Got this email the other day:
“Good article, but for us non-techies, could you explain how a new, faster AMD CPU is going to really benefit most users? In particular, what are the forecasts for new software (aside from the upcoming Microsoft XP 64 bit OS) ? And, what about new motherboards & chipsets in the works and how will they improve the new AMD CPU performance ? In short, a complete picture would be really helpful. Until I can get someone to clear up the cloudy pictures, I will just wait to upgrade my PC.”
Reasonable questions. Let’s see if we can answer them, and if we can’t, why we can’t.
“Could you explain how a new, faster AMD CPU is going to really benefit most users?”
The simple, shallow answer to this is “DUH, it will work faster.”
The far better answer is another question: Do you do anything that takes any real time on your current computer, or otherwise seems slow?
If anything you do seriously is just taking a split-second for a CPU to do, being able to do it in a split-split-second is hardly a compelling reason who’d rather not spend the money to spend hundreds of dollars for a replacement.
Then again, probably not too many people in that position are visiting this website.
The average reader of this website is far more likely to be someone who wouldn’t mind something faster, whether for objective reasons or not, but doesn’t want to spend much money for little improvement.
Many in this boat start getting interested when something new is twice as fast as something old.
For most of those folks, a new, faster AMD Hammer capable of running 3GHz ought to at least approach that 2X factor in at least raw CPU speed over most late model Athlon XPs.
Combine that with a reasonable price, and we think a very sizable chunk if not majority of people will bite by that point. Some won’t, and the proportion of the “won’ts” will probably be larger than in past generations, but that’s when we think we’ll see the crossover to Hammer.
That’s what we think will happen with upgraders. For those interested in new computers (by that we mean a computer that isn’t replacing an older one), the price/performance ratio of current Hammers now justifies (at least for overclockers not really strapped for cash) the extra cost of such systems over XP systems, simply because of their likely longer useful life.
“In particular, what are the forecasts for new software (aside from the upcoming Microsoft XP 64 bit OS)?”
The simple, shallow answer to this is “Look here.” This is where AMD is compiling lists of those software companies that have committed to x86-64 software.
The far better answer is “We don’t know, because we don’t know whether or not x86-64 is going to become the next general standard for software.”
There’s two critical unknowns hanging over the future of x86-64.
The first is “How hot is Intel going to be for x86-64?” Yes, they’re going to make some x86-64 capable chips, but will they embrace the standard, or just do the minimum necessary to keep those willing to defect to AMD just for x86-64 from doing just that.
Right now, it looks more like the latter. While early adapters of x86-64 in the server market probably have little to worry about; the desktop is another story. If Intel just puts out a token number of x86-64 capable chips for the desktop, don’t expect the run-of-the-mill software firm with run-of-the-mill software to come out with x86-64 software any time soon.
The second unknown is what will happen with 64-bit Windows, even among those with x86-64 capable systems. How well will it work? Will people in general want to buy it? Will people in general want to start replacing their software with 64-bit versions?
Please don’t write me saying essentially, “Well, I do, therefore everyone else will.” That just tells me you’re exceedingly self-centered (and please, really don’t write me if you think x86-64 is going to make Grandma a Linus Torvald groupie). Joe Sixpack will decide this one, and if they don’t buy it, the software won’t come.
Personally, if I had to recommend a new personal system today to anybody, I’d recommend a platform that was at least x86-64 compatible as insurance in case x86-64 does become the standard down the road (btw, that list includes Intel’s 915/925 Express chipsets).
But I’m just putting “safe” over “sorry.”
And, what about new motherboards & chipsets in the works and how will they improve the new AMD CPU’s performance?
The simple, shallow answer to that is “We don’t know.”
The far better answer is there is no far better answer.
Over the next month or so, we’ll see PCI Express motherboards coming out. With one big exception, all these initial boards and cards will offer over current boards is greater useful life, not greater performance.
The one big exception is SLI, which is only going to do you any good if you end up buying two video cards, and even then it may not be worth the cost for many if not most.
After that, sometime, we’ll see AMD shift over to DDR2. This isn’t going to do a whole lot of good, either, just add a light sprinkling of performance percentage points, but when? Who knows? Nobody, including the mobo makers, are talking yet, and all we really know is that AMD planned on doing it eventually a while back, and “eventually” could come when DDR-667 becomes both available and affordable.
It may happen six months from now with the second generation of 90nm processors, but that’s just a guess. There isn’t any factual basis backing it up so far.
What might be of far more concern down the road is whether or not the mobo you buy today or six months from now, AGP or PCI Express, DDR or DDR-2, will be dual-core compatible.
AMD has sort of said that socket 940 boards ought to be, but that leaves the other sockets up in the air.
It may end up being a silly question, since the initial 90nm dual cores are likely to be sitting high up on the pricing food chain, and who knows whether or not a 2006 65nm dual core will be compatible with a mobo you’ll be able to buy anytime soon?
Unfortunately, we ought to have answers to at least some of these questions by now, and it is AMD’s fault that we don’t.
For instance, AMD may not have made up its mind as to exactly when they’re going to shift over to dual DDR, but they ought to know enough by now to be able to say definitively whether a DDR-2 capable Hammer will require a new socket or not, and whether these “San Diegos” and other second-generation 90nms will come in just DDR-2 flavor, DDR or DDR-2 flavors, or DDR and DDR-2 flavors. We don’t care what the answer is, just give us an answer, if not for us, then for all those corporate types you’re trying to impress.
“In short, a complete picture would be really helpful. Until I can get someone to clear up the cloudy pictures, I will just wait to upgrade my PC.”
Do you think this fellow is being unreasonable? I don’t, and I don’t think anybody except a rabit AMDroid would think so, either.
But AMD apparently does. This is why I just laugh when I hear Hector Ruiz talk about AMD becoming “customer-centric.” It’s a joke, a meaningless slogan. You can’t be customer-centric when you leave the customer in the dark all the time, and when it comes to informing the intelligent customer about future intentions so they can plan ahead, AMD usually picks the Dark Side.
Maybe that works for kids and CompUSA buyers, but if you want serious grown-ups, especially corporate beancounters, to spend serious money, giving them the finger as to your future intentions is not too bright. Especially when the competition has been doing just that for a long time.
If you want Intel’s corporate business, you’d better conduct business like Intel and give customers what they need to know to plan, and “Nyaah, nyaah, we’re not going to tell you” isn’t the way to do it.
Or does AMD just want dumb customers?
Perhaps those responsible for corporate accounts are a bit more forthcoming with details, but aren’t individuals customers, too? Aren’t they your main customers? So why treat them like they were the enemy?
After all, it’s not people need new computers like they need air. If you don’t want to talk, they can do what this guy and many others are doing. They’ll wait you out, and maybe wait longer than they would had you been forthcoming in the first place.
Not smart. Not smart at all.