Tuesday, June 16th 2020
Tachyum Reveals HPC/AI Prodigy Motherboard at ISC 2020 Digital
Dr. Radoslav Danilak, founder and CEO of semiconductor startup Tachyum Inc., will discuss the company's "Gamechanger" universal processor reference design solution for HPC and AI environments as part of ISC High Performance 2020 Digital event June 22-25.
ISC 2020 Digital is an inaugural online event that focuses on bringing the most critical developments and trends in high-performance computing, machine learning and data analytics for the benefit of the global HPC community. Danilak presents "4 Exaflops AI Training With a 32 Rack Tachyum System" to help virtual attendees learn about the Tachyum universal processor architecture, performance, reference design motherboard, chassis, rack, switches, management software and system software that make up the reference design. The pre-recorded presentation will be available for viewing June 24 from 4 p.m. CEST to registered participants of the event and remain online for 14 days after it debuts.
Tachyum's 4 Exaflops AI Training system is based on the Tachyum Prodigy Universal Processor Chip, a 64-bit general purpose processor combining the best of CPU, GPGPU for HPC and AI accelerators like TPU.
A supercomputer based on Tachyum's reference design will be installed in 2021 at Slovak Academy of Science (SAV) in Bratislava, Slovak Republic. In addition, the reference design for customers will be available at Tachyum locations in Bratislava, Slovakia, and Santa Clara, California, USA. All hardware, software and switch designs will be open sourced, allowing end users to build low-cost scalable open systems dramatically reducing cost of acquisition and operation.
"From the Prodigy Universal Processor to the reference motherboard, to the reference 1U and 2U servers, our reference rack with Edgecore and SONiCS switches, and our 4 Exaflops AI Training system is a gamechanger by allowing for a simple programming model and environment based with a single chip within a coherent multiprocessor environment," said Danilak. "I look forward to sharing with attendees of ISC 2020 how our total solution will revolutionize hyperscale data centers in terms of offering significant improvements in performance, power consumption, server utilization and space requirements."
A frequent presenter at technology trade shows around the world, Danilak has more than 25 years of industry experience and over 100 patents designing state-of-the-art processing systems. Danilak is a U.S. citizen born in Slovakia and serves on the Slovak government's Innovation Advisory Board. He is a member of the IDC Technical Computing Advisory Panel, the Forbes Technology Council, and a contributor to TechTarget. He holds a Ph.D in Computer Science and an MS in Electrical Engineering from the TUKE Slovakia, in Kosice.
Prodigy outperforms the fastest Xeon processors at 10x lower power (core vs. core) on data center workloads, as well as outperforming NVIDIA's fastest GPU on neural net AI training and inference. Prodigy's 3X lower cost per MIPS compared to other CPU competition, coupled with its 10X processor power savings, translates to a 4X reduction in Data Center TCO (Annual Total Cost of Ownership: CAPEX + OPEX). Even at 50 percent Prodigy attach rates, this translates to billions of dollars per year in real savings for hyperscalers such as Google, Facebook, and Amazon.
Since Prodigy can seamlessly and dynamically switch from data center workloads to AI or HPC workloads, unused servers can be powered up, on demand, as ad hoc AI or HPC networks - CAPEX free, since the servers themselves are already purchased. Every Prodigy-provisioned data center, by definition, becomes a low-cost AI center of excellence, and a low-cost HPC system.
ISC 2020 Digital is the largest online HPC event of the year, combining invited and contributed programs with presentations from vendor partners sharing their latest product developments in the Vendor Showdown, Exhibitor Forum and in-house events.
ISC 2020 Digital is an inaugural online event that focuses on bringing the most critical developments and trends in high-performance computing, machine learning and data analytics for the benefit of the global HPC community. Danilak presents "4 Exaflops AI Training With a 32 Rack Tachyum System" to help virtual attendees learn about the Tachyum universal processor architecture, performance, reference design motherboard, chassis, rack, switches, management software and system software that make up the reference design. The pre-recorded presentation will be available for viewing June 24 from 4 p.m. CEST to registered participants of the event and remain online for 14 days after it debuts.
A supercomputer based on Tachyum's reference design will be installed in 2021 at Slovak Academy of Science (SAV) in Bratislava, Slovak Republic. In addition, the reference design for customers will be available at Tachyum locations in Bratislava, Slovakia, and Santa Clara, California, USA. All hardware, software and switch designs will be open sourced, allowing end users to build low-cost scalable open systems dramatically reducing cost of acquisition and operation.
"From the Prodigy Universal Processor to the reference motherboard, to the reference 1U and 2U servers, our reference rack with Edgecore and SONiCS switches, and our 4 Exaflops AI Training system is a gamechanger by allowing for a simple programming model and environment based with a single chip within a coherent multiprocessor environment," said Danilak. "I look forward to sharing with attendees of ISC 2020 how our total solution will revolutionize hyperscale data centers in terms of offering significant improvements in performance, power consumption, server utilization and space requirements."
A frequent presenter at technology trade shows around the world, Danilak has more than 25 years of industry experience and over 100 patents designing state-of-the-art processing systems. Danilak is a U.S. citizen born in Slovakia and serves on the Slovak government's Innovation Advisory Board. He is a member of the IDC Technical Computing Advisory Panel, the Forbes Technology Council, and a contributor to TechTarget. He holds a Ph.D in Computer Science and an MS in Electrical Engineering from the TUKE Slovakia, in Kosice.
Prodigy outperforms the fastest Xeon processors at 10x lower power (core vs. core) on data center workloads, as well as outperforming NVIDIA's fastest GPU on neural net AI training and inference. Prodigy's 3X lower cost per MIPS compared to other CPU competition, coupled with its 10X processor power savings, translates to a 4X reduction in Data Center TCO (Annual Total Cost of Ownership: CAPEX + OPEX). Even at 50 percent Prodigy attach rates, this translates to billions of dollars per year in real savings for hyperscalers such as Google, Facebook, and Amazon.
Since Prodigy can seamlessly and dynamically switch from data center workloads to AI or HPC workloads, unused servers can be powered up, on demand, as ad hoc AI or HPC networks - CAPEX free, since the servers themselves are already purchased. Every Prodigy-provisioned data center, by definition, becomes a low-cost AI center of excellence, and a low-cost HPC system.
ISC 2020 Digital is the largest online HPC event of the year, combining invited and contributed programs with presentations from vendor partners sharing their latest product developments in the Vendor Showdown, Exhibitor Forum and in-house events.
1 Comment on Tachyum Reveals HPC/AI Prodigy Motherboard at ISC 2020 Digital
That just sounds like a modern GPU.