AMD has squandered a golden opportunity

ADVANCED MICRO DEVICES (AMD) continues to emphasise the advantages of its native quad core design, which has four processing cores on one piece of silicon. This chip is supposed to ship to customers this month.

But Intel announced last month that it had already shipped one million quad core devices, which began shipping to customers last November. The chip giant's design uses two dual-core dies packaged together.

Worryingly, AMD has said that its quad core parts will launch at frequencies up to 2GHz. But just recently, the company quietly added four dual-core Opterons, two of which operate at 3.2GHz.

Hindsight is always twenty-twenty. But knowing what we know today, AMD should have launched a dual-die quad-core months ago.

This design should have been launched last December, when the company moved to 65 nanometre manufacturing technology. That would not have been the perfect quad core, but it would have entered AMD into the quad-core race. As things stand at the moment, the company won't have a full quad-core line-up until close to year-end.

Considering that most of the AMD core technologies were in place to make a dual-die quad-core possible, would it have been that hard and cost that much to bring to market? I don't think so.

If that design had been launched, it could have averted much of the negative press that the company has had to endure. But more importantly, it could have helped to reduce its previous two quarter losses by bolstering its average selling prices (ASPs) and keeping its customers onboard.

PC World reported last month that Falcon Northwest doesn't even stock AMD chips, but the company will special order them. The publication also said that prior to the Core2 launch, the company's performance minded customers had bought AMD almost exclusively. An AMD dual-die quad-core chip may have made a difference.

If a dual-die quad-core design had been released, two socket eight core servers would have been launched. As that didn't happen, this space has been left to Intel to gobble up. In the four socket market segment, a 16 core design would have propelled AMD's performance advantage further.

One lesson that AMD should have learned from this is the importance of keeping momentum. Because the company failed to do that, the chip giant has now regained important lost ground. A dual-die quad-core could have given the company breathing space until its native quad-core design was market ready.

From a purist's standpoint, a monolithic quad-core design is the way to go. But AMD's pursuit of that has proved costly, as the company continues to lose money. So when Barcelona starts shipping for revenue, it's going to have to start bringing home the bacon. µ

See Also
AMD takes big hit on Opteron sales
Channel grumbles over further AMD Barcelona delays