Tachyum has revealed plans for its future supercomputers that are expected to offer several Exaflops of compute performance despite chip delay.
Tachyum reveals new 20 Exa-Flop & 10 AI Zetta-Flop supercomputer designs, offering up to 30 times the expected performance targets of the United States Department of Energy
The Important Projects of Common European Interests (IPCEI) program for Slovakia selected the Tachyum Prodigy 2 chip to assist with exascale high-performance computing (HPC) & zetta-scaling artificial intelligence (AI) for funding in Europe. The European Commission has also approved the company's 26.4M EUR Supercomputer in financing & is awaiting the finalized notification process.
Tachyum's reference designs are based on the company's systems, engineering teams, and developments to alter data centers & supercomputers into more universal computing centers, allowing for exascale and higher workloads utilizing HPC and AI to process simultaneously and within the same architectural family. The expected designs will be available in 2025.
Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores to deliver up to 4x the highest-performing x86 processors for cloud workloads, up to 3x that of the highest-performing GPU for HPC, and 6x for AI applications.
Tachyum's supercomputer designs are expected to produce 20 EXAFLOPS of FP64 vector performance, the top of the Department of Energy's target, and within the 60MW power envelope within an area of 6,000 square feet.
This supercomputer will also produce above 10 ZETAFLOPS of AI performance, which exceeds the DOE's target by 30 times the expected target. FP64 performance of up to 25 EXAFLOPS can be accomplished within the same area, depending on the power envelope's flexibility. The company has been claiming some big performance numbers against NVIDIA's H100 and Intel 3rd Gen Xeon Scalable processors but nothing has come out in physical form yet.
Tachyum provides leading-edge solutions from silicon to complete systems to address the ever-increasing demands for both HPC and AI. Tachyum-designed supercomputers push the forefront of HPC performance while crossing the zetta-scale barrier for AI, transforming data centers into universal computing centers.
— Radoslav Danilak, founder and CEO, Tachyum
Prodigy 2 data centers contain new infrastructure generations that are produced for networking and storage to employ the most advanced technology in the industry while assuring that the entire cluster runs at maximum efficiency and performance.
If you want to know more about Tachyum and the company's data center designs for high-performance computing and AI, you can read more about the chip here which was last talked about by the company back in October but no word of production has been out since then.
Refference- https://wccftech.com
0 Comments