AMD takes SledgeHammer to beat Intel's Merced

But curiously, the strategy's a lot subtler than the codename...

AMD is today poised to build on the buzz surrounding the Athlon by announcing its 64-bit successor, SledgeHammer. It's scheduled to ship in 2001, after Intel's H2 Merced/Itanium target, but it appears AMD has been reading Intel's roadmaps diligently, and has an ambush planned. For over a year now Intel has been talking down prospects for Merced/Itanium (we'll decide what to call the renamed chip once the market decides). The company says it expects Merced (screw it...) to be important as a development platform for 64-bit systems, but that it's going to be a while before there's sufficient 64-bit software around for customers to derive major benefits from it. With the OS vendors all not going much further than promising shipping software by the Merced launch, that's a pretty reasonable take. Some developers will be running with beta 64-bit OS software, but development won't gather momentum until a fair while after Merced ships. Intel therefore points people gently in the direction of Mckinley (or "Ickel," as it'll no doubt be known) as the first high volume Intel 64-bit processor, and in the meanwhile the company will continue to develop and sell higher speed, high volume IA-32 (there's something you can pronounce with confidence) chips. But if Intel expects 32-bit to remain king through 2001 and maybe beyond, what should AMD do? It's currently collecting 32-bit speed trophies for Athlon, and if it can build on this up to the Sledgehammer rollout then it might just be in a commanding position. SledgeHammer, marketing VP Dana Krelle tells us in a blatant leak to the WSJ this morning, will be 64-bit, but will be designed to run 32-bit apps at blisteringly high speeds. Alongside that AMD's x86-64 architecture is intended to be simpler to deal with than Intel's. So here's the script, as we see it. AMD sees a possible window whereby it can carry and increase current Athlon momentum through 2000, and then capitalise on that with what will in operation effectively be a faster 32-bit chip in 2001. The possible gotchas here are first, that the Athlon mo gets blunted (it is a long way to run, and AMD doesn't usually manage to stay ahead for long), and second, that any delays in SledgeHammer would give Intel the opportunity to establish its own roadmap instead. If AMD avoids these, and manages a seamless Athlon-SledgeHammer transition, then the mindshare it can capture with the latter will help it with the big one - use SledgeHammer market momentum to establish the AMD 64-bit architecture while leaving Intel floundering. That's superficially plausible as a strategy (this is not the same as saying it's achievable), but we can see another gotcha. Whether or not AMD 64 is easier to develop for than Intel 64, the Intel architecture currently has scads of software development resource going into it. Solaris, Monterey, Linux, Windows 200x (no, really), and by 2001 at least three of these will be shipping. So what's AMD doing about getting development going for Sledgehammer? ®

Sponsored: Designing and building an open ITOA architecture