There’s no doubt that Intel has been wildly successful ever since they launched their line of Core processors. Previous to that, the desktop processor crown belonged to AMD for a good few years and their Athlon line-up, but unfortunately they didn’t manage to keep up with the rising costs of fabrication. The faux-pas with AMD’s quad core processors and subsequent rebranding into the tri-core Phenom shows just how challenging the fabrication and manufacturing process is.
But while AMD recuperates and sells off its fabrication business, Intel continues to dominate with their intellectual property by introducing the new evolution of Core. As you may have noticed, Intel launches a new architecture in one year, and in the next enhances it via a die shrink. This is common practice as shrinking the die allows engineers to optimize the existing architecture for maximum power savings and reduce heat output; a more refined model of the previous so to speak.
We’ve seen Intel do this with the Core and Core 2, so now, in just a few months time, the world will again bear witness to Intel’s new architecture, a different beast from the existing Core 2: the Core i7. And it will be big, with the main feature being the much improved multi-threaded performance scaling.
The i7 brings to the table more of everything; bandwidth, threads, efficiency, all the things that Intel expects to contribute to an explosion of parallel programming. If you thought having two cores was great, and four is WAY cool, expect to see up to eight and sixteen cored with the i7!
One of the problems with existing Core 2 Quad processors is the way Intel approaches it… by clobbering together two dual core CPUs and calling it a Quad. Contrast this to AMD’s monolithic design, where all four cores share the same die-space. This is akin to Intel having four departments in two different buildings while AMD brings all four departments into the same building; obviously we know which one will be more efficient. This will be the first mistake the the i7 corrects.
Another improvement is again an idea that AMD came up with first: on-board memory controllers. AMD have been using this approach ever since the first Athlon launched way-back-when in 2003. And if that wasn’t enough, Intel will also have what they are calling QPI – Quick Path Interconnect. This is a serial interface that connects different components of the motherboard together in a single high-speed high-bandwidth channel, very similar to AMD’s own HyperTransport, which was again launched way-back-when. I guess Intel’s finally admitting that their rival’s approach to computing is better. But of course, Intel won’t just be copying blindly, as each feature will be even better, faster, more efficient; after all, one would find it hard pressed to be impressed with just a carbon copy.
Interestingly enough, the i7 will also come with what is being marketed as Fusion, with the processor having an integrated graphics core in the same die, providing on board graphics without actually having a separate graphics card.
Finally, we have something called Turbo Mode, where when the CPU detects that only a few cores are being actively used, they will divert all power and increase the clockspeed of these cores, giving you the performance of a single, very fast core, with the flexibility of having multi-core processors when the need arises. The only reason why you won’t be able to run all four cores at their maximum theoretical speed is because that would cause too much heat and fry your computer. It’s an interesting feature that sounds a little bit like smoke and mirrors, but we shall see what the hardware enthusiasts think of it when the i7 launches.
Finally, the new processor will be running on a different socket, which means you will need to shell out more money for a new motherboard, and possibly new RAM too. It will be an expensive experience if you’re an early adopter, but the i7 should be an interesting beast to see, to say the least. 🙂