Pairwise sequence alignment is an important application to identify regions of similarity that may indicate the relationship between two biological sequences. This is a computationally intensive task that usually requires parallel processing to provide realistic execution times. This work introduces a new framework for a deadline constrained application of sequence alignment, called MASA-CUDAlign, that exploits cloud computing with Spot GPU instances. Although much cheaper than On-Demand instances, Spot GPUs can be revoked at any time, so the framework is also able to restart MASA-CUDAlign from a checkpoint in a new instance when a revocation occurs. We evaluate the proposed framework considering five pairs of DNA sequences and different AWS instances. Our results show that the framework reduces financial costs when compared to On-Demand GPU instances while meeting the deadlines even in scenarios with several instances revocations.

A Fault Tolerant and Deadline Constrained Sequence Alignment Application on Cloud-Based Spot GPU Instances.

. The exponential growth in demand for digital services drives massive datacenter energy consumption and negative environmental im-pacts. Promoting sustainable solutions to pressing energy and digital infrastructure challenges is crucial. Several hyperscale cloud providers have announced plans to power their datacenters using renewable energy. However, integrating renewables to power the datacenters is chal-lenging because the power generation is intermittent, necessitating ap-proaches to tackle power supply variability. Hand engineering domain-speciﬁc heuristics-based schedulers to meet speciﬁc objective functions in such complex dynamic green datacenter environments is time-consuming, expensive, and requires extensive tuning by domain experts. The green datacenters need smart systems and system software to employ multiple renewable energy sources (wind and solar) by intelligently adapting computing to renewable energy generation. We present RARE ( R enewable energy A ware Re source management), a Deep Reinforcement Learning (DRL) job scheduler that automatically learns eﬀective job scheduling policies while continually adapting to datacenters’ complex dynamic environment. The resulting DRL scheduler performs better than heuristic scheduling policies with diﬀerent workloads and adapts to the intermittent power supply from renewables. We demonstrate DRL scheduler system design parameters that, when tuned correctly, produce better performance. Finally, we demonstrate that the DRL scheduler can learn from and improve upon existing heuristic policies using Oﬄine Learning.

RARE: Renewable Energy Aware Resource Management in Datacenters

Abstract The world is becoming increasingly dependant in computing intensive appliances. The appearance of new paradigms such as Internet of Things (IoT), and advances in technologies such as Computer Vision (CV) and Artificial Intelligence (AI) is creating a demand for high performance applications. In this regard, Graphics Processing Units (GPUs) have the ability to provide better performance by allowing a high degree of data parallelism. This devices are also beneficial in specialized fields of manufacturing industry such as CAD/CAM. For all this applications, there is a recent tendency to offload this computations to the Cloud, using a computing offloading Cloud architecture. However, the use of GPUs in the Cloud presents some inefficiencies, where GPU virtualization is still not fully resolved, as our research on what main Cloud providers currently offer in terms of GPU Cloud instances shows. To address this problems, this paper first makes a review of current GPU technologies and programming techniques that increase concurrency, to then propose a Cloud computing outsourcing architecture to make more efficient use of this devices in the Cloud. 

Efficient GPU Cloud architectures for outsourcing high-performance processing to the Cloud

Software approaches for resilience of high performance computing systems: a survey

HEGrid: A High Efficient Multi-Channel Radio Astronomical Data Gridding Framework in Heterogeneous Computing Environments

The share of the top 500 supercomputers with NVIDIA GPUs is now over 25% and continues to grow. While fault tolerance is a critical issue for supercomputing, there does not currently exist an efficient, scalable solution for CUDA applications on NVIDIA GPUs. CRAC (Checkpoint-Restart Architecture for CUDA) is a new checkpoint-restart solution for fault tolerance that supports the full range of CUDA applications. CRAC combines: low runtime overhead (approximately 1% or less); fast checkpoint-restart; support for scalable CUDA streams (for efficient usage of all of the thousands of GPU cores); and support for the full features of Unified Virtual Memory (eliminating the programmer’s burden of migrating memory between device and host). CRAC achieves its flexible architecture by segregating application code (checkpointed) and its external GPU communication via non-reentrant CUDA libraries (not checkpointed) within a single process’s memory. This eliminates the high overhead of inter-process communication in earlier approaches, and has fewer limitations.

CRAC: Checkpoint-Restart Architecture for CUDA with Streams and UVM

The share of the top 500 supercomputers with NVIDIA GPUs is now over 25% and continues to grow. While fault tolerance is a critical issue for supercomputing, there does not currently exist an efficient, scalable solution for CUDA applications on NVIDIA GPUs. CRAC (Checkpoint-Restart Architecture for CUDA) is new checkpoint-restart solution for fault tolerance that supports the full range of CUDA applications. CRAC combines: low runtime overhead (approximately 1% or less); fast checkpoint-restart; support for scalable CUDA streams (for efficient usage of all of the thousands of GPU cores); and support for the full features of Unified Virtual Memory (eliminating the programmer's burden of migrating memory between device and host). CRAC achieves its flexible architecture by segregating application code (checkpointed) and its external GPU communication via non-reentrant CUDA libraries (not checkpointed) within a single process's memory. This eliminates the high overhead of inter-process communication in earlier approaches, and has fewer limitations.

Autonomous unmanned aerial vehicles are complex systems of hardware, software, and human input. Understanding this complexity is key to their development and operation. Information visualizations already exist for exploring flight logs but comprehensive analyses currently require several disparate and custom tools. This design study helps address the pain points faced by autonomous unmanned aerial vehicle developers and operators. We contribute: a spiral development process model for grounded evaluation visualization development focused on progressively broadening target user involvement and refining user goals; a demonstration of the model as part of developing a deployed and adopted visualization system; a data and task abstraction for developers and operators performing post-flight analysis of autonomous unmanned aerial vehicle logs; the design and implementation of DATA COMETS, an open-source and web-based interactive visualization tool for post-flight log analysis incorporating temporal, geospatial, and multivariate data; and the results of a summative evaluation of the visualization system and our abstractions based on in-the-wild usage. A free copy of this paper and source code are available at this http URL

Data Comets: Designing a Visualization Tool for Analyzing Autonomous Aerial Vehicle Logs with Grounded Evaluation

A system and method for software checkpoint-restoration between distinctly compiled executables is disclosed. A first compiled version of the software, such as Version A, is executed. After which, checkpointing is performed in order to generate a checkpoint image. After checkpointing, restarting execution is performed with at least some of a second compiled version of the software, such as Version B, being executed using a switching function that is configured to switch execution upon restart at least partly to the second compiled version of the software. In this way, different executable versions may be used during the restart than during the initial execution, such as an unoptimized build during the restart versus an optimized build during the initial execution, so that software testing and/or debugging may be performed more efficiently.

Software checkpoint-restoration between distinctly compiled executables

The SPAdes assembler for metagenome assembly is a long-running application commonly used at the NERSC supercomputing site. However, NERSC, like many other sites, has a 48-hour limit on resource allocations. The solution is to chain together multiple resource allocations in a single run, using checkpoint-restart. This case study provides insights into the "pain points" in applying a well-known checkpointing package (DMTCP: Distributed MultiThreaded CheckPointing) to long-running production workloads of SPAdes. This work has exposed several bugs and limitations of DMTCP, which were fixed to support the large memory and fragmented intermediate files of SPAdes. But perhaps more interesting for other applications, this work reveals a tension between the transparency goals of DMTCP and performance concerns due to an I/O bottleneck during the checkpointing process when supporting large memory and many files. Suggestions are made for overcoming this I/O bottleneck, which provides important "lessons learned" for similar applications.

Twinkle Jain

Papers

CRAC: Checkpoint-Restart Architecture for CUDA with Streams and UVM

CRAC: Checkpoint-Restart Architecture for CUDA with Streams and UVM

Data Comets: Designing a Visualization Tool for Analyzing Autonomous Aerial Vehicle Logs with Grounded Evaluation

Software checkpoint-restoration between distinctly compiled executables

Checkpointing SPAdes for Metagenome Assembly: Transparency versus Performance in Production.