The two-day online event offered Community members the chance to connect over several hours each day to hear from HPC thought leaders on topics such as quantum computing and neuromorphic computing, AI, multi-cloud and large-scale simulation and data analytics.
From there, attendees were able to listen to partner keynotes from NVIDIA, AMD, and Intel. The topics covered the latest performance-based EPYC architecture from AMD and NVIDIA's new cloud-native supercomputing architecture, as well as introduced new HPC forums, such as Intel's OneAPI initiative.
The event itself may have been online, but the content - from the keynotes, to the breakouts, to the CTO panel discussion - was happening in real time. Members were able to watch the live keynotes and then use the parallel breakout tracks to engage with peers on the topics that most interested them. Online events can get be difficult to navigate, but the feedback was overly positive.
"The platform was better than I have seen so far, with breakouts and the stage very easily navigated," said
While each breakout track had moderators to initiate and drive discussion, it was the insight from community members that made these tracks more conversational and thus more rewarding. Attendees found the content to be "well balanced with input from/for customer, partner,
During the breakouts, many members used the live chat tool to offer insight into their own best practices. Additionally, members shared where they see opportunities to improve and expand the HPC ecosystem in the future.
Several themes emerged over the course of the event:
Interoperability: many attendees highlighted the need for an infrastructure that can meet the performance and scalability required for a HPC environment. At the same time, there's a need to maintain open standards flexibility to support multiple workloads both in traditional on-premises, as well as emerging cloud-native environments. Many noted that the preference is to keep the infrastructure on-premises, as adoption of storage-as-a-service models - including cloud - is low. However, some members suggested that bursting during peak times when the cloud can help off-load some burden from the on-premises infrastructure.
Several attendees acknowledged that they include heating, cooling and power consumption as a part of purchasing decisions. Due to this, along with other considerations, such as lifecycle management, funding models, compliance and security, attendees discussed the value of adopting hybrid cloud strategies to leverage both on-premises and cloud infrastructure for their HPC workload needs.
Especially as use cases for HPC are merged with simulation and AI modeling workloads, data sets are growing larger. HPC customers seek the ability to abstract from silos towards heterogeneous resource pools while still providing a standard platform where they can run different applications. As multi-cloud environments grow, so does the need for the focus to remain on data portability and manageability to truly take advantage of a hybrid environment.
It's this type of dialogue - from partners, customers, and subject matter experts - which drives the mission of the
"
Attendees agreed. In fact, many said it was their favorite event they attended since the beginning of the pandemic. One member wished she could have "cloned myself" to attend more of the sessions in real time.
While the hope is for in-person events to sprout up again soon, it's clear that hybrid events - with both remote and in-person portions - will remain a part of the future for HPC communities. The good news is, either way, members of the
"While virtual Conferences are not the same as in-person events, the
To become a member, please visit www.dellHPC.org.
.
(C) 2021 M2 COMMUNICATIONS, source