higher efficiency of polling Work Completion than existing busy polling while maintaining the low CPU overhead of event trigger. Our implementation of a remote paging system with RDMAbox outperforms existing representative solutions with up to 4? throughput improvement and up to 83% decrease in average tail latency in bigdata workloads, and up to 83% reduction in completion time in machine learning workloads. Our implementation of a user space file system based on RDMAbox achieves up to 5.9? higher throughput over existing representative solutions.
Many real-world systems, such as social networks, rely on mining efficiently large graphs, with hundreds of millions of vertices and edges. This volume of information requires partitioning the graph across multiple nodes in a distributed system. This has a deep effect on performance, as traversing edges cut between partitions incurs a significant performance penalty due to the cost of communication. Thus, several systems in the literature have attempted to improve computational performance by enhancing graph partitioning, but they do not support another characteristic of real-world graphs: graphs are inherently dynamic, their topology evolves continuously, and subsequently the optimum partitioning also changes over time. In this work, we present the first system that dynamically repartitions massive graphs to adapt to structural changes. The system optimises graph partitioning to prevent performance degradation without using data replication. The system adopts an iterative vertex migration algorithm that relies on local information only, making complex coordination unnecessary. We show how the improvement in graph partitioning reduces execution time by over 50%, while adapting the partitioning to a large number of changes to the graph in three real-world scenarios.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
