Core i7-3960X features explained

The 2nd-generation Intel Core i7 Extreme Edition processor is described as “super-smart, ultra-threaded, and absolutely uncompromising” by its marketing mumbo-jumbo. So what does that really mean? Well, it supports Turbo Boost for dynamically increasing the CPU clock speed when needed, Hyper-Threading to process two tasks per each of its six cores, and it is the first to support quad-channel RAM configurations. It has completely unlocked multipliers for overclocking flexibility, the largest amount of on-board cache memory of any Intel processor, and the companion X79 chipset supports 40 lanes of PCI-express bandwidth for true 2x16 SLI and CrossFire setups. Finally, it’s debuting alongside Intel’s brand new liquid-cooling solution, consisting of a waterblock, reservoir/pump, and heatsink/fan in closed, pre-assembled system that is simple to install.

If any of those things interest you, we’ll be going into more detail on them right now -- otherwise, you might just skip ahead to the benchmarks and conclusion.

Socket LGA 2011 & X79 Express Chipset

As mentioned on the previous page, these new high-end Core i7 processors (AKA Sandy Bridge-E) are using Intel’s latest LGA 2011 specification. Aside from the obvious coincidence with the year it’s being released, it also literally means there are 2,011 pins on the motherboard that come in contact with the bottom of the CPU. If you don’t believe me, you can count it in the picture on the previous page!

This new socket, along with the accompanying X79 Express chipset, carries with it support for quad-channel DDR3 RAM and up to 40 PCI-Express lanes. This provides for a massive boost in available bandwidth on both fronts, which is good news for everybody, whether you’re concerned with your workstation or gaming rig. Granted, most users’ setups don’t even approach the bandwidth limits of existing platforms, but enthusiasts and professionals alike will certainly enjoy the prospect of 16x16x8 Tri-SLI or combined 51.2 GB/s of memory throughput.

Six processor cores on 32-nm process

Like the first generation of high-end Core i7 processors (code-named Gulftown), the i7-3960X and the i7-3930K both have six physical cores on one chip with a TDP rating of 130W. That’s the same amount of power draw (and thus heat production) as the quad-core Bloomfield chips, but the difference lies in the switch to a 32-nm process from 45-nm. That’s the beauty of progression in fabrication technology -- you can get more performance for the same costs because everything becomes smaller and [comparatively] cheaper to produce. We’ll be waiting a bit longer on the switch to 22-nm, though, which will be coming with the Ivy Bridge CPUs early next year that will replace the mainstream Sandy Bridge line (and eventually this SNB-E line with IVB-E!).

Turbo boost

In not so many words, Turbo Boost is used to dynamically increase the clock speed of the Core i7 processor whenever the workload demands it. It’s basically the opposite of Intel’s Speed-Step tech, which will reduce power and lower the clock speed of the CPU whenever it’s not being used, to save energy and cut down on heat. They’re two sides of the same coin, constantly adjusting the Core i7 processor based on your computing needs at any given time. You can disable them, of course, if you’d prefer more consistent performance at your fingertips.

Hyper-threading

This bullet point has been around since the Pentium 4 era, when dual-core processors didn’t exist and Intel decided to create a second “virtual” core on their CPUs to promote multi-tasking. Hyper-threading basically allows the CPU to handle twice as many application threads as it has processing cores. So, for example, the i7-3960X can handle 12 threads with its 6 physical cores, the i7-2700K 8 threads with its 4 cores, etc.

It sounds like a great thing in theory, yet unless you’re running half a dozen processor-intensive applications while you play games, it’s not going to be of much use. Unfortunately, there still aren’t very many individual games and applications that make use of 4 cores simultaneously, let alone 6, so being able to run 8 or 12 threads at once remains more of a curiosity or point of boasting than anything. Hopefully, with most new CPUs these days being quad- or hexa-core ones, we should be seeing more and more programs use them as time goes on. For gaming, I’ll bet we see a huge spike in multi-thread usage once the next generation of consoles comes out, provided they continue to try to push the envelope of hardware efficiency.

New liquid cooling solution

Starting with the Sandy Bridge-E family of high-end Core i7 processors, Intel is no longer bundling a heatsink/fan combo into the retail package. Perhaps they finally realized that anybody spending anywhere from several hundred to a thousand dollars on a CPU is just going to opt for a higher-quality aftermarket cooler anyway, so they decided not to bother trying to force a stock one on them. They’ve partnered with Asetek to create a high-quality liquid cooling solution that is compatible with all Sandy Bridge and upcoming Ivy Bridge processors on any of the LGA 2011/1155 or 1366/1156 sockets. It’s a fully-enclosed package with copper waterblock, pump, radiator, and propylene glycol reservoir and comes with a 120mm fan, as pictured above.

Intel estimates that one in particular will cost between $85-100 at retail, but if that’s too much for you, or you’re not interested in overclocking, there will also be a barebones Intel-brand heatsink/fan combo for LGA 2011 that will cost less than $20. Of course, there will also be plenty of other choices across the spectrum from cheap heatsink/fans to more expensive all-in-one liquid-cooling solutions from third parties such as Corsair, Antec, Cool-It, and Cooler Master, among others.