httpv://www.youtube.com/watch?v=S_mMles-vcI
In this video from EE Times, Nvidia’s Sumit Gupta describes why energy-efficient ARM processors is better option for supercomputing going forward.
The number one consideration for x86 has always been to make operating systems like Windows run much faster and to be able to respond to unpredictable tasks, such as a mouse-click or a keyboard entry,” said Gupta, noting that the need for branch prediction and speculative execution was the reason x86 processors had such sizeable cache. “It’s a terrific processor for everyday computing, not the right device as we go towards high performance computing,” he maintained.
Read the Full Story.
This is just stupid. x86 processors have been optimized for a wide range of workloads over the years, from windows desktops to computationally intensive applications. The typical x86 CPU just has one “coprocessor”: The GPU. ARM on the other hand is a platform optimized for smartphones and embedded systems with lots of coprocessors, offloading as most as possible. The standard ARM CPU does not even come with a full integer ALU and has to emulate simple divisions. Apart from all the other problems, like no 64 bit support and an MMU which cannot handle more than 4 GB of RAM for a single operating system instance.
We have yet to see an ARM compute node which actually performs. Up to that point, this seems more like a desperate stunt from NVIDIA. And the combination ARM/CUDA is certainly not making software development easier for anybody.
What a commercial… Sumit doesn’t seem to having heard about iPSC, Paragon, ASCI, etc. Fair to say that it’s the wide spectrum of capabilities and standards of x86 which made HPC mainstream.