INTEL IS AHEAD of schedule with Larrabee. Initial boards are going to be shipping to devs in November. Before you get your hopes up, these are not consumer parts, nor were they meant to be. Dev means developer, and chances are the epic MMO that you are whipping up in your mom's basement does not qualify.
The initial boards are meant to code against, and they will do that nicely. They will not be at the final clock, final core counts, or final anything however. If you want to get your feet wet, Intel should have a board for you in a few months.
Don't expect stunning performance though, Larrabee is a new paradigm, and that takes a long time for most devs to wrap their brains around. If you are just going to code DXxx or OpenGL, you don't need one of these.
If you are going to do something interesting, or write engine/middleware code, well you are probably already talking to Intel about Larrabee. µ
Cancer Victims of Lung might state:"I Should Never Have Smoked At ALL". Why is it this article gives same impression.

My views on Larrabee are well known, So geting thru Haze, past paper shredder & Chemical Eraser, & Finally to Product, well sort of Product, is amazing.

Although it sounds like mononucliosis, when Kissing was promised result. It could be anything, at this point. Yet, Intel is NOT Famous for Developement of Its' OWN Inventions.
TS von Drashek
Larabee is one of the most exciting things that has happened in PC graphics for a long time. If the board realy does have a 280watt TDP then it must pack a serious number of transistors especialy if built on a 45nm process. This gives me high hopes that it packs some serious power - no idea how it is going to work but wont be buying a new PC until this comes out - multi GPU larabee on a nehalem x58 board sounds very interesting even if I do need a 1600w PSU.
Is this where EVGA and XFX are supposedly "defecting" to...???

A whole new paradigm... Cool, just like the CELL is to most. A whole year has been spent slating technology just like this ,and yet just around the corner, it's about to be the next paradigm shift. WTF. Programming always used to e about pushing the best new technology to the limits, now it seems to be about falling flat in the face of change and complaining 'cos it's too hard. Get over it.
Hi Scoop,

Thanks for the info.

How many cores/shaders/execution-units on the first product, do ya figure? I assume the programming model or instruction set is x86, but what do you know about the programming model? Is it Nvidia Win/Lin and mind yer 16K cache per block-o-processors, or AMD Win I'm-still-not-sure, or yer Cell PPU & and 8 SPUs and mind yer 256K per PPU?

I don't want to write compilers, I just want to use them, so the per (heh) stream processor cache/local store/what-have-you is of some interest here.

Bob's yer uncle!
256K per SPU I meant to say!
"...and chances are the epic MMO that you are whipping up in your mom's basement does not qualify."

hilarious. this guy just plainly rules!!! i love you! If i was a chick, I'd wanna have hundreds of your babies, Charlie!!!! But then again, I'm not a chick, but I'd still wanna make love to you, just for teh heck of it!!! YaY!!!1
Now it's clear who the sponsor here. TheIntellinquirer, nice. Very nice
Intel's past of 3D hardware (back to the i740 even) I'm not exactly going to expect much. Intel has always talked a big game in terms of 3D this or that. Yet, without an exception, their parts have always fallen far short of the promised performance.

Now if the Intel part being discussed here is the unit I saw mentioned in another rag a few weeks back, then it's supposedly a ray-tracing design. Leaving out the fact that no game supports ray-tracing right now (instantly rendering the card useless unless the card still performs ray-tracing, every engine is translated on the fly or re engineered) and the fact that most developers right now don't think ray-tracing real-time for gaming is seriously possible unless your cheating something serious. I guess if I were DAAMMIT or Nvidia I wouldn't exactly be losing sleep over it.

In the end it will collapse as all Intel graphics initiatives have, with much fanfare followed by deafening silence, and finally a mocking of Intel for producing (yet again) sub-standard graphics hardware for the gaming world.
Its a common misconception that larrabee is for ray-tracing. I'm amazed that people are still saying it. Bet ya there is a quick apology to intel about that in the next issue ;)
i just read on wikipedia that the team that designed the pentium 4 are working on this one. I wonder if it will be slow and run like a 5800 dustbuster?
While true the P4 team designed the "cores" for Larabee (because they are based off the Atom), they went the opposite direction of NetBurst, especially considering the Atom is in-order (the first from Intel since the original Pentium!). It is rumored there will be 16 cores, tied together to a large vector SIMD unit. SO, it still uses the SIMD vector unit(s) just as a GPU does. The CPU's are for scheduling, conditionals, etc. The CPU's will provide greater conditional capability, e.g. programmability. So for graphics, this will hardly matter, and judging by Intel's graphics track record....it could be very slow.
You keep writing Charlie.
If it wasn't for ATI, we'd be paying $600 for a GTX280. Well, anyone with a loose enough wallet to buy one. I sure hope AMD makes it through, otherwise, CPU prices are going to be out the monopoly wazoo.

3 players in the GPU market is good. 3 competitive players in the CPU market would be better.