The next generation AMD CPUs and GPUs will use some of the same technologies, optimizations, and interconnects designed for the supercomputers of the US DOE, Oak Ridge National Library. According to Lisa Su, chief executive officer of AMD, the Frontier system, scheduled to arrive in 2021 is currently being produced by AMD and Cray. The system will be loaded with AMD EPYC along with Radeon Instinct's GPUs to optimize the modeling as well as the simulation of complex structures, further research, and weather and surprisingly, the rest of the world can benefit from the company's latest massive project.
During the Hot Chips Symposium 31, AMD's chief executive officer shared some details about the future of the semiconductor industry. But, the discussion was immediately sidetracked by different queries, including Crossfire returning and the 7nm Threadripper, which were also discussed at the said event. Additionally, amazing tidbits of the next generation silicon development were also talked about.
Lisa Su announced AMD's collaboration with Oak Ridge National Labs, the Department of Energy, and Cray which will produce the next-generation supercomputers at Oak Ridge. The CEO revealed that the project, codenamed Frontier, is the system that will utilize all optimizations that she earlier discussed. Moreover, she shared that the public could expect a leadership system with highly optimized GPU and CPU, and highly enhanced coherent interconnect between the GPU and the CPU.
The CEO also shared that these technologies are designed for the highest performance supercomputers that have massive system capabilities and power. Interestingly, the AMD boss announced that these AMD supercomputers will be available in more commercial systems and will be included in the company's next-generation CPUs and also in its next-generation GPUs to boost computing efficiency. The CEO highlighted what it takes to optimize computational efficiency in 2019, and it no longer includes bigger, denser, and a better processor.
Moreover, she noted several areas that are left behind the trajectory of the customary computer. Communication between the GPU and the CPU as well as memory bandwidth are being developed at a slow pace, and the ball is now in the hands of AMD, to boost the pace of change and close the gap towards a more customized, application-tailored accelerators and ICS.
When it comes to interconnect technology, AMD already taps its very own Infinity Fabric interconnect between dies on its AMD EPYC CPUs and Ryzen 3000. The AMD CEO also said that there is a greater need to have hardware and software co-optimization so users could get the best out of the performance capabilities.
