A Q&A with David Pepper
The last eight months have been tough for David Pepper, one of our senior product managers. He had a serious bicycle accident that left him immobile. A spell in a manual wheelchair – using his legs for propulsion – saw him back on his feet three months ago, although still with mobility challenges. That’s meant he’s been working part time for the past few months as he works his way through a recovery process which will take at least a year and a half.
David joined VMIC – which became part of GE’s Intelligent Platforms business, the embedded computing part of which became Abaco – in 1998. As such, he has a unique perspective on developments in the embedded computing world – so The Connected Battlefield took advantage of his enforced downtime to ask him to share his thoughts.
What are your memories of what embedded computing looked like two decades ago?
VME was really the dominant platform at that time. An interesting dynamic we saw was the emerging interest in moving away from Motorola-based computing to other processing architectures with Intel being the most interesting one. Intel was a challenge for VME as it is a little endian versus big endian format. This led us at VMIC to developing our patented VME byte swapping circuitry, which served us very well for several generations of Intel-based product.
CompactPCI also began gaining a lot of traction back in those days. Although originally slated for telecom applications, the 3U form factor became very successful in military and aerospace applications that required a smaller footprint.
What kind of applications do we take for granted today that were not possible 20 years ago?
I would say high performance embedded computing (HPEC) in rugged applications is something that has become mainstream. These days, high performance Linux-based computing clusters can be built in both 3U and 6U form factors. The ability to have multiple computing nodes interconnected via a high-speed fabric is fairly common, with speeds now up to 40 Gb per second be typical between nodes.
What has been the most fun thing you’ve been involved in since you joined VMIC?
Really, that would be involvement with the standards community working on things like VPX and OpenVPX and seeing the technology evolve in both computing, memory architectures and high-speed fabrics has been a lot of fun. Although I do admit: some days, I wish IT would slow down, for now there seems to be no end to it.
VME was a bus-centric architecture. Now, with VPX, we have a fabric-centric architecture? What do you think the latter brings?
Really that goes back to the second question. You really have the ability now with high-speed interconnects to build a very high performance computing platform.
For you, what has been the most exciting technology breakthrough of the last 20 years?
I would say the ability of a company like Intel to continue to shrink the transistor such that we can have either twice as many components or twice as much performance in the same footprint. That’s really been a game changer. Honestly: I thought they would run out of steam 10 years or more ago, but breakthroughs in transistor technology and material science has kept the trend going. Now that we see geometries of 10 nm and less, I think we might be about out of steam - but we may be just another breakthrough way from yet more capability.
What do you foresee in embedded computing in the next two decades?
I just recently picked up a book about quantum physics as I find that a fascinating subject. As we learn to work at the subatomic level, I think things are really going to get interesting in terms of compute capability. We could easily see a breakthrough that could allow for orders of magnitude of performance increase over technologies that are available today.
Many people have reached out to me since my accident – something I truly appreciate,” David concluded. “My recovery continues, I’m getting better all the time – and I’m encouraged by the improvement I’ve seen so far.”