| The Exiled |
30.09.2021. 12:34 |
AMD announces ambitious goal to increase energy efficiency of processors running AI training and High Performance Computing applications 30x by 2025:frend:
Citiraj:
AMD today announced a goal to deliver a 30x increase in energy efficiency for AMD EPYC CPUs and AMD Instinct accelerators in Artificial Intelligence (AI) training and High Performance Computing (HPC) applications running on accelerated compute nodes by 2025. Accomplishing this ambitious goal will require AMD to increase the energy efficiency of a compute node at a rate that is more than 2.5x faster than the aggregate industry-wide improvement made during the last five years.
|
Izvor: TechPowerUp i The Next Platform
Citiraj:
US closes in on Exascale: Frontier installation is underway
Citiraj:
The staff at the Oak Ridge Computing Facility, backed by the Exascale Computing Project community and technology partner HPE, is hard at work fielding the United States’ first exascale system this year, and ensuring that the system is ready for real science on day one. Built by HPE, Frontier will span 9,000+ Cray EX nodes, each consisting of one third-gen AMD Epyc CPU plus four Radeon Instinct MI200 GPUs, connected via Slingshot 11 networking. Frontier is projected to provide more than 1.5 exaflops of HPC and AI processing performance, according to ORNL.
|
Izvor: HPCwire
|
Citiraj:
NCAR prepares for Derecho, its third-generation Weather and Climate Supercomputer
Citiraj:
Derechos, the namesake of the new supercomputer coming to the National Center for Atmospheric Research (NCAR), are fast-moving, widespread bands of thunderstorms. Derecho, built by HPE, will be water-cooled and predominantly powered by third-generation AMD EPYC Milan CPUs and nVidia’s 40GB A100 GPUs, with 2,488 CPU-only dual-socket nodes (256GB of memory per node) and 82 single-socket heterogeneous nodes (four A100s and 512GB of memory per node). In total, the system is equipped with 692TB of total memory, 328 A100 GPUs and 5,058 Milan CPUs, all connected by HPE Slingshot v11 networking. This combined hardware will deliver 19.87 peak petaflops – more than triple the performance of Derecho’s predecessor, Cheyenne (5.34 peak petaflops). Cheyenne, installed in 2016, was itself preceded by the 1.26-peak petaflops Yellowstone system in 2012.
|
Izvor: HPCwire
|
Citiraj:
ASML Analyst Day: The future with DUV, EUV and High-NA is bright:kafa:
Citiraj:
ASML plans to introduce new extreme ultraviolet (EUV) lithography equipment that will extend the longevity of Moore’s Law for at least ten years, according to executives at the world’s only supplier of the tools, which are crucial for the world’s most advanced silicon. Starting in the first half of 2023, the company plans to offer customers equipment that takes EUV numerical aperture (NA) higher to 0.55 NA from the existing 0.33 NA. The company believes that the new equipment will help chip makers reach process nodes well beyond the current threshold (2nm) for at least another 10 years, according to ASML vice president Teun van Gogh, in an interview with EE Times.
The company, which has now grown to around 31,000 employees and is headquartered in the Netherlands, is in the spotlight like never before. It goes hand in hand with the foundries, which hardly anyone outside the technology sector knew before the chip crisis of recent times. But now it is also becoming clear there how important, but also large, these companies already are. And this demand will not really subside, ASML believes, the demand for more wafers will continue to increase. Thus, ASML also opposes the assumptions that overcapacities could quickly occur. According to their analyses, this will not apply to the market as a whole, because it needs chips in structure size from 3 nm to 350 nm, as ASML illuminates with some components on the new iPhone alone. However, because ASML earns a lot from capacity building, overcapacity would also be the worst conceivable scenario.
|
Izvor: ComputerBase i EE Times
|
|
