Yesterday, we mentioned that HKEPC.com reported that the initial 65nm CPUs from AMD would be slow.
Today we hear a different story from the Inquirer:
“The voices from the Far East are claiming that by the end of 2006, AMD will have 65 nanometre cores running at 3.0, 3.2 and 3.4 GHz, which are set to be clocked faster than any Intel product at that time.”
What odd phrasing. Normally, when AMD talks about getting a product out the door, they use phrases like “commercial production will begin” or “we’ll start shipping for revenue.” It’s corporatese, but they clearly mean “we’ll start making/selling.”
They’ve never, ever said, “We’ll start running chips at X speed.”
“Running” a chip at a certain speed means nothing. After all, there are people today with a freeze unit that are running Hammers at 3.4GHz.
“Selling” a chip is quite another matter. If you want to keep people away from Conroe, and you have a 3.4 pocket rocket that will be ready at Santa time, you don’t tell people you’ll be “running” a 3.4GHz chip by then. You tell them you’ll be selling one.
It is only if you don’t have a 3.4Ghz pocket rocket that will ready by Santa time that you start using words like “running,” just to give people reason to hold back (and if antsy people don’t realize that “running” and “selling” aren’t quite the same thing, so much the better).
Something else to consider when evaluating the likelihood of the two reports is that the low-speed prognosis seems to come from some official AMD document, while the other is just a few oddly-phrased words most likely coming second-hand from an unknown source.
How To Read This
The problem with assessing items like these is that the situation by definition can change fairly quickly. SOI has always been tricky stuff to play with. You may be stuck in April, and be OK in May, or June or July.
The best read on these two stories is that AMD is being conservative in what they say officially/semioffically, and only promising what they’re pretty sure they can do by a certain point in time.
Reports about chips “running” at higher speeds indicate that some progress is being made, while leaving the sale date open.
The way we look at this, anyone who wants a new computer can either buy a Conroe sooner, or AMD’s answer to it later. We’ve assumed up to now that “later” meant around the middle of 2007, and we see no reason at the moment to change that. Maybe they’ll get a few low-unit FXs or high-end Opterons out before then.
The “what” isn’t really in question, if for no other reason than AMD is going to be in real trouble if they don’t have a 65nm answer to Conroe by the middle of next year. It’s the “when” that’s in question, and whether you wait nine or twelve months after Conroe introduction is really just a minor difference. In either event, you’re going to wait a while.
It’s not like Conroe is this all-conquering beast that AMD cannot possibly answer. Godzilla never completely destroyed Tokyo in those movies, you know. 🙂
If, for whatever reason, you want to wait for AMD, that’s a reasonable course of action, provided you understand you’re going to have to wait a while, probably a year, maybe a little less than that.
What will probably not be a good idea for most is to not wait for AMD’s answer and buy an AM2 system sooner rather than later, hoping to upgrade later. Compatibility is always a potential factor when you’re hoping to upgrade a CPU a year from now, but more importantly, it’s hard to see how AMD isn’t going to be under a lot of pricing pressure the second half of the year and early into next. AM2 systems are likely to be rather cheaper late in the year than next month.