Timing is everything. With the world in an economic crisis, unemployment levels soaring in New Zealand, and the wallet-lightening Christmas period just around the corner, those handsome devils at Intel have picked a fine time to drop their latest saliva-inducing CPU range at our doorstep.

But before you go re-mortgaging your home (assuming your home is still worth more than your mortgage) and head down to PlayTech to get your next hardware fix, allow me to show you what these new products have to offer and see if the Intel fanboys' recent chant of "wait for Nehalem" rings true.

So what's the big deal? Well unless you've been living under a rock for the last couple of years you'll know that Intel's "Core 2" series of processors have been the dominant range of CPU's in the marketplace.

When these chips were released they went head-to-head with, and then subsequently annihilated, AMD's "Athlon X2" range of CPU's. AMD have been on the back foot ever since, and even their much-hyped "Phenom" range of chips launched late last year fell embarrassingly short of reaching the raw performance levels of the mighty Core 2 architecture. Since then the Intel chips have been getting cheaper, faster, and more efficient. AMD do have a new line up of chips named "Deneb" scheduled for a Christmas launch but in true kick-em'-while-they're-down fashion Intel have decided that now is the time to unleash the next iteration of the Core architecture: the Core i7.

The initial range of Core i7 CPU's consist of 3 chips - the 2.66GHz 920, the 2.93GHz 940, and the 3.2GHz 965 Extreme Edition. All 3 chips feature four cores, HyperThreading, 256KB of ultra-fast L2 cache per core and a large 8MB L3 cache shared across all cores. All these chips are designed for the latest motherboard socket called LGA1366. The '1366' corresponds to the amount of pins in the socket which connect to the CPU and is almost twice as much as the Core 2's LGA775 socket used, mainly due to Intel integrating the memory controller onto the CPU itself.

Speaking of memory, these CPU's only support the new range of DDR3 memory, and also have the ability to run in "Tri-channel" mode so now you can have 3 sticks of high-bandwidth RAM inflating your credit card bill instead of just 2 (and we'll find out if this offers any performance improvement later on). All the current chips are built using a 45nm manufacturing process, and this will move to 32nm sometime next year following Intel's 'tick tock' strategy. I won't go into detail on what other architectural improvements Intel have done on their chips but it is suffice to say that many improvements have been made and clock-for-clock we should see the Core i7 chips outpacing their Core 2 predecessors.

As for the motherboards, the only chipset currently available to support Core i7 is X58, also from Intel, which happens to be able to support both Crossfire and SLI graphics solutions which is pretty damn cool. LGA1366/X58 systems are markedly different to anything we've seen from Intel thus far (but strikingly similar to recent AMD systems funnily enough). Instead of a North Bridge (which is now built into the CPU itself) connecting the RAM, CPU, graphics subsystem, and other devices via a Front Side Bus, we now have an Input/Output Hub (IOH) connecting the CPU and RAM to the rest of system via a Quick Path Interconnect (QPI).

Remember, I mentioned that the memory controller is now integrated into the CPU itself, so CPU-RAM data no longer needs to go through the North Bridge/IOH. The current operating speed of the QPI is 25.6GB/s which is twice what Core 2 systems could do at 1600MHz Front Side Bus. Three channels of DDR3-1066 memory equates to roughly 25.5GB/s maximum theoretical bandwidth so this leap forward in technology was required for Intel to move forward in performance as far as system bandwidth is concerned.

Overclocking the Core i7 is, well, same same but different. You still take the main bus frequency (The QPI speed, currently 133MHz), and multiply it by the CPU multiplier (20x for the Core i7 920) to get your final CPU speed (2.66GHz in this case). The current CPU's only officially support DDR3-1066 memory speeds, so you'll have to overclock the CPU to get up to the higher memory speeds. More on how far I got with overclocking later. Right now I'm itchy for some figures, so let's jump right into the benchmarks.

Test System

.
Core 2 CPUs: E8200 (2.66Ghz Dual Core), Q9450 (2.66GHz Quad Core), QX9770 (3.2GHz Quad Core)
Core 2 Motherboard: Asus P5Q Pro
Core 2 RAM: 2x2GB G.skill PK Series @ DDR2-800 CL4
Core i7 CPUs: 920 (2.66GHz Quad Core), 965 (3.2GHz Quad Core)
Core i7 Motherboard: Asus P6T Deluxe
Core i7 RAM: 3x2GB G.Skill NQ Series @ DDR3-1066 CL7
GPU: Sapphire 512MB HD 4850, Catalyst 8.10 Drivers
PSU: Silverstone Olympia OP700
O/S: Windows Vista Home Premium 64-bit SP1

3D Rendering

This is where a lot of people are willing to spend a lot of cash to get high performance, so we'll start here. The difference between waiting a few hours for a scene to render as opposed to overnight or even a couple of days is certainly a tangible benefit for some people, so I can fully appreciate the investment. Remember, the scores you're seeing here is the time it takes for just one frame to render - now multiply that by 30 or 60 frames per second and you can see how long it can take to render even a short 3D scene.

A good start for the Core i7 chips - the Core 2 Quad cPUs are up to 33% slower in these rendering tests. The poor old dual core E8200 is left well behind in the multi-core Cinebench department. Only two cores? Pfffft, go back to 2006!

Video Encoding

This is where you take video from one format - from a digital handycam or DVD for instance - and compress it into a smaller format suitable for PC playback without losing too much quality. x264 is a High Definition format with DVD-like quality (I'm using version 0.58.747 if you're wondering, encoding 1442 frames at roughly 3900kb/s bitrate), whilst Windows Media Encoder converts the video into WMV format which can lose a lot more image quality but has better playback compatibility. Encoding is done in two passes - the first pass is a blind conversion which simply converts the video from one format to another, whereas during the second pass the video data is fine-tuned, capturing a lot more quality where it's needed and compressing more where it won't be noticed, thus it takes a lot longer. The x264 result is recorded by how many frames per second the CPU can encode, while the Windows Media result is recorded by how many seconds it took to encode both passes.

Again the Core i7 lays the smack down on the older chips with the x264 result, with an almost 50% improvement at the same clock speed on that all-important second pass. On the WME test, however, it's actually a hair slower than the Core 2 equivalent. Once again the E8200 fails miserably.