Intel pulls back from FB-DIMMs

A FEW DAYS ago, Eva told you about microbuffers and how they bring many of the benefits of FBDs without the cost or heat. What they are not telling you is that FB-DIMMs are basically dead on Intel platforms come 2008.

Intel is telling everyone who will listen that they are going back to rDIMMs in 2008 over about 90% of their server line. This is due to them not being able to hit power numbers, seemingly a problem with the tech itself vs implementations, and of course cost. Looking out at the numbers, 4S servers are about 10% of the server market, 2S make up most of the rest, so this would seem to indicate that FBDs will be relegated to high socket counts.

This move actually makes complete sense, the benefits of FBDs, high capacity and high bandwidth are a good match for that market. Likewise, the down sides, high cost, power and latency are less of a problem there. I can also see them putting it in niche 2S servers that need lots of DIMMs but go easier on the CPUs.

Whoever pushed the whole concept of making Nehalem totally modular, pat yourself on the back, you probably saved the project. When that chip does roll out, look for FBDs to be more or less off the table.

The only worrying thing is if FBDs can survive with 10% of the market. Will that be enough for continued R&D on the AMBs, or will they just putter along until no one cares, then fade away? In any case, the clock is ticking. µ