I think one interesting point here is that we don't know whether the custom chips for eBay and Facebook were designed from scratch. Because if so, that's a lot of work. I am sure Intel was comensated handsomely for it. But I note that Waxman made it clear that buiilding off-the-shelf chips remains Intel's primary focus, obviously. But I would worry that these custom projects could make Intel take it's eye off the ball.
>> But I would worry that these custom projects could make Intel take it's eye off the ball.
Very curious to know what the ball is here. The future of our industry is all about customization and individualism. From medical device to chips, that is where the world is going. There is no reason wht eBay and Shell will need the same type of server. If Intel can offer them unique products, I think that is a good strategy.
I mean like speed grades or what we called binning, taking specific kinds of parts with certain yields or certain killed blocks out of the mix and calling them custom. Anyway, it's all just speculation now. I am waiting for an interview. If you know something, tell us all.
I agree about customization being important. But that has to be more difficult for Intel to do. If the company's main focus remains on the off the shelf products, I am simply saying I hope that all of the customization work doesn't distract Intel from its work on those products. As we've discussed on this forum many times, Intel needs to succeed with Atom and its other off the shelf chips.
we don't know whether the custom chips for eBay and Facebook were designed from scratch.
I very much doubt they were. I'd guess modifications of existing designs. The X86 core will be the same, but peripherals will differ.
Intel got press years back by talking about reusable library code for chip design. I'd guess that on a low level, they are aiming for a modular design that is extensible and customizable. Customers with special requirements can be accomodated by building a chip with a different base components list than the off-the-shelf design.
Of course, simple doesn't mean easy, and no doubt it's a complex and expensive effort to make a one-off design, but it's still an order of magnitude cheaper than a design from scratch.
I think this is taking ASIC and embededded electronics but in this casem the ASIC is a monstrous processor. That customization must be a trade secret for Intel which has no serious competitor in this trade.
Some details about Avoton can be found on page two of this story:
Intel revealed that Avoton, a.k.a. the C2000, will include eight Silvermont-class 64-bit Atom cores. It also integrates four Gbit Ethernet controllers, four serial ATA 2.0 controllers, two SATA 1.0 controllers, four controllers supporting 16-lanes of PCI Express Gen 2, and support for DDR3 memory. A related SoC for storage systems called Rangely will include a crypto accelerator using Intel's Quick Assist technology.
Intel can decide to pack chips inside a processor and then decide to activate the one that will work based on how much a customer pays. With the cost of the transistor minimal, there is no reason they cannot do this.
What are the engineering and design challenges in creating successful IoT devices? These devices are usually small, resource-constrained electronics designed to sense, collect, send, and/or interpret data. Some of the devices need to be smart enough to act upon data in real time, 24/7. Are the design challenges the same as with embedded systems, but with a little developer- and IT-skills added in? What do engineers need to know? Rick Merritt talks with two experts about the tools and best options for designing IoT devices in 2016. Specifically the guests will discuss sensors, security, and lessons from IoT deployments.