Got this note overnight. Some people seem puzzled as to why I keep mentioning PCI Express, so maybe I’ll should explain in a bit more detail.
I think your analysis from a hardware standpoint is pretty good. A lot of us are waiting on the PCI express bandwagon for fear of obsolescence.
But I think there may be some flaws in your analysis, whether they be large or small is a consideration I’ll leave for you.
A lot of us have been waiting . . . for the ‘right time’ to make the video card plunge. To hopefully get in when a good card can be had for a price that does not break our meager banks. I’m speaking to anyone in our audience using something earlier than current-generation parts. Hell I’m 30 and a professional and I still can’t go tooling around for a top-end video card.
So here I sit, with an aging Socket A system using an Athlon XP 2000+ and a 64 MB GF3, probably not atypical of the gaming enthusiast-on-a-budget trying to get the most life out of what they’ve got. Return to Castle Wolfenstein: Enemy Territory plays just fine, thank you very much, and if that was the only game out between now and PCI Express, most of us would be happy like clams.
But it’s not, is it?
So advice seems to be to wait for PCI Express if at all possible, but I think you’re going to find for a lot of us that it’s just not going to be possible for three reasons:
1. When I switch to a PCI Express motherboard, it seems likely that there are going to be additional expenses above and beyond the board. Will the first PCI Express boards take my Socket A CPU? Would I even want one that did at that point, or will I also be upgrading my CPU? My RAM? Usually a big motherboard switch means more, sometimes a lot more, expense than just the board. So now the recommendation is not just to wait, but that at the same time I’m going to upgrade my video card? Maybe I’ll wait longer after that, to space things out. BUt that means until at least mid-next-year, or longer, I’ll still be running my GF3. How many of us will find that acceptable?
2. The first generation of any part, generally, sucks. That should be the second or third universal rule of Overclockers.com. I don’t know what the other ones are, a sort of Ten Commandments of Overclockers.com, but that one should definitely be in the top three. I guarantee that when the first generation of PCI Express video cards come out, you (and by that I mean personally, you, Ed) will write an article about how we should all hold on to our last-generation AGP cards because PCI Express just isn’t ready yet and the hardware and drivers are all plagued with bugs, not the least of which is that ATI and NVidia will probably spring new chips on us as well for the occasion so we’ll have crappy drivers, crappy hardware, and PCI Express that won’t be ‘ready’ yet for prime time.
3. The games we will be playing this Christmas will need more than what most of us have to look the way the developers intended. Believe it or not, there will be games out this Christmas that will require more processing power, more RAM, or more features in hardware than what most of us are carrying in order to play the way they should, or the way we’ll want. And by ‘want’ I don’t mean ‘wish’, but rather a practical benchmark by which we decide whether it’s time to look for a little more power somewhere.
So what are we supposed to do? My advice to gamers is the following: Save your birthday money, save your holiday cash gifts, save up your allowance and when the next big game you really have to have comes out this fall or winter–and for almost all of us I think there is going to be one by Christmas–and buy the best of what you can afford, and figure that’s the card you’re going to use when PCI Express first comes out. For better or for worse, I don’t think there is going to be more waiting for any of us caught behind the current generation of cards.
The reason why I’ve been stressing the emergence of PCI Express recently is not because PCI Express is going to be stunningly wonderful in and of itself. It’s because it is part of a new generation of motherboards.
The motherboards that will be released next year will make a jump we haven’t seen since the jump from Pentium-style to Pentium II style mobos back in 1998. We’re going from an AGP, DDR, PATA, PCI motherboard system to a PCI Express, DDR2, SATA environment.
This not only makes most current equipment fairly useless in future generations; it also reduces the length of time competitive replacements for the older generation motherboards will be available.
The reality is the next serious upgrade after mid-2004 is going to be a whopper for everyone, and it’s unavoidable. (There may be some transitional mobos initially, but they won’t last long.) New video card, new RAM, most likely by then SATA drives, besides the CPU and mobo.
If you want to take your sweet time about taking that next jump, I’m actually rather inclined to agree with you. Next June, I probably will be saying precisely what this gentlemen predicts I’ll be saying about PCI Express. 🙂
This is why I’ve been recommending buying now. The machines available a year from now won’t be considerably faster than what you can affordably get and easily overclock today. It will probably take two years for that to happen. Buy now, and get some good use out of the machine before you end up replacing most of it next go-round.
The only exception to the “current stuff is capable and quite affordable” is the video card area.
This gentleman is essentially saying, “I have a GF3; I can’t wait that long.” That’s quite understandable, but understand that any video card you buy will only be as good for as long as you own that platform. New platform, obsoleted video card.
Past surveys have indicated that people like to hang on to video cards for a few years, moving it to new CPU/mobos. A lot of people like to buy computer upgrades one piece at a time.
This really isn’t going to be an option. If you’re thinking, “Well, I’ll buy a real hot, expensive video card today, and move it into my next computer system, you are likely in for a very unpleasant surprise when you do.
Will there be transitional AGP/PCI Express mobos in 2004? Probably, but dual-capable systems in the past have tended not to work very well. Will video card makers make both AGP and PCI Express version of the latest hot card? Maybe for a while, but again, history says, “Not for very long.”
This is you need to look ahead and plan a little before you buy. Otherwise, you get blindsided, and end up spending more of your hard-earned money than you ought to because you didn’t look before you leaped.
Why You Should Look Before You Leap
When I say such things, I usually get a few letters from people who get personally offended and say in essence, “How dare you tell me the way I buy equipment isn’t good?”
I’m reminded of one forum post I saw a while back. The gist of the message was “I bought a socket 423 system. Then socket 478 came out just after I bought it. Those #$#%%^@#, I’ll never buy from them again.”
This is what happens when you don’t look before you leap. That’s why I say such things.
If the budget-conscious decide, “OK, I need a new video card, and since I want to get at least two years out of it, I won’t change the platform for another two years,” that’s good. If someone else says, “I really need a new video card now, but I also need a new system in a year. I’d better spend less and buy a video card that will be good enough for the next year, then I’ll buy a new system and replace it all,” that’s good, too.
It’s certainly a lot better than saying today or tomorrow “OMG, I need a hot video card.” then a year later, “OMG, I have to have a new platform . . . . WTF, there’s no slot for my video card!!! Those #$#%%^@#!!!!”
No, they’re not #$#%%^@#, you’re a dumb . . . well, if your brain charges a toll of several hundred dollars before it lets something new in, maybe you ought to consider turning it into a freeway. 🙂
I mean really, if experience is the best teacher, why don’t we throw kids off short buildings to teach them that jumping off tall ones isn’t a good idea.
What we’re out to do here is encourage good buying habits (which, btw, is good for any major purchase), and that requires looking ahead and planning. We can’t give a single piece of advice that will fit everyone, but we can give you good advice on how to go about what’s right for your particular set of circumstances.
Hammer: Over and Above
Looking and planning ahead is also a big reason why I’m very skeptical about the average reader buying the first-generation of Athlon 64/FX systems. They at least look to have future compatibility problems over and above the ones mentioned above.
1) Will the memory controller in Athlon 64/FXs work with DDR2? Don’t know, probably not.
2) Will socket 940 mobos work with future socket 939 processors? Don’t know, current indications are they won’t.
3) Will socket 754 have a long life, or will it be supplanted by socket 939? Don’t know, tend to doubt it, but I wouldn’t even take my word one way or the other for it. In any event, see item 1) above.
To me, these platforms look like nine-month wonders, which given the amount of money AMD expects you to lay out for them, isn’t going to be too good for a lot of people. If I’m wrong, AMD is perfectly free to say so, and I’ll be more than happy about it.
If these issues do not bother you, as long as you’re informed about them, fine by me.
But if these things do bother you, you ought to know for sure before you buy. This is one area where it is foolish to “think positive.”
If you don’t know for sure from the company itself, don’t buy, and if someone tells you, “Don’t worry about it,” ask him, “Will you buy me a new CPU/mobo if you’re wrong?”
Yes, there’s a similiar situation with Prescotts. Socket 478 is on its way out, but that’s the difference between Intel and AMD. We know socket 478 is on its way out, so we can plan accordingly. Twelve months from now, Intel will make a big leap and have socket 755 AND all the PCI Express/DDR2 etc. mobo, with a socket 755 Prescott standing in until Tejas arrives. How well it will work is anyone’s guess, and the more prudnet or budget conscious may want to wait until Tejas or beyond, but at least we know it’s going to be a big break, and know well ahead of time.
(Hint, hint, this is a big reason why corporations buy Intel. Yes, yes, I know Intel isn’t perfect when it comes to this, but they are better about it. This kind of secrecy hurts AMD in the long-run; it doesn’t help it.)
There’s nothing wrong with the answer “No” when it comes to looking ahead and planning. What is wrong is not providing the answer to those who need to know when they need to know it.