SANTA CLARA, Calif., June 14, 2017 — To succeed in the face of aggressive global competition, semiconductor design companies must minimize time-to-market without sacrificing the quality of their critical chip designs. This balancing act has become increasingly difficult as semiconductor technology evolves to necessitate more complex, integrated design workflows. Often, the speed and efficiency of data management is the limiting factor, according to industry experts at Elastifile.
Semiconductor companies leverage a variety of electronic design automation (EDA) tools to manage their workflows for commercial chip design, simulation and verification. Those tools can generate huge amounts of data, placing tremendous pressure on an internal IT infrastructure and creating bottlenecks that delay product release. With data volume continually increasing in proportion to design complexity, traditional storage infrastructure is no longer up to the task.
Meanwhile, the emergence of the public cloud has enabled on-demand, elastic access to an unlimited pool of compute and storage resources. Historically, chip design firms have been slow to embrace cloud integration, due to concerns regarding IP security and tool licensing costs. That landscape is now changing, however. Security measures in the public cloud security are now far more robust and sophisticated than those typically employed in on-premises data centers. Also, tool vendors, such as Cadence®, are now introducing “cloud-ready” EDA platforms with flexible licensing options to facilitate cloud-scale deployments. This evolving ecosystem is now catalyzing a strong desire to burst EDA workloads into the cloud for accelerated, massively parallel processing.
Considering these trends, it’s now clear that siloed, vendor-specific storage arrays do not provide the performance and flexibility needed to properly support modern chip design workflows. As a result, semiconductor IT departments need better solutions to efficiently store, process and share their critical data.
Crucial characteristics of an effective EDA data management solution include:
- Consistent, high performance – Faster “time to result” means reduced time-to-market
- Scalability to handle large data sets – Increasing data volume necessitates seamless capacity scaling
- Simple, unified data management – Unifying data within a single global namespace reduces management complexity
- Future-proofed, cloud-ready architecture – An enduring IT solution must flexibly and efficiently integrate emerging IT technologies (e.g. public cloud, flash storage)
To meet the challenges presented by modern design workflows, semiconductor companies should pursue software-defined, elastically scalable, parallel file systems. These file systems must be designed to deliver the performance and flexibility required to remove storage bottlenecks and accelerate mission-critical EDA workflows.
“Modern chip design often requires hyper-detailed simulation and verification of electronic circuits, sensors, IC packages, and more…generating vast amounts of EDA data,” said Shahar Frank, Elastifile CTO and co-founder. “To avoid costly time-to-market delays, the underlying IT infrastructure must scale to accommodate massive data volume, while also delivering the performance required to accelerate EDA analysis. As a result, efficient chip design now requires flash-native, software-defined file systems with unconstrained scalability. These next-generation file systems will support the necessary acceleration of semiconductor IT whether running on-premises, in the cloud, or in hybrid cloud environments.”
To learn more about Elastifile’s solution for EDA infrastructure and the benefits of a high-performance, cross-cloud data fabric, please see the following resources:
Elastifile solution brief on Parallel Computing for EDA: http://www.elastifile.com/wp-content/uploads/epub/pdf/Solutions-Brief-Parallel-Computing-for-EDA.pdf
Elastifile blog discussing the evolution of semiconductor IT: https://blog.elastifile.com/respect-the-silicon-and-bring-on-the-cloud