Chicago (IL) – Intel’s switch to an integrated memory controller with the upcoming Nehalem core will offer enthusiasts to finally take advantage of fast memory devices – which are constrained by the front side bus (FSB) today. The memory controller bandwidth is catching up with performance memory, which means that we should see dramatically improved memory performance one the new CPU arrives.

The performance memory industry is alive and kicking these days, despite the fact that the advantages of higher clock speeds, which currently reach to 2000 MHz, are fading and are virtually non-existent, at least on Intel platforms. The simple reason is that the memory bandwidth is constrained by the FSB, which limits the bandwidth to 8.53 GB/s in a 1066 MHz FSB (266 MHz QDR), to 10.66 GB/s in a 1333 MHz (333 MHz QDR) version and 12.8 GB/s in those rare 1600 GHz (400 MHz QDR) models.

However, dual-channel memory can offer a bandwidth that exceeds what FSB can take: 12.8 GB/s (DDR2-800) to 32 GB/s (DDR3-2000). Even if you buy those ultra-expensive DDR3-2000 devices you won’t see a dramatic increase in performance, at least if you don’t overclock the FSB at the same time. To support that 32 GB/s bandwidth, you would need a CPU capable of running a 4 GHz FSB (1000 QDR).

Intel Nehalem architecture uses 64-bit memory controllers that are directly connected within the CPU silicon, eliminating those "FSB brakes". Expect you bandwidth utilization jumps from current 50-60% to 90%.

Industry sources now indicated that the mainstream Nehalem processor code-named Lynnfield will be able to almost double the memory bandwidth – to about 18.5-20.1 GB/s when DDR3-1333 modules are used.

If you own or plan to buy DDR3 memory, prepare yourself mentally and financially for Nehalem. Intel is currently preparing two different desktop parts: Bloomfield will become the new Core Extreme and feature a triple-DDR3 controller. Using three or six DDR3-1333 modules you should be able to achieve 30 GB/s, while Lynnfield will arrive in Q1 2009 and offer a regular dual-channel DDR3-1333 controller.

So, keep the modules, and wait for their time. It will come.

holy moly

May 14, 2008 15:23     

guacamole! that is amazing, Intel is going to dominate.
robert report abuse vote down vote up Votes: +5

Intel...

May 14, 2008 16:41     

Already is dominating...
JackCrackerMan report abuse vote down vote up Votes: +7

Ahh...market power in action...

May 14, 2008 17:26     

So if I understand this correctly, Intel is going to make it harder to overclock their products, but easier to overclock other party's products??? And they are going to present it in a manner so that some people will actually thank them for this? Nice.
dan report abuse vote down vote up Votes: -8

Need to leave other's room to wiggle

May 14, 2008 23:07     

Is Intel not going to offer an Extreme Edition Nehalem that can be tweaked? I suspect some board makers and the likes of Voodoo and Alienware will want to have the capability too.
itsallgood report abuse vote down vote up Votes: +0

why Q9450

May 14, 2008 23:11     

sales has stopped? saw several customer complaints on this and it is same 45nm technology.
Oa report abuse vote down vote up Votes: +0

Copying from AMD, again.

May 14, 2008 23:48     

AMD CPUs have the memory controller integrated years ago... Intel also copied AMD's "GHz doesn't matter, only performance matters" idea and introduced the "Core" design. Intel licensed AMD's 64 Bit instruction set (x64). Intel's first Pentium-D was a joke comparing with the AMD's first X2. But I have to say, Intel's current Core2 Duo is fast enough even with the burden of the north bridge. Eliminating the north bridge and put the memory controller inside of the Core2 Duo/Quad would certainly make it even faster. This would put even more pressure on AMD with its current situation.

Doug_d report abuse vote down vote up Votes: -6

Love AMD but

May 15, 2008 00:50     

I've always been a hardcore AMD fan but Intel just has the edge right now. If i was building a system right now, it'd probably be Intel. Infact i'm trying to suck more life out of my AMD system to wait for Nehalem. I'm sure AMD will eventually pull itself out of this rut its in... but they better stop putzin around with crap they don't know or at least do a better job with it and focus back on the processor world or they're going to lose it.

Epsilon report abuse vote down vote up Votes: +3

copying the 386SX

May 15, 2008 01:10     

copying the 386SX, not AMD
nono report abuse vote down vote up Votes: +8

what's next?

May 15, 2008 02:15     

I don't know how many of you think of this. As of now, intel can beat AMD without breaking a sweat on benchmarks... even with their highly controversial FSB bottleneck (that's how AMD called it). And.. what would happen once they remove it? will be be insanely speedy ?? or the other way around? something interesting to observe..
alien report abuse vote down vote up Votes: +6

Nehalem Arrives in Q1 2009, Long Way Off

May 15, 2008 02:35     

Nehalem will arrive in Q1 2009, that's a long way off. That means Nehalem will be affordable in Q2 2009. For those of you who don't like upgrading every year, I guess that means you're going to be keeping your present system for another year. "Lynnfield will arrive in Q1 2009"
Cypherdude report abuse vote down vote up Votes: -7

Look it's business as usual.

May 15, 2008 03:49     

If I hear another crazed Intel or NVidia fanboy break out in a knee jerk fit of zealous touretts then I'm going to start bashing my head against a wall. I swear they must all be hypnotized.
Sq7 report abuse vote down vote up Votes: +1

Stealthy theory

May 15, 2008 05:15     

With DDR2-800 ram, increasing the FSB from a default (say) 266 to 400 will uncap my ram for a 33% increase in ram performance? Well, let's see. As an experiment. I ramped the FSB, but lowered the CPU's multi to maintain the same CPU clock. There was no change in system performance. I'm finding this 'ram liberation' theory a bit stealthy.
Rod report abuse vote down vote up Votes: -3

Correction: Copying from AMD, again. Ohter way around...

May 15, 2008 14:46     

Intel Timna - see http://www.cpu-world.com/CPUs/Timna/ which had an integrated memory controller and intergrated graphics (Fusion truely an AMD idea?) Anyway, Timna should've been the P6 uArch direction that Intel should've followed years ago and returning to the P6 uArch with the Core 2 paid off big time - Funny how some of the Timna features will be only now intgerated on in Nehalem - Can't wait for it...

Nixo report abuse vote down vote up Votes: +6

Expensive & Bandwidth wont matter

May 31, 2008 22:57     

I don't think the extra bandwidth will make much difference (its got 16MB cache). It should help when utilizing multiple cores thought, but it will be hard to use all 6 cores. Also this thing is going to be crazy expensive. lets look at a transistor/price comparison between an e8500 and Nehalem. e8500: 400 million transistors, $200 Nehalem: 1.9 billion transistors. if it scales 1:1 a Nehalem processor should cost $1000.

matthew report abuse vote down vote up Votes: -1