Clarete Crasta

Clarete Crasta

Senior Principal Engineer, HPE HPC/AI BU. Distributed Systems | HPC | Disaggregated Memory | FAM | Resiliency | Programming Models.

Posts by Collection

portfolio

publications

New Time Features on OpenVMS and Read-Write Consistency Data Check Algorithm

Published in OpenVMS Technical Journal, Vol. 15, 2010

Describes new time feature implementations on the OpenVMS operating system and presents a read-write consistency data check algorithm for ensuring data integrity in multi-processor environments.

Recommended citation: C.R. Crasta, B. Fisher. (2010). "New Time Features on OpenVMS and Read-Write Consistency Data Check Algorithm." OpenVMS Technical Journal, Vol. 15.
Download Paper

A Method and Apparatus for Faster, Reliable and Consolidated Logging of Information During System Failure

Published in TD Commons (Technical Disclosure), 2016

Proposes a method for faster, more reliable, and consolidated logging of diagnostic information during system failures, reducing recovery time and improving root-cause analysis in enterprise systems.

Recommended citation: C.R. Crasta, S.S. Yelandur, S. Shivanna. (2016). "A Method and Apparatus for Faster, Reliable and Consolidated Logging of Information During System Failure." TD Commons.
Download Paper

OpenFAM: A Library for Programming Disaggregated Memory

Published in Workshop on OpenSHMEM and Related Technologies (OpenSHMEM 2021), 2021

Introduces the OpenFAM library for programming disaggregated memory in HPC systems, presenting its API design, programming model, and initial performance results on fabric-attached memory hardware.

Recommended citation: S. Singhal, C.R. Crasta, M. Abdulla, F. Barmawer, D. Emberson, R. Ahobala, et al. (2021). "OpenFAM: A Library for Programming Disaggregated Memory." Workshop on OpenSHMEM and Related Technologies.
Download Paper

MODC: Resilience for Disaggregated Memory Architectures Using Task-Based Programming

Published in Workshop on Resource Disaggregation and Serverless (WORDS 2021) & arXiv:2109.05329, 2021

Presents MODC (Memory-Oriented Distributed Computing), a task-based programming framework for disaggregated memory architectures that achieves resilience through task-level recovery and selective re-execution of failed tasks, drastically reducing mean time to recovery versus whole-system checkpointing.

Recommended citation: K. Keeton, S. Singhal, H. Volos, Y. Zhang, R.C. Chaurasiya, C.R. Crasta, et al. (2021). "MODC: Resilience for Disaggregated Memory Architectures Using Task-Based Programming." Workshop on Resource Disaggregation and Serverless (WORDS 2021). arXiv:2109.05329.
Download Paper

Extending Chapel to Support Fabric-Attached Memory

Published in Chapel User Group (CHIUW 2022), 2022

Presents language extensions to Chapel that enable transparent access to Fabric Attached Memory (FAM), allowing Chapel programs to seamlessly exploit disaggregated memory resources in HPC environments.

Recommended citation: C. Amitha, B.L. Chamberlain, C.R. Crasta, S. Singhal. (2022). "Extending Chapel to Support Fabric-Attached Memory." Chapel User Group (CHIUW 2022).
Download Paper

OpenFAM: Programming Disaggregated Memory

Published in Cray User Group (CUG 2023), 2023

Presentation of the OpenFAM framework at the Cray User Group, covering programming disaggregated memory at scale with a focus on deployment lessons and practical performance results from HPC systems.

Recommended citation: S. Singhal, C.R. Crasta, et al. (2023). "OpenFAM: Programming Disaggregated Memory." Cray User Group (CUG 2023).
Download Paper

Coupling Chapel-Powered HPC Workflows for Python

Published in Chapel User Group (CHIUW 2023), 2023

Demonstrates coupling of Chapel-powered high-performance computing workflows with Python, enabling HPC practitioners to leverage Chapel parallel computing capabilities from Python environments.

Recommended citation: J. Byrne, H. Kuno, C. Ghosh, P. Shome, C. Amitha, S. Singhal, C. Riana, et al. (2023). "Coupling Chapel-Powered HPC Workflows for Python." Chapel User Group (CHIUW 2023).
Download Paper

Record Based Pointers to FAM

Published in Chapel User Group (CHIUW 2023), 2023

Introduces record-based pointer types for Fabric Attached Memory in the Chapel programming language, enabling efficient and type-safe access to remote disaggregated memory regions.

Recommended citation: C. Amitha, C.R. Crasta, B.L. Chamberlain, P. Shome, S. Singhal. (2023). "Record Based Pointers to FAM." Chapel User Group (CHIUW 2023).
Download Paper

Fabric Attached Memory — Hardware and Software Architecture

Published in SNIA Software Developer Conference (SDC 2023), 2023

A comprehensive overview of the hardware and software architecture of Fabric Attached Memory (FAM), presenting its design, integration with OpenFAM, RDMA-based communication protocols, and support for persistent memory hierarchies.

Recommended citation: C.R. Crasta, D. Emberson, S. Singhal. (2023). "Fabric Attached Memory — Hardware and Software Architecture." SNIA Software Developer Conference (SDC 2023).
Download Paper

Fabric Attached Memory — Hardware and Software Architecture

Published in SNIA Compute Memory Storage Summit (CMSS 2023), 2023

Positions Fabric Attached Memory within the broader landscape of storage innovation and emerging disaggregated memory models, discussing integration with OpenFAM and support for persistent memory hierarchies.

Recommended citation: C.R. Crasta. (2023). "Fabric Attached Memory — Hardware and Software Architecture." SNIA Compute Memory Storage Summit (CMSS 2023).
Download Paper

OpenFAM: Programming Disaggregated Memory

Published in Concurrency and Computation: Practice and Experience, 35(28), e7910, 2023

A comprehensive journal paper on the OpenFAM library for programming disaggregated memory at scale, presenting the full API, memory model, and performance results for latency optimization and fault tolerance in distributed memory hierarchies.

Recommended citation: S. Singhal, C.R. Crasta, M. Abdulla K, F. Barmawer, G. Bhat, R.A. Rao, S. PN, et al. (2023). "OpenFAM: Programming Disaggregated Memory." Concurrency and Computation: Practice and Experience, 35(28), e7910.
Download Paper

FAM support in OpenSHMEM

Published in Cray Users Group Australia (CUG Australia 2024), 2024

Extension of OpenSHMEM to support Fabric Attached Memory (FAM), enabling seamless integration of remote disaggregated memory into PGAS programming models for HPC applications.

Recommended citation: C.R. Crasta, S. Singhal, F.I. Barmawer, R.C. Chaurasiya, S. KV, D. Emberson, et al. (2024). "FAM support in OpenSHMEM." Cray Users Group Australia.
Download Paper

Global Distributed Client-side Cache for DAOS

Published in Proceedings of the Cray User Group (CUG 2025), pp. 50–58, 2025

A global client-side caching architecture for DAOS (Distributed Asynchronous Object Storage) that aggregates DRAM, SSDs, and network bandwidth from compute nodes to eliminate server-side bottlenecks and achieve significant performance gains for HPC and AI workloads.

Recommended citation: C.R. Crasta, J.L. Byrne, A. Dwaraki, D. Emberson, H.A. Kuno, S. Lee, et al. (2025). "Global Distributed Client-side Cache for DAOS." Proceedings of the Cray User Group (CUG 2025), 50–58.
Download Paper

A Query Engine for Scientific Data Exploration using Theory, Simulation, and Artificial Intelligence Models

Published in Proceedings of the SC'25 Workshops, 2025

A query engine designed for scientific data exploration that integrates theory, simulation, and AI models to support large-scale data-driven discovery.

Recommended citation: A. Dwaraki, S.R. Sukumar, C.D. Rickett, C.R. Crasta, H. Kuno, M. Neal, et al. (2025). "A Query Engine for Scientific Data Exploration using Theory, Simulation, and Artificial Intelligence Models." Proceedings of the SC'25 Workshops.
Download Paper

talks

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.