Network Computing readers were the first to learn about Intel's efforts to pack a data center onto a single chip, via my recent interview with Intel chief technology officer Justin Rattner. (See Q&A: Intel CTO Justin Rattner.) Now, the chip behemoth has taken things one step further, formally announcing its single-chip cloud research project.
The official squib from Intel (here) is that: "Researchers from Intel Labs demonstrated an experimental, 48-core processor. . .Intel Labs has nicknamed this test chip a 'single-chip cloud computer' because it resembles the organization of datacenters used to create a 'cloud' of computing resources over the Internet."
Now, this is fine news, but it's nothing to get all aTwitter about, in terms of thinking that we're going to see single-chip cloud machines all over the place tomorrow. Remember, this is a research project. Also, one should note that this is simply the latest step in Intel's ongoing Tera-scale effort, wherein the company previously demonstrated an 80-core prototype processor.
You could say that Intel has enough neat stuff in its labs that it can simply roll out something anytime it wants and say, "Hey, here's something nifty, and it means such-and-such is coming." (In this case, you'd replace "such-and-such" with "cloud on a chip.")
Which is not to say there's not some reality behind Intel's astute publicity engine. One can assess the feasibility of cloud on a chip via the realization that it's on a continuum with what we've got today -- namely, cloud on highly virtualized servers. (The point here is that highly virtualized server present thousands of physical and logical cores within a single container, so you could say that putting something analogous onto one chip is simply a packaging issue.)
As well, Intel's "single chip cloud computer" can be seen as an intellectual exercise, which rolls up a lot of ongoing research in the broader bucket of data centers on a chip. (Cloud is the sexier word du jour, so it makes for a better press release.)
Here's some salient perspective on that data center on a chip research, from my aforementioned recent interview with Rattner:
Rattner: We just won the best-paper prize in the Symposium on Operating System Principles with our collaborators at Carnegie Mellon on something called FAWN, which stands for "fast arrays of wimpy nodes." It's the idea that, if we could build tomorrow's processors out of arrays of relatively simple cores, we could deliver data-center-class solutions. It would be data centers on chips, and then arrays of those chips.
Wolfe: Does this set up a possible race between virtual and physical processors, because with the chips you're talking about, you'll have so many physical cores you won't need virtual instances?
Rattner: If an individual core is so inexpensive, why go to all the trouble to virtualize it? Just allocate some number of physical cores to the problem. What we're also trying to understand is, what leads to the most energy-efficient solution? Am I more energy efficient if I take a big core and virtualize it many ways than I would be if I took lots of simple cores and handed them out as the workloads demanded? I can't tell you what the answer is, but things are looking pretty good for the small cores.
There's more here.
My final thought on Intel's single-chip cloud computer is that the announcement Intel made is less about letting you know you're going to be getting cloud chips real soon and more about generating buzz for upcoming product. We see this in the press release line: "Intel will integrate key features in a new line of Core-branded chips early next year and introduce six- and eight-core processors later in 2010."
I should note that I've neglected to mention AMD, which is a co-equal player in graphics via its acquisition of ATI, which was folded into the company several years ago.
Read more about the cloud chip from the horse's mouth, here.
Follow me on Twitter. What's your take? Let me know, by leaving a comment below or e-mailing me directly at [email protected].
