Planning your next HPC system? Some thoughts to consider.
Tuesday, June 10, 2025 5:00 PM to 5:20 PM · 20 min. (Europe/Berlin)
Hall H, Booth L01 - Ground floor
HPC Solutions Forum
Data Center Infrastructure and Cooling
Information
Most HPC systems today are planned based on inevitable restrictions such as budget, space, or cooling capacity.
System compute architecture is often a “given”, a common denominator for all applications and stakeholders, a safe compromise.
But for those willing to optimize, there are good practices to enhance a system within existing restrictions.
Some of these practices directly affect the processing architecture, others make systems more efficient by sharing resources.
Some approaches require long-term thinking and courage. The reward may be the best system you ever had.
System compute architecture is often a “given”, a common denominator for all applications and stakeholders, a safe compromise.
But for those willing to optimize, there are good practices to enhance a system within existing restrictions.
Some of these practices directly affect the processing architecture, others make systems more efficient by sharing resources.
Some approaches require long-term thinking and courage. The reward may be the best system you ever had.
HPC Solutions Forum Questions
Will Ethernet solutions (including Ultra Ethernet Consortium) be significant for HPC-AI supercomputers? Why or why not?What types of applications are not well served by “standard” processing architectures, and what is the solution?For organizations pursuing both traditional HPC and new AI workloads, to what extent should they have shared infrastructure / budget / personnel versus separate?
Format
On Site
