<font color=red>** Warning **</font color=red>, kinda long, could be boring, definitely rambling, kinda guessing, and if you are going to read this, go take a wiz first, and maybe grab a coke or something in case you get thirsty. :lol:
What is the point of releasing new technology, if you can't even offer it to the public? sure I'll just get myself a new alienware or Voodoo PC everytime a new videocard is announced.
Well actually the public can get the latest and greatest usually, and sadly, from DELL. I got my flyer last week from them with the X850XT as an upgrade on the XPS system for only $50CDN more than the GF6800-256mb (not GT, not Ultra like I initially thought). Now Dell probaly not only moves more chips than any other single vendor for ATI, but they probably also sell more 'high end cards' or their XL/GTO variants than Voodoo and Alienware comdined.
Now while I agree that these things are coming fast and furious, right now demand far outstrips supply, and while that sucks for us, it's far better for nV and ATI than the opposite.
They go and shove some more stuff up the R520 and release some uber PE gold silver with crazy ram with crazy fan with crazy box edition.
Actually that's the opposite of what I see, I see the R520 as being a likely UNDERCLOCKED card because it will have natural headroom thanks to smaller process and low-Kd, nothing nV has will match that (they don't even us 0.13 low-K now). So this allows you to release a core, that like the R9600PRO has a wide range of performance characteristics (from those like mine to those like yours :tongue: [sorry just had to
). Now the refresh on that is simple, move the natural stock clocks up to where the OC'ers are putting them already anyways, and the maturity of the process should yield you more chips that reach the higher levels. This of course makes the 'reviews' see a jump which is as insignificant as the R9800Pro to R9800PRO-256 and X800XTPE to X850XTPE; add the uninformed who think they have a refresh worth spending more money on (outside the at all cost moders). Now that would be the wise way to introduce the next generation core IMHO. This built in 'room to spare' creation should also allow for far more chips to meet the standard speeds, as would even their cast off I'd think. Any pro version and non-pro or SE version wouldn't have to do much to best our current options in those areas.
One other thing to remember that the ATI SM3.0 refresh HAS to happen, now while it doesn't have to happen soon, it does have to happen sooner rather than later; because the 'futureproof' timelines are really starting to make a play with the next great engines being toutted as less than 12months away from the expected launch date.
nVidia and ATi don't give a [-peep-] if they can provide the chips with now, they're just wanting to kick eachother's ass.
I don't think that's true, call me the cynical economist. I see this from a different perspective (allow me great lattitude to unfold my personal conspiracy theory or commensal collusion). Both ATI and nV have their weaknesses, and to me alot of this is a holding pattern where neither benifits, but their losses are less than what it could be. Neither seems to be able to figure out what the heck is happening with PCIe, nV likely happy they aren't forced to get their HSI problematic NV40 to market right now, and ATI unsure of how much they need to produce for this generation either (especially if it is shown to be lacking in anyway for future games). Neither wants to get caught with a glut of cards that they can't offload except for at a loss, so keep the market stoked (although annoyed) and only produce the bare minimum you need to make until you can figure out which way the wind blows and what you need to produce. Now this is a tough issue because your long term plans are conflicting with these short term PCIe/AGP issues as well as other irriations like HSI bridges Rialto bridges, etc. nV has the luxury of cancelling or delaying chips, ATI on the other hand is running into 'inevitability' Mr. Anderson.
. They desgined and devloped the X850 and X800XL series card, the X800/XL address your production and cost issues and give you much needed market pressence that appeals to a good target market. It also addresses the short comings and production issues associated with the X700XT. Now while all this is going on you have the R520 program, still going forward like it has been for over a year. However they are long finished and simply waiting for a green light to produce. However what about all these old chips, have we moved enough yet? Well no not as many as we'd like, but putting off the R520 in order to try and recoup any more for the R420/423, only means delaying the returns for the R520. Now the dicision to go forward likely involves alot of factors, even included something so small as the idea that it was announced and to miss that date would be negative. Sure the biggest factors are other things, like expected yields, benifits and efficiencies as well as possible advantages over the competition. Now in addition to that ATI has alot of chips that don't meet certain aspects of the competitor's roadmap. No SM3.0, and no current chip with AMR/SLI capabilities (even in the very latest), so in order to introduce those, you have to bring in a new chip. Why bother with a low end when you're already toying with a chip design for M$, and you can benifit from watching it's production and from letting M$ fab and thereby test your basic designs and new ground. Working on a similar chip makes more sense than working on an RV5xx instead and playing it safe in such a way as to devlop a stepping stone or something completely different. Sure the R500 and R520 are different, but they also have enough in common to learn from.
With this wide shortage of X800 and 6800Ultra, I can tell you the situation is not going to get better, no matter how TSMC, IBM can improve the yield rates, as GPUs becomes more complex, they need time to figure things out.
While that may be true I think that the yields of the first few R520s don't need to be as high as they would be if they transitioned the new design in something that did need high volume like the mid-end. While moving to a new process is likely best done on the mid and low ends, moving to something that may need alot of respins to get right might be better done with a core that you can profit more of individually with a low initially start. Also really what would it say if ATI only came out with a low end SM3.0 card, which really likely would be as pointless as an FX5200's or GF6200's 'full feature-set'. It might make alot of sense for test purposes, but would likely also lead to the question of why the heck you can't do it with the cards I'm paying 2-4 times as much for!
I think that's more at issue than actual yields, because directly related to yields in the selection of 'stock' speeds, and by lowering target speeds significant yields <i>should</i> be achieveable, and IMO if there is no mV reply ATI doesn't have to play the silly 'let's see what their clock speeds are then we'll chose ours game that they played with the launch of the X800 and X700 series cards.
Of course nV's response will be interesting to see because there is no way they are simply going to sit back and do nothing, but likely their path will closely follow the NV40 in the way that the R420 closely followed the R300's design, rather than coming out with a completely new GPU.
Anywhoo, that's my view of it. Of course it could be 100% wrong, well maybe 99.44% wrong, but still I think there's more than just yield issues involved here, which I doubt would cause all these shortages alone for such a long period of time. But once again, just my two frames' worth.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK