Dread there is a reason @MD used a rating scheme, cause the 3000+ was supposed to compete with a 3GHz. Or perhaps ur a person who still is of the megahurtz myth... dude get a clue, when ur beloved 1ntel starts using a numbering scheme too (cause speed doh mean crap NO MORE) yuh cyah come in here and buss down no 400Mhz FSB cause it keeping up with ur 800Mhz FSB, or if u like we'll get someone with a A64 and the equivalent of a 1600MHz FSB and shut yuh up. Oh and one more thing yuh have a 5700 Ultra.... yuh not in the same PLANETARY system in terms of graphics so hold it down.
HAHAHAHA>LOL>Hard Luck Vinion2000!NOWAY a 5700Ultra could come round a 6800 and add what W1nTry say dey bout 3000+ competing with P4 3Ghz...and damn.I wish those were really some girl's on GATT's Tac Up....blast.
Graphics card names are the bitter end Come on chaps, get honestBy Charlie Demerjian: Monday 06 September 2004, 09:12GRAPHICS CARDS ARE STARTING to bother me. No, not because they are expensive, not because they are cheap, and not because I don't have a good one, it is a problem with names. There are too damn many names for each chip line, and each bi-annual refresh, there seem to be twelve more models. XT, GT, PE, FU, they are multiplying faster than rabbits, and faster than I can keep up with.To be fair, there are good reasons why there are nineteen different flavors of each chip, this week, and it is one of the things I hate most in the world, marketing. Yes, those little critters are at it again, and we all pay the price. Know the memory width of an ATI 9800XT? How about when you step down to a 9800Pro? Does it drop when you go to the non-Pro? How many pixel pipes does a NVidia 6800Ultra have? It goes down when you go to the normal, non-suffixed 6800, but where does the 6800GT lie in this scheme? So, if you head hurts from trying to juggle this little bit of info, lets talk clock speeds. A single chip line can almost double in core clocks from the lowest to the highest parts of the line, and the memory speeds also take a wild swing in speed. Combine the two, and you get a massive flowchart that would make my accountant cringe.But wait, it gets worse, much much worse. When you try to compare a low end part of a mid-range chip line to the low end part of a high-end chip line, the results are obvious. When you compare a high/medium to a low/high, you get.... Ummm... more headaches.Of course, you can always pick your favorite site and read up on the last test they did, and pick your card out. You run down to the store, and quickly find to your horror that none of those are in stock, but there are 73 other models there, and almost a third of those boxes are labeled intelligibly (link: http://www.theinquirer.net/?article=16470) as to what is in them. The expression is 'Sucks to be you', and the headaches begin.So, it is pretty much impossible to keep up with these things anymore, and no one I talk to actually likes the situation. So why does it happen, and why is it getting worse? Competition and marketing.Here is how the game is played, and you end up losing. There are two players in the high end of the graphics card business, with a few up and comers, but we will ignore them. These two companies are annoyed at each other because they are bitter competitors, but that should come as a shock to no one. They play games back and forth. Again, no shock here.They launch products at about the same time, usually within a few weeks of each other, twice a year. This tends to happen at about the peak buying season for computers, imagine the coincidence. Due to all the factors that go into making up a multi-million transistor semiconductor, one company is almost always later than the other in getting parts out the door. This usually means they are considered a follower, and end up having their last product line being bashed up and down by the competitions new parts.While it may seem bad, it can be a good thing. They know where their competitor will end up in the benchmarks, and have a little window to tweak their cards so they are just a little bit better than the other guy. It happens all the time, and rarely you can even see it happening.Due to the vagaries of making a semiconductor, speeds of the chips coming out of the fab vary, and usually end up in a bell curve. You set the speed that you want to sell the chip at so the majority of the chips you get back meet that goal. You want to set that number as high as you can make it, so you are the fastest, but also as low as is economically feasible. There is wiggle room. If you set it to high, and only 10% of the cards make that cut, you are going to have a very expensive chip to sell, and not many of them at that.One solution is to have many speed grades. The top 10% go to the gamer card, the middle 80% to the mainstream part, and the low 10% to some poor vendor who you caught leaking to Fudo or something. They call it a special line, unique to them, charge a premium, and laugh at you. Soon shooting erupts.Getting back to reality, there are good reasons why you want to have more than one speed grade for a line, it allows you to use more parts, and sell some at a premium. Both are good things, and gives the consumer a choice.The problem comes in when you are in second place in the release date game, and your architecture is slower. Imagine you have a clock speed set for 500MHz, and you know your chip will achieve 1000ArbitraryMarks2005. You tell your board vendors that this is the speed you will be running at, give them a few sample chips, and off they go, designing for a 500MHz part.Then your competitor comes out with an architecture a month earlier than you, and they get 1200AM on their new part. Once bladder control is reachieved, you then formulate a cunning plan. We up our clocks to 610MHz and like magic, our cards hit 1220AMs, and we win. Yay, great plan. Champagne all around, promotions and stock options abound, and it is time to buy that new BMW.Then, walking into the office the next morning, the problems start, and it doesn't help that those four bottles of champagne you downed cost $4.99 at the corner drug store. Right now light hurts, not to mention sounds. Emails from your partners all revolve around one central theme, namely 'you did what?!?'. It seems they all designed their boards for 500MHz with a 10% safety margin, and 610MHz won't work.But it gets worse. The fab says that a little less than .043% of the chips you ordered make that 610MHz cut. So, your old plan, having three lines suddenly changes. You have your old 300MHz 'blue collar' edition, the big seller 400MHz 'white collar' edition, and the 500MHz 'executive'. To make those ever important benches, you have to make a new model, but what to call it? Wait until a riced out Honda drives by, and look at the window stickers for clues, but remember, VTEC is taken, and will probably get you sued also.Enter the new top of the line model, the 610MHz VR-XM. You can make 57 of them with the first batch of chips from the fab, and the next batch due in 3 months may have a better bin split, so you might get a few more then. What do you do now?Simple, each hardware site gets one to review and pummel the competition senseless. That is called marketing. Your partners get the other 7, one each. That is called allocation. If you do the math, that is about one per continent, or about how many of the last round of high end cards hit the stores recently. Don't worry, in 90 days, you will have at least double that many to sell, and that should satisfy the pent up demand from almost a single metropolitan area.So, can it get worse? Yup, your competition, pissed off that you play games, releases their new overclocked card, the 1300AM XXL-43 model, they saw the name on a t-shirt or something, or it could have been tequila induced. They can only make 93 of them, but they pipped you in the benches. Solution this time? When the next batch comes out, so does the 1450AM scoring VR-ZT_Mega. You can only make 32, but damn they are fast.It goes on and on and on. Each new game brings a new model, and each new model makes the situation worse for you and me. They know the game and they know the pain it causes them, their partners, and their customers. Hardware site owners just love it, few things draw readers like a headline about 34 graphics card roundups testing everything iD has ever made from Commander Keen to Doom3. Yup, a winner at last.So, what is to be done? This one is so simple it will never happen. When a graphics company announces a line of products, announce it all, and announce the clocks too. You don't necessarily have to say when they will ship, but lay it out. It will prevent fiascos like when reviewers who get a card one week get a given clock, and the next batch get a faster one, and the stragglers get the fastest cards yet. They all have the same 'super-duper edition' stickers on them, and the PR people all swore that the clocks were right when you talked to them a couple of weeks ago. Strangely, a month or two after launch, you still can't buy the damn things. Gosh.Yeah, you have all seen it before, many times if you were paying attention. It shows no sign of improving, and no sign of changing. Sadly, even if I bitch out loud, I don't think anything will happen. Those benchmarks are so powerful that no chip maker can resist flying circles around the lightbulb of marketing games.
