Core i7 Up Close

Core i7 Up Close

Tick tock? More like ding-dong, mutha—shut your mouth. What baby? We’re talkin’ about Core i7.

Our apologies to Isaac Hayes, but if he were alive, we’re almost certain he would have been tapped to hammer out a theme song for Intel’s most significant CPU launch in, well, ever.

Why is this CPU more significant than the 8088, Pentium, or Pentium M? As the second new chip produced after a series of embarrassing losses to archrival AMD, the Core i7 will answer for the world whether Intel is prepared to ride the momentum of its Core 2 launch with another winning chip or if it’s content to rest on its laurels, as it did with the Pentium 4.

Core i7 also represents a major new direction for Intel, which has stubbornly clung to the ancient front-side-bus architecture and discrete memory controller for years. Indeed, with its triple-channel integrated DDR3 memory controller and chip-to-chip interconnect, the block map of a Core i7 looks more like an Athlon 64 than a Core 2 chip.

Intel actually has three quad-core Core i7 CPUs ready: the top-end 3.2GHz Core i7-965 Extreme Edition, the performance-oriented 2.93GHz Core i7-940, and the midrange 2.66GHz Core i7-920. For the most part, all three are exactly the same except for clock speeds, multiplier locking (only the Extreme is unlocked), and QuickPath Interconnect speed. See the chart on page 42 for details.

The bigger issue is how Core i7 performs. To find out, we ran the Extreme 965 against AMD’s fastest proc as well as Intel’s previous top gun in a gauntlet of benchmarks. Read on for the results.

Intel takes a bold approach to processor architecture, multi-core computing

As a buttoned-down company, Intel rarely likes to make sweeping changes, but its upcoming Core i7 CPU is a major break from the past. Gone is the ancient front-side bus that connects all of the current-gen CPU cores. Instead, cores will communicate via a high-speed crossbar switch, and different CPUs will communicate via a high-speed interconnect.

Also on the outs is the need for an external memory controller. Intel, which has relied on gluing two dual-core chips together under the heat spreader to make its quad-core CPUs, is now placing all four cores on a single die.

Even overclocking, which was once verboten to even talk about within 10 miles of Intel’s HQ, is now automatically supported. Intrigued? You should be. Intel’s Core i7 is the most radical new design the company has taken in decades.

An Inside Job

One of Core i7’s most significant changes is the inclusion of an integrated memory controller. Instead of memory accesses going from the CPU across a relatively slow front-side bus to the motherboard chipset and finally to the RAM, an IMC will eliminate the need for a front-side bus and external memory controller. The result is dramatically lower latency than was found in the Core 2 and Pentium 4 CPUs.

Why can’t the memory controller on the motherboard simply be pushed to higher speeds to match an IMC? Remember, when you’re talking about a memory controller residing directly in the core, the signals have to travel mere millimeters across silicon that’s running at several gigahertz. With an external design, the signals have to travel out of the CPU to a memory controller in the chipset an inch or so away. It’s not just distance, either—the data is traveling across a PCB at far, far slower speeds than it would if it were within the CPU. In essence, it’s like having to go from an interstate to an unpaved, bumpy road.

Of course, if you’re an AMD loyalist, you’re probably bristling at the thought of Intel calling an IMC an innovation. After all, AMD did it first. So doesn’t that make AMD the pioneer? We asked Intel the same question. The company’s response: One: An IMC isn’t an AMD invention and, in fact, Intel had both an IMC and graphics core planned for its never-released Timna CPU years before the Athlon 64. Two: If AMD’s IMC design is so great, why does the Core 2 so thoroughly trash it with an external controller design? In short, Intel’s message to the AMD fanboys is nyah, nyah!

Naturally, you’re probably wondering why Intel thinks it needs an IMC now. Intel says the more efficient, faster execution engine of the Core i7 chip benefits from the internal controller more than previous designs. The new design demands boatloads of bandwidth and low latency to keep it from starving as it waits for data.

Memory a Trois

The Core i7 CPU is designed to be a very wide chip capable of executing instructions with far more parallelism than previous designs. But keeping the chip fed requires tons of bandwidth. To achieve that goal, the top-end Core i7 CPUs will feature an integrated tri-channel DDR3 controller. Just as you had to populate both independent channels in a dual-channel motherboard, you’ll have to run three sticks of memory to give the chip the most bandwidth possible. This does present some problems for board vendors though, as standard consumer mobos have limited real estate.

Most performance boards will feature six memory slots jammed onto the PCB, but some will feature only four. On these four-slot boards, you’ll plug in three sticks of RAM and use the fourth only if you absolutely have to, as populating the last slot will actually reduce the bandwidth of the system. Intel, in fact, recommends the fourth slot only for people who need more RAM than bandwidth. With three 2GB DIMMs, though, most enthusiast systems will feature 6GB of RAM as standard.

Although it may change, Core i7 will support DDR3/1066, with higher unofficial speeds supported through overclocking. Folks hoping to reuse DDR2 RAM with Intel’s budget chips next year can forget about it. Intel has no plans to support DDR2 with a Core i7 chip at this point, and with DDR3 prices getting far friendlier to the wallet, we don’t expect the company to change its mind.

Hyper-Threading Revisited

A CPU core can execute only one instruction thread at a time. Since that thread will touch on only some portions of the CPU, resources that are not used sit idle. To address that, Intel introduced consumers to Hyper-Threading with its 3.06GHz Pentium 4 chip. Hyper-Threading, more commonly called simultaneous multi-threading, partitioned the CPU’s resources so that multiple threads could be executed simultaneously. In essence, a single-core Pentium 4 appeared as two CPUs to the OS. Because it was actually just one core dividing its resources, you didn’t get the same performance boost you would receive from adding a second core, but Hyper-Threading did generally smooth out multitasking, and in applications that were optimized for multi-threading, you would see a modest performance advantage.

The 45nm-based Core i7 will pack all four cores on a single die. The cores will communicate via a high-speed crossbar switch. An integrated memory controller and Quick Path Interconnect links to other CPUs also make the Core i7 very AMD-like.

The problem was that very few applications were coded for Hyper-Threading when it was released and performance could actually be hindered. Hyper-Threading went away with the Core 2 series of CPUs, but Intel has dusted off the concept for the new Core i7 series because the transistor cost is minimal and the performance benefits stand to be far better than what the Pentium 4 could ever achieve.

Intel toyed with the idea of redubbing the feature Hyper-Threading 2 but decided against it, as the essential technology is unchanged. So why should we expect Hyper-Threading to be more successful this go around? Intel says it’s due to Core i7’s huge advantage over the Pentium 4 in bandwidth, parallelism, cache sizes, and performance. Depending on the application, the company says you can expect from 10 to 30 percent more performance with Hyper-Threading enabled. Still, Intel doesn’t force it down your throat because it knows many people still have mixed feelings about the feature. The company recommends that you give it a spin with your apps. If you don’t like it, you can just switch it off in the BIOS. Intel’s pretty confident, however, that you’ll leave it on.

Intel's Core i5 Analyzed and Tested (150+ Benchmarks)

Memory Loss

Enthusiasts will also question the move from tri-channel DDR3 to dual-channel DDR3—why go back if wider is better? Intel’s decision is based on pragmatism and cost. The tri-channel circuitry in the CPU doesn’t add much cost to the processor, but it’s not cheap to implement when building a motherboard. Those added traces from the socket to the RAM slots mean more layers and pricier boards. That’s one of the contributing factors to excessively priced X58 boards this past year.

Should you be concerned about shifting down to dual-channel? Generally, no. For the most part, only the most memory-bandwidth-intensive apps will actually see any performance hit. The fantastic latency and overall bandwidth capacity of the Nehalem design is more than adequate for today’s applications. This doesn’t mean it’s a non-issue. The vast majority of Lynnfield motherboards we’ve seen are opting for four-DIMM slots. That means a maximum of 8GB using affordable 2GB DIMMs (4GB DIMMs are currently cost-prohibitive). The only board we’ve seen with more DIMM slots is Gigabyte’s GA-P55-UD6, which features six, for a total of 12GB, using 2GB DIMMs. The memory controller in Lynnfield tops out at 16GB, while Bloomfield maxes out at 24GB. Realistically, 24GB of RAM is way overkill for 99 percent of us. Our experience has shown us that most apps do not consume that much RAM; 4GB to 6GB is the sweet spot today.

PCI-E at the Core

Another significant change for Lynnfield is that PCI-E comes directly off the CPU core. In X58/Bloomfield, X48/Core 2 Quad, and 790FX/Phenom II, PCI Express is external to the CPU, residing in the north-bridge chipset. Data is pumped out to the chipset where it must then be directed to the PCI-E slots that connect to the GPUs. As you can imagine, the extra hop creates a latency issue—that’s eliminated with Lynnfield. Lower latency is better, right? Yes and no. While latency is better on Lynnfield, the CPUs feature but a single x16 PCI-E 2.0 lane in the chip. That’s fine for a single GPU, but multi-GPU configurations will have to split the bandwidth. That means an SLI or CrossFire X rig will run both cards in x8 mode. Remember, however, that we’re talking x8 at PCI-E 2.0 speeds, which is 8GB/s for each card. From what we’ve seen and from what vendors have told us, only at the highest resolutions with antialiasing do you see any impact.

But, you say, what about tri-SLI? That, too, can be theoretically supported. Even though Lynn-field supports but a single x16 PCI-E 2.0 lane, additional PCI-E lanes are added through the P55 chipset—at the cost of latency. Previous designs that had PCI-E plumbed from both the north bridge and south bridge were dinged for doing just this. Some vendors are working around the lack of bandwidth by adding Nvidia’s nForce 200 chip to their boards. The nForce 200 doesn’t magically add bandwidth but it does manage the available bandwidth across multiple slots. Thus, a board vendor could add three or even four x16 PCI-E slots and have the nForce 200 chip manage the load for multi-GPU configurations. This would eliminate the need to have the GPUs feed off of the higher-latency connection in the chipset.

CrossFire X support is a given in the vast majority of P55 motherboards. And SLI will be as well for the board vendors who pay Nvidia to “certify” that their boards are SLI-ready. Fortunately, the big names are already onboard with that, including Asus, Gigabyte, MSI, EVGA, and even Intel.

If you’re wondering why Intel doesn’t just fix the lack of bandwidth by adding additional PCI-E lanes in the CPU, consider this: Intel would have to add additional traces from the CPU and the socket, and even Intel is loathe to introduce yet another new socket standard so soon. The company also doesn’t think it’s worth it as anyone who truly wants full dual-x16 GPU support should really be buying X58: Remember, folks, Lynnfield and P55 are for the “mainstream.”

The Lynnfield LGA1156 again rejiggers the design of the modern CPU. With the Athlon 64 and the Core i7, the memory controller was moved from the north bridge directly into the CPU core. With Lynnfield, the PCI-Express ports are now handled directly by the CPU as well. Furthermore, relatively low-speed data from the hard drives and USB ports flows though a single DMI link at 2GB/s with Lynnfield.

P55: Downsizing the Chipset

In the old days, new core-logic chipsets were almost as exciting as a new CPU. Not so today. With X58 and Bloomfield, the chipset got fired from its job of managing the memory controller. With Lynnfield, it even loses its responsibilities for managing PCI-E, as both features are now integral to the CPU itself. Today, the chipset is nothing more than a glorified south bridge, managing USB, SATA, PCI, PS/2, and other decidedly unsexy items. The P55 might have been sexier if it had USB 3.0 or SATA 6 functionality, but that won’t appear until next year. And even if it had those features, it’s pretty clear where the P55 stands: Its desk has been moved to the basement and its red Swingline stapler has been confiscated.

Intel’s latest troika of new CPUs brings Nehalem goodness to the masses

Nehalem for everyone! That simple sentence best explains Intel’s brand-new series of CPUs, which is sure to please budget users everywhere while confounding power users.

Why would a new CPU that gives you the best bang for the buck in town be greeted nervously? Because Intel’s new CPU brings with it a new socket as well as a new infrastructure. This new infrastructure is essentially a fork in the road that forces users to make a difficult choice: Save money today but get locked out of the high-end, or splurge today knowing that the budget CPU is damn near as good as the top-end part.

For the details on Intel’s new budget monster, savor our full report, consume the specs, and then digest the benchmarks to see just which path your next PC should take.

Meet Lynnfield

We normally don’t use a CPU’s code-name once its real name is known, but to help keep your head from popping off over Intel’s confusing naming scheme we’re going to rely on some code-names here for clarity.

Intel’s newest CPU family is code-named Lynnfield. The lineup includes the 2.93GHz Core i7-870, the 2.83GHz Core i7-860, and the 2.66GHz Core i5-750. Lynnfield chips use essentially the same microarchitecture as Intel’s original Core i7 CPUs, which were code-named Bloomfield, but the new CPUs are incompatible with existing Core i7 motherboards. That’s right, you could walk into a store and buy a Core i7 CPU that will not work with the Core i7 motherboard you just bought. Likewise, the Core i7 heatsink cooler you bought may not work with a new Core i7, either.

Despite its smaller size, Intel’s new Lynnfield Core i7/Core i5 CPU (right) actually features an increased transistor count of 774 million and a larger die size of 296 square millimeters, compared with the 731-million-tranny Bloomfield Core i7 (left) and its 263 square-millimeter die.

Socket Switcheroo

The most notable difference in this new crop of Core i7s is the socket. For Lynnfield, Intel is introducing the LGA1156 socket. This socket is, as stated, incompatible with the current LGA1366 motherboards and CPUs. To irk you even more, even the heatsink cooler mounting holes are incompatible, so you probably couldn’t use an LGA1366 cooler, even if you happened to have one. And even more annoying to enthusiasts, LGA775 coolers are also incompatible. Earlier this year, Intel execs told Maximum PC that LGA775 cooler compatibility was being considered for the new chip, but obviously the company has since ruled that out. To break it down: LGA1366 uses a 12cm gap, LGA1156 uses an 11cm gap, and LGA775 uses a 10cm gap. Yes, one centimeter difference and you have to dustbin your pricey high-rise cooler even though it’s capable of handling the thermals of the new chip.

We asked Intel if it was doing this just to piss people off and the company said no, it did it for legitimate engineering reasons. Intel actually lowered the height of the new direct socket load mechanism that clamps the CPU in place, which required moving the mounting holes out. Existing heatsinks capable of the thermal load should work, Intel said, so long as consumers obtain updated mounting brackets from the cooler maker. We have to also note that very new high-end coolers are coming with mounts for LGA1156 too. Still, make sure that if the box says Core i7, support for LGA1156 is included.

Although Intel wouldn’t confirm this, we’ve been told by high-end system builders that certain LGA1366 motherboards and coolers would flex enough to create a gap between cooler and CPU. The new design presumably fixes that problem.

Sockets Compared

Lynnfield Socket

Bloomfield Socket

The new budget Core i5/Core i7 CPUs use a new LGA1156 socket design (top image), which is incompatible with existing LGA1366 Core i7 motherboards and coolers (bottom image). The design eliminates traces for the third memory controller, but builds in wires that will accommodate forthcoming CPUs with integrated graphics cores.