Can Nvidia jump the regulatory hurdles to capture Arm—and how?; OpenCAPI is the right answer for memory and I/O—so what’s missing?; Future-proofing a semiconductor/software strategy for edge AI; How one neuromorphic chip startup sees a path.
The IT sector has been digesting the technical and product line implications of the $40 billion deal for Nvidia to acquire Arm. And now we need to understand the global economic and political implications to assess whether the governments in the United States, the United Kingdom, Europe, and China, who all seem to have a say, will allow this deal to go through. We walk through the issues with David Bicknell, thematic research analyst at GlobalData.
Also on the program, FPGAs have long since moved to very high performance signaling that converges I/O and memory traffic so they can be configured in a flexible fashion as a system is being created and not on some spec sheet that locks the ratio of memory and I/O down. So why don’t CPUs do the same thing? We talk to Allan Cantle, newly anointed technical director of the OpenCAPI Consortium about this and other I/O issues.
Later in the show we take a look at how one AI chip startup is looking at future proofing from an architectural, software, and business point of view in the crowded computer vision market—an area that’s even more jam-packed at the edge. We delve into this with Sima.AI founder and CEO, Krishna Rangasayee, a veteran in the semiconductor space.
We conclude today’s program with a chat about the opportunities in neuromorphic computing segment from an applications/workload perspective as well as a technology one. This conversation is with Jack Kendall, CTO at Rain Neuromorphics.
Thanks as always for tuning in. Timestamps for the segments can be found below.
0:49 Can Nvidia jump the regulatory hurdles to capture Arm—and how?
17:04 OpenCAPI is the right answer for memory and I/O—so what’s missing?
30:46 Future-proofing a semiconductor/software strategy for edge AI
43:25 How one neuromorphic chip startup sees a path.