"...It's our position that 3U VPX is superior to 6U VPX."
Editor’s note: When you think of Extreme Engineering Solutions (X-ES), you probably think “chassis and cards provider.” But, as an utterly honest interview with Extreme Engineering Solutions’ President and CEO Rob Scidmore reveals, the company also sees itself as a systems integrator contending with GE Intelligent Platforms or Curtiss-Wright. Scidmore also tells where X-ES sits in the Intel versus Freescale paradox and why he feels VME is a dying art. Edited excerpts follow.
VME: Extreme Engineering popped onto our radar screen a couple of years ago: a small company, middle America, building VME boards and other cool stuff. Let’s start with some company metrics.
SCIDMORE: This summer we should reach about 120 employees. We’ve been around since 2002, and last year we did about $32 million in revenue.
VME: So your company is known as focusing on customized COTS. How is such a custom solution faring in an increasingly COTS era?
SCIDMORE: Basically, really quite well. If you follow our growth path – we doubled our revenue last year. But more importantly, we do have standards-based products. We’ve found that most customers don’t really need customized versions of their products immediately. Usually we can lead with the standards-based products and provide the modified COTS in short order.
VME: How long does it take Extreme Engineering to design something new from scratch? Tell me about that process.
SCIDMORE: Well it really depends on what you mean by “design from scratch.” There are certain products out there like a brand-new Freescale or a brand-new Intel product, where it literally takes probably six months to a year to bring something to the market. The majority of the modifications we see are not, “Oh by the way, I want this new processor.” If you look at our road map, we support every major Freescale high-performance processor, and we’re supporting all the embedded Intel high-performance processors. And once you’ve gotten the bulk of the hard work done (developing and supporting that from a software perspective and understanding how their product works), you’re turning that into a modified version where it’s wrapped around CompactPCI, VME, or VPX, which is fairly trivial.
VME: You mentioned three form factors you build your boards in accordance with. Which other form factors?
SCIDMORE: PrPMC and XMC and have really been the strong suit of our company for a lot of years. We have a strong presence in the 6U CompactPCI field, but I think we are the leading company worldwide in terms of developing PrPMCs.
VME: We don’t see PrPMCs very often. Explain why you’re a leading supplier there.
SCIDMORE: Every embedded-level form factor supports PMC and XMC. It’s always there, but in one regard, it’s fairly rare for someone to plug a processor on a processor board. So a lot of the times where we’re being pulled in is – I’ll use the analogy that’s the same as COM Express: Somebody wants to add a processer to their design, their special sauce, and they need a way to do that on a custom or standard form factor. Well, that would be where PrPMC would fit very well. It’s a standard product that can plug onto their custom or standard form factor card, much like COM Express plugs onto a custom or standard form factor card.
VME: Extreme Engineering was “into” the P.A. Semiconductor processors in the pre-Apple days, but looks like you’ve turned to Core i7 and Freescale offerings like the PowerQUICC and QorIQ. Which processor do your VME customers ask for most these days?
SCIDMORE: If you look at Freescale and Intel, we have a road map that basically says we are going to take on all the high-performance QorIQ solutions like the 4080 and the follow-on beyond that, or the Core i7 and the follow-on beyond that. We’re going to be there no matter what, and we already had those products before P.A. Semi. We’ve been with Freescale since 2003, and Intel probably two years before P.A. Semi came along. In terms of the customers who came to us as a result of P.A. Semi, I would say about 80 percent of them have gone to Intel and about 20 percent of them have moved on to Freescale. But people who used P.A. Semi and VME – I see those people sticking PowerPCs.
VME: So new designs then overall – all your form factors – which processor are you finding is most popular?
SCIDMORE: I think it’s a split. We’re doing about 50/50 [Intel and Freescale].
VME: OK, moving on, tell me about VPX versus OpenVPX … as you’re seeing it.
SCIDMORE: I don’t really see a distinction for us at this point. All of our products are OpenVPX. There are basically two profiles that address the majority of our customers – and the profiles pretty much define the backplane topology, not really the processor topology.
VME: Are you doing 3U or 6U OpenVPX, and which is more popular?
SCIDMORE: With OpenVPX, clearly 3U for us – that’s the most popular. Of course, it’s smaller, and then, more importantly, it’s our position that 3U VPX is superior to 6U VPX.
VME: Oh really? Why is that?
SCIDMORE: You put the same processor on the board, right, 3U or 6U? You’re taking the same amount of power and you’re putting it to the rail, taking it to the wedge lock. If that power has to go a longer distance, you’re going to have a larger drop. So in terms of thermal performance of a box, a 3U solution is better than the 6U. And what makes it even more complicated is when they write the standards, they somehow think that 6U should be able to handle more power. It’s just the opposite. If you’re talking about high-performance processors, it’s always a challenge to get the thermal solution that pulls that heat off the boards, the rails, and out of the box.
In terms of the 6U solution, you’ve got to take that heat further and you’re talking about putting more power in it. So I think what it comes down to is – if your application has so much I/O that you can provide a more dense solution with 6U, then you should use 6U. If your application is thermally limited, then 3U makes more sense. For air-cooled solutions, 6U probably has the advantage in that you can get more I/O out. But, if you’re looking at a conduction-cooled solution, then 3U is clearly much better.
VME: Do you consider yourselves a system integrator or a provider of cards and chassis?
SCIDMORE: We have 3U chassis. We do not have a 6U chassis. I would say we’re doing system-level integration, as well, and there’s a desire for the market to see us as a systems integrator. I would say if the industry doesn’t see us that way now, they will shortly.
VME: Who do you consider your biggest two competitors?
SCIDMORE: I would think that the customer doing it themselves is the largest competitor right out of the gate. I would say we probably are more of a match for GE and Curtiss-Wright in terms of which products we provide than Kontron or Mercury.
VME: OK. Getting back to OpenVPX, what are your predictions? How much of a difference will it really make in the VME culture?
SCIDMORE: Since the OpenVPX announcement [at MILCOM] and that meeting, it’s almost like somebody flipped a switch that’s turning off CompactPCI and VME and gives system designers in the company the excuse to go back to their management and say, “No, VME and CompactPCI don’t make any sense.” Now they can go after these new standardized serialized buses that are so popular today.
Conversely, VME is a very, very old technology – a giant parallel bus. There’s no new money being put into it, so I can’t understand why anybody would go there. Ironically, with the majority of people we’ve talked to, they want VME as a form factor; they don’t want the bus.
VME: What do you think is the future of VME itself?
SCIDMORE: I think it’s a slowly dying part of the industry. The slowly dying applications and the people who are boxed in – no pun intended – will continue to use it.
VME: Switching gears, VITA/VSO is getting into small form factor initiatives (VITA 73, 74, and 75). Do you use small form factors, or are you planning to?
SCIDMORE: The fact of the matter is this: They fall into a camp. There’s a group of people out there who want to define it at the box level. They want to say, “You define it at the connectors,” and, “What do you care what’s inside the box?” Those people have a good point. When I look at the new VITA standards, I see people trying to drive a new board-level standard in addition to the box-level standard for small form factors. From our perspective, that’s basically a mistake. They’re biting off more than they can chew, and that means the standard is going to take years to get anywhere. The mistake is that people are using it to define something new, rather than figuring out how to use what’s already out there.
VME: Does Extreme Engineering play in any of the VSO working groups?
SCIDMORE: We definitely did with OpenVPX.
VME: All right. What changes do you see coming in mil/defense in the next 5 to 10 years?
SCIDMORE: I say the biggest challenge for the mil/aero guys is the disappearing of leaded solutions and what that’s going to bring. That’s probably the No. 1 thing on my list. Probably No. 2 would be that SERDES interfaces in excess of 5 GHz per second is another big problem coming down the pike. Routing them on boards and backplanes inside the box and supporting those high-performance serial links in large systems – making that stuff work reliably across the full spectrum of temperature and environmentals could be a challenge.
VME: Okay, what should we look for from Extreme Engineering going forward?
SCIDMORE: More VPX, more system-level products, just sort of an overall expansion. We’re doing quite well. Just more of the same. CS
Extreme Engineering Solutions (X-ES)