Skip to content (access key 's')
Logo of Technion
Logo of CS Department
Logo of CS4People
Events

The Taub Faculty of Computer Science Events and Talks

ceClub: Fast and Efficient Genomics Compute with the First Available Processing in Memory (PIM) on DRAM
event speaker icon
Gilles Hamou (UPMEM)
event date icon
Thursday, 18.11.2021, 14:30
event location icon
Zoom Lecture: 97649267246
Fast and efficient genomics compute with the first available Processing In memory (PIM) on DRAM Processing In memory (PIM), as UPMEM developed it - programmable processors integrated on DRAM die and orchestrated through the host processor -, are now widely accepted, and promoted as the next evolution in compute architecture. Memory vendors, JEDEC, now processor vendors, users, IT labs… are now exposing PIM in their plans. PIM’s benefits are based on structural considerations about the energy dominant cost of data movement in the compute node, and the inherent bottlenecks associated with it.
PIM can disrupt the compute node architecture as PIM provides unmatched performance, efficiency, costs – one order of magnitude better for each than app servers - and scalability. Those are now demonstrated with UPMEM PIM product with leading customers and IT labs on widespread big data and AI apps, with PIM app servers dispatching hundreds of GBs of PIM-DRAM embarking thousands of PIM units.
Scalability must be built in to ensure PIM successful development. That means Hardware, System and Software scalability. Designing and manufacturing within DRAM process capabilities, within DDR protocols, within server motherboards and BIOS capabilities, including DIMM slots, leveraging compilers and other SW tools, within programmers’ skills… are key pillars. Scalability of PIM, and thus its outlook is work in progress, with JEDEC DDR-PIM standardization groups.
Since early days, DNA sequencing workloads such as NGS mapping and variant calling and several others, have been tackled worldwide by genomics customers, labs and UPMEM application engineers. PIM porting and results will be presented, showcasing 10x improvements in speed, energy consumption and cost. Bio: Gilles is a French senior high-tech entrepreneur with several start-up and scale-up successes ranging from Internet to telecoms and semiconductors.
Gilles led the development of UPMEM since it was founded in 2015, from ideation to productization and early sales. Gilles has successfully created and consolidated a unique expert team and top tier global eco-system aligned with the disruption Processing In Memory is bringing to the market.
Gilles’s entrepreneurial success leverages its Engineer from Ecole Centrale Paris and INSEAD MBA training as well as its early career in strategy & management consulting at the Boston Consulting Group.