Very recently, Intel released their sixth-generation of Core processors, codenamed Skylake. More of an iteration of a long-term plan than a new path, Skylake is very similar to Haswell and Ivy Bridge in that it is tuned more for the race to idle rather than for pure speed. Its the first real consumer product based on Intel’s 14-nanometer production process (no, Broadwell doesn’t count), and it is responsible, along with Haswell-E, for accelerating sales of DDR4 memory and new products like NVME-compatible solid state drives. I was invited to Intel’s media briefing at the rAge 2015 expo, and if you follow me after the jump, you’ll hear some of the answers to the questions that I asked.
For the briefing, Intel was represented by Hitendra Naik, director of innovation for South and Sub-saharan Africa, and Mohammed Fareed, business development manager for the software and services group for Intel South Africa. Intel had two demo units for the media to tinker with after the briefing – one was a Core i7-6700K ITX system built into a Deepcool Trinity chassis, which Mohammed admitted was very difficult to build in to, and the other was an HP All-In-One system with a Intel RealSense camera and a few demos for it.
The first half of the discussion went through a quick run through of the Skylake family and what made it different to Haswell. Since I’ve written about Skylake before, nothing was new to me. The only thing I did learn is that technically, Skylake isn’t really launched locally – while there are a handful of parts from the K-series chips to the starting Core i3 lineup available online, Intel doesn’t consider a processor family launched until it features in notebooks and extends down to the Celeron brand for the desktop platform. There does seem to be a quicker ramp-up of product being sold locally than in the past, so I don’t think it’ll take too long for all 48 of Intel’s Skylake products to become available over the course of the next few weeks or months.
Once all the PR and product-speak was over, the floor was open to questions. Here’s what I asked.
Q: Given that its your most popular budget processor, will we ever see another Pentium G3258-like product?
Mohammed: “Well, when we think about our previous product rollouts, the G3258 was a celebration of the Pentium’s 20th anniversary, 20 years since people started taking overclocking more seriously. We created the G3258 as a note to that celebration, and its really unique in that regard.”
My translation: Another unlocked Pentium processor? Nope!
Q: Why did Intel start off with high-end chips for Skylake instead of a full Core i5 and i7 rollout?
Mohammed: “When we look at previous launches, we’ve started off with high-end processors as well. The cadence of releases for Skylake is really not very different from Haswell, or Ivy Bridge, and so from our perspective there’s really nothing different for this launch for any other.”
My translation: Well… what he said, I guess. Its probably better to launch as many working, high-clocking, unlocked processors as possible for enthusiasts while binning the rest for slower speeds, which would help with a slow ramp-up of initial releases of working chips made with an advanced process that’s barely out of the taping stages.
Q: Will Intel have Skylake-compatible motherboards for the Thin-ITX and 5×5 form factors?
Mohammed: “We still have a cadence of processors coming out for Skylake and I can’t speak for what other manufacturers are doing for motherboard products supporting Skylake.”
My translation: Intel doesn’t have a 5×5 motherboard compatible with Skylake, and probably doesn’t have anything planned for Thin-ITX platforms either. Or, well, not yet at least.
Q: On a larger scale, how important is South Africa to Intel?
Mohammed: “So, Intel South Africa has been around for quite a few decades. We started out as a two-man team and have now grown into a 25-30-person team, and we continue to grow and evolve. What I can say is that Intel Software, a division that’s been around for decades worldwide, has seen great potential in software and software development in South Africa, and has dedicated resources into South Africa to improve the software coming out locally, which creates an adjacent market that adds value to compute.”
My translation: Yep, we’re big enough for them to care. I mean, hey, there were Core i7-6700K chips ready to purchase within two weeks of the international launch. That’s impressive.
Q: When something launches locally, how quickly does that gain traction for a market such as ours?
Mohammed: “The beautiful part of an emerging market, and an entry-level low-cost market, is that these are typically the first markets to take up product and put it out there. The mature markets, because of their scale, and their size, their adoption of newer technology is not as rapid as a market that carries varying amounts of stock with no guarantees of availability. They’re a very risk-averse market [unlike our one locally].”
My translation: New stuff gains traction quickly here because South Africans are always happy to get their hands on new tech and new stuff. If the long lines for Oculus demos at rAge were anything to go by, we’ll snap up VR technology just as quickly.
Q: What does Intel think the future of computing will look like?
Hitendra: “That’s a very open-ended question…”
Q: Well, say that you have a PC with a single pool of coherent memory, based on flash memory, with the CPU and GPU working in tandem coherently, and everything is much, much faster than today’s hardware. What will that PC look like?
Hitendra: “Okay, well… With those changes, it probably won’t be a traditional computer at all. We might not interface with that computer like we do with desktops and other devices today. It would have to be AI-controlled because all of the information on that computer is available almost instantaneously, at any time, and no human can manage a data set that large. We have been working for a while on cognitive computing and on pre-emption. Microsoft’s Cortana is something that’s similar to what we’re doing and where the tech needs to go, where it’s pre-empting what you might want from it, or context-aware of what you’re doing.”
“Having data that is ‘live’, in a sense, and always available, really needs to have an AI applied to it to pre-empt what you might use that data for. When you give the AI a problem or a question, you don’t just want a single answer, you want the answer plus all of the related answers to questions you might ask, or tasks that you might have wanted it to perform. When such a device comes around, it might still be functionally a compute device. There’ll still be an input-output relationship where you ask it to do something, and it’ll go off and come back with the answer a while later. But if, or when, it starts to ‘think’ for itself, making decisions based on the inputs its been given, then it’ll no longer be a computer.”
My translation: Someone inside Intel is already looking into how they can turn machine ethics into machine code. The T100 will probably be powered by a Core i7-10700K when the robot uprising starts and Skynet is activated.