Designing Power-Aware Collective Communication Algorithms for InfiniBand Clusters

2012 IEEE 24th International Symposium on Computer Architecture and High Performance Computing

Gaenko

2012

Although high-performance computing has always been about efficient application execution, both energy and power consumption have become critical concerns owing to their effect on operating costs and failure rates of large-scale computing platforms. Modern microprocessors are equipped with the capabilities to reduce their power consumption using techniques such as dynamic voltage and frequency scaling (DVFS) and CPU clock modulation (called throttling). Without careful application, however, DVFS and throttling may cause a significant performance loss due to system overhead. This work presents design considerations for a runtime procedure that dynamically analyzes blocking point-to-point communications, groups them according to the proposed criteria, and applies frequency scaling by analyzing both communication and architectural parameters without penalizing the performance much. Experiments, performed on NAS parallel benchmarks verify the proposed design by exhibiting energy savings of as much as 11% with a performance loss as low as 2%.

Section: Related Workmentioning

confidence: 99%

Runtime Procedure for Energy Savings in Applications with Point-to-Point Communications

2012 IEEE 24th International Symposium on Computer Architecture and High Performance Computing

Gaenko

2012

“…But as the number of cores within a node keep on increasing, this approach of forcing the sockets to remain idle during communication, can introduce significant performance overheads. The power saving achieved in [12] are be equivalent to executing the two sockets in a node at the minimum frequency and throttling state T 4 , whereas the approach proposed here achieves better power saving by keeping both sockets at minimum frequency and throttling to a higher state T 5 . The detailed experimental comparison of the two algorithms is left as future work.…”

Section: Related Workmentioning

confidence: 99%

“…However, they differ significantly with the approach presented in this paper. Specifically, [12] assumes that throttling has a negative effect on internode communication. Thus, the authors of [12] redesign the all-to-all operation, such that a set of sockets does not take part in communication a some point of time, and thus, may throttled.…”

Section: Related Workmentioning

confidence: 99%

“…Specifically, [12] assumes that throttling has a negative effect on internode communication. Thus, the authors of [12] redesign the all-to-all operation, such that a set of sockets does not take part in communication a some point of time, and thus, may throttled. But as the number of cores within a node keep on increasing, this approach of forcing the sockets to remain idle during communication, can introduce significant performance overheads.…”

Section: Related Workmentioning

confidence: 99%

“…In [9], authors have developed a tool which estimates power consumption characteristics of a parallel application in terms of various CPU components. In [12], algorithms to save energy in the collectives, such MPI Alltoall and MPI Broadcast, are proposed. However, they differ significantly with the approach presented in this paper.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Per-call Energy Saving Strategies in All-to-All Communications

Recent Advances in the Message Passing Interface

2011

With the increase in the peak performance of modern computing platforms, their energy consumption grows as well, which may lead to overwhelming operating costs and failure rates. Techniques, such as Dynamic Voltage and Frequency Scaling (called DVFS) and CPU Clock Modulation (called throttling) are often used to reduce the power consumption of the compute nodes. However, these techniques should be used judiciously during the application execution to avoid significant performance losses. In this work, two implementations of the all-to-all collective operations are studied as to their augmentation with energy saving strategies on the per-call basis. Experiments were performed on the OSU MPI benchmark as well as NAS and CPMD application benchmarks, in which power consumption was reduced by up to 10% and 15.7%, respectively, with little performance degradation. Abstract. With the increase in the peak performance of modern computing platforms, their energy consumption grows as well, which may lead to overwhelming operating costs and failure rates. Techniques, such as Dynamic Voltage and Frequency Scaling (called DVFS) and CPU Clock Modulation (called throttling) are often used to reduce the power consumption of the compute nodes. However, these techniques should be used judiciously during the application execution to avoid significant performance losses. In this work, two implementations of the all-to-all collective operations are studied as to their augmentation with energy saving strategies on the per-call basis. Experiments were performed on the OSU MPI benchmark as well as NAS and CPMD application benchmarks, in which power consumption was reduced by up to 10% and 15.7%, respectively, with little performance degradation.

Achieving energy efficiency during collective communications

Concurrency and Computation

Zhao

2012

SUMMARYEnergy consumption has become a major design constraint in modern computing systems. With the advent of petaflops architectures, power‐efficient software stacks have become imperative for scalability. Techniques such as dynamic voltage and frequency scaling (called DVFS) and CPU clock modulation (called throttling) are often used to reduce the power consumption of the compute nodes. To avoid significant performance losses, these techniques should be used judiciously during parallel application execution. For example, its communication phases may be good candidates to apply the DVFS and CPU throttling without incurring a considerable performance loss. They are often considered as indivisible operations although little attention is being devoted to the energy saving potential of their algorithmic steps. In this work, two important collective communication operations, all‐to‐all and allgather, are investigated as to their augmentation with energy saving strategies on the per‐call basis. The experiments prove the viability of such a fine‐grain approach. They also validate a theoretical power consumption estimate for multicore nodes proposed here. While keeping the performance loss low, the obtained energy savings were always significantly higher than those achieved when DVFS or throttling were switched on across the entire application run. Copyright © 2012 John Wiley & Sons, Ltd.