Fred Weber, Chief Technology Officer for AMD Computational Products Group, confirmed at Microprocessor Forum 2003 today earlier reports about AMD K9 processor’s show up in 2005, a little bit more than three years after the first K8 – AMD code-named SledgeHammer and ClawHammer – chips saw the light of the day in early 2002. No definite technical details were said at this time, moreover, there is a feeling that AMD has not set the final targets for the K9 yet.

It was indicated that the first samples of AMD K9 will be released in the second half of 2005, 3.5 years after the AMD showcased its early K8 processors. Historically it took about three to four years to develop a micro-architecture of a desktop microprocessor, but the first indications of the K9 product development start belong to March 2003, right after the Opteron went into mass-production. Today’s claim of Mr. Weber points toward a tight schedule for AMD K9 development process.

AMD is looking at adding multi-threading pattern into its future microprocessors in order to improve overall performance, it transpired, but Fred Weber declined to confirm if any real decision has been made on the matter. He added that multi-core design would give a more substantial speed bump compared to that brought by technologies similar with Intel’s Hyper-Threading.

We do know that Sun, IBM and Intel are planning to add or develop multi-threading technologies into their CPU designs because of cost-efficiency and remarkable performance gain provided by those techniques. Therefore, since AMD is not an exception from the economical point of view, I would expect the company to add the feature into its next-generation K9 conception as well.

It is interesting to note that during AMD Athlon 64 launch in late September AMD confirmed that the first dual-core AMD Opteron processors featuring AMD64 aka x86-64 technology would be manufactured using 90nm SOI technology in 2005 (see this and this news-stories). In the same year, AMD starts transition to 65nm fabrication process, it was indicated. 

The AMD K9 chip is surely targeted on server, workstation and desktop applications at the same time. Thus, in case the dual-core AMD K8 chip will make it into the market before the K9, the latter will have to sport either a very efficient multi-threading pattern, or two cores per one chip, so to outperform its predecessor in all kinds of applications. Because this is only a guess, take it with a grain of salt.

All-in-all, we still know nothing about AMD K9, but, at least, we can ascertain AMD’s commitment to rapid development of its CPU architectures.