AMD’s road map boasts superb cartographer

The path that brought 64-bit x86 computing to the mainstream was cut by a man you may have never heard of. Fred Weber, CTO of AMD, is a visionary in the most practical sense. His vision is AMD’s road map for CPUs, and in laying out that map, Weber has proved an uncanny predictor of customers’ needs.

As a chip architect, Fred Weber led the engineering team for the AMD project code-named Hammer. With a tiny staff and budget, adding up to about 10 percent of the resources that Intel dedicated to creating Itanium, Hammer was the independent film going up against a blockbuster. But Weber and his team built a not-so-secret weapon — it was published in very early specifications — into Hammer. What looked to casual observers like an Athlon plus some 64-bit machine instructions turned out to be an x86-compatible workhorse — x86 remade to deliver a level of compute and throughput performance the volume market didn’t think was possible.

Weber’s career-defining event occurred when Itanium, Intel’s heir apparent to x86 for commercial computing, stumbled in the market because of its lack of compatibility with x86 and  design flaws that made Itanium systems uncommonly hot, noisy, and difficult to code for. In 2003, Weber drove Hammer — then branded Opteron — straight through the hole left open by Intel. In 2005, Opteron made the leap to dual cores, and other members of the AMD64 processor family are following in quick time.

Weber shows how far ahead he was thinking back in 1999 by talking about AMD’s upcoming server virtualization technology, Pacifica, as though it’s been on the market for years. It is clearly the way virtualization should be done. When manufacturers want Pacifica and other Opteron advances, they’ll pluck out their existing CPU and drop in a new one. That hallmark of Weber’s architecture is unique and underappreciated.

Very little rattles Weber, but he shows frustration over Microsoft’s delay in bringing 64-bit Windows to market. Asked if Intel had anything to do with that, Weber shot back an incredulous glare.

Then he said, “There’s no way Intel can ever catch up.” And that, in one line, is Fred Weber.