When you have 54.2 billion transistors to play with, you can pack a lot of different functionality into a computing device, and this is precisely what Nvidia has done with vigor and enthusiasm with the new “Ampere” GA100 GPU aimed at acceleration in the datacenter. …
As the name of this publication suggests, we are system thinkers and we like to watch the evolution of a collection of tools into a platform. …
It is hard to remember sometimes way back when, in 2008, as Nvidia first took a stab at GPU compute in the datacenter with the original Tesla GPU accelerators and a very rudimentary CUDA programming environment for offloading parallel algorithms from CPUs to GPUs. …
A new CPU or GPU compute engine is always an exciting time for the datacenter because we get to see the results of years of work and clever thinking by hardware and software engineers who are trying to break through barriers with both their Dennard scaling and their Moore’s Law arms tied behind their backs. …
Companies invest in platforms over a decade or more, and that is why architectures persist longer than we might think given technological differences and economic forces. …
The GPU has become a standard platform for accelerating high performance computing workloads, at least for those that have had their code tweaked to support acceleration at all. …
Like their US-based counterparts, Google and Amazon, the Chinese Internet giants Baidu and Alibaba rely on GPU acceleration to drive critical parts of their AI-based services. …
It is funny to think that in a certain light, AMD has Big Blue to thank for its resurgence in the datacenter. …
The days when the X86 processor could do just about every kind of processing in the datacenter are gone. …
It would be convenient for everyone – chip makers and those who are running machine learning workloads – if training and inference could be done on the same device. …
All Content Copyright The Next Platform