OK, this thread is so huge I just stopped reading a few posts after my reply. I'll post this reply in case people are interested, and as a courtesy to those who replied (relatively quickly) to my first post.
Ben,
But it is much simpler then getting ATi's AFR parts working under any currently supported OS
You got me there.
The SLI idea from nV is a fairly solid move for them from numerous angles. They can push their "nForce4" mobos along with its dual PEG slot configuration to buyers along with two of their video cards, not to mention it makes them a complete lock on top of benchmark charts for some time to come.
It's a PR bonanza for nV, no doubt, but I'm more interested in what it'll do for me.
Wavey is rubbing off on you a bit much Pete, you are taking on the ATi apologist roll here. If I could SLI my R9800Pro right now I would, another $200 to ~double up on performance would be an absolute no brainer. Most of the upgrades the enthusiast market take part in tend to be along the lines of trying to double performance, and for vid cards that means shelling out ~$500 lately. Now that figure can be ~$200 for a lot of people if they build their next rig with it in mind ignoring completely the uber rigs that some will build for considerably more.
It's quite possible--I've been spending way too much time in video forums, and at B3D in particular.
But my point was that even SLI'ing your 9800P "right now" won't happen, as you don't have the MB or possibly even the PSU to make it happen. OK, you say, let's start over from when dual-PEG MBs are available, and let's assume they're only ~$20 more than a similar single-PEG board (I'd be impressed if this'll be true, as they'll need all the extra traces and power components to support another PEG slot, and they probably won't be getting mass-market pricing for these boards). Let's even assume you have a beefy PSU. So you buy a $200 128MB 6600GT. It starts to drag later on, so you add another a year later. Now you have 16 pipes, 500MHz core, 500MHz DDR--basically a 6800U+. Except for one thing--only 128MB RAM. I'm not sure how much good all that speed will do for you with that little RAM. Plus, I'm looking at this from the perspectiev of a brand new tech generation. SLI looks a lot less attractive at the high end when we're switching from SM2.0 to SM3.0, or SM3.0 to SM4.0 cards, IMO.
It's a really, really cool idea, and awesome for the rich &/ determined PC gamers and potentially for workstation markets (speed is a factor, but so are drivers--but SLI Quadros are a definite possibility), but I'm not convinced it's a better upgrade path than we have now in the $200-300 midrange. Quadruple DVI, as jiffylube mentioned, is intriguing for the financial set, and possibly for Longhorn, if it supports 200+dpi displays, as I've read elsewhere at AT.
Rollo,
1. It's not as simple, but if I'm upgrading motherboards anyway, a nForce4 board and a PCIE 6800GT is a nice way to go this year, and next year I can add one for less than the cost of two $400 cards I bet.
It's a great option for those who can afford it.
2. How is it a "lot of effort"? And why wouldn't someone with a 450> psu want to try for two 6800NUs? That's 24 pipes for under $600? Beats paying $500+ for 16?
Well, ensuring you have a dual-PEG MB and a sufficiently studly PSU isn't no effort. It's a lot more effort compared to a Voodoo 2, but I guess those carefree days of not having to find spare molexes are long gone.
24 pipes for <$600 sounds great, but only 128MB seems like a huge limiter. Not show-stopping, but it seems important. I haven't entirely thought it through yet, though. Again, this is looking at it from the perspective of those who buy a new $200 card every year or two, not those who buy a new $400 card every year.
3. There is something wrong. You can't use the lack of ps2 games as the main fault of 5900s and then in the next sentence say the lack of SM3 games is the main reason you don't need SM3.
Your logic isn't entirely clear to me there, but I'll grant you that touting SM2.0 and then downplaying SM3.0 is a bit double-faced. But the leap from SM1.x to SM2.0 still seems more important than that from SM2.0 to SM3.0, IMO.
I've got $20 that says there are more SM3 games out by June 2005 than there were PS2 games out by June 2004?
I don't think we're looking at comparable situations, so I'd prefer to spend that $20 on that swanky Sandisk 8-in-1 reader, if you don't mind.
4. As for my "weak excuse", I said it's an example of why I pimp nVidia these days?
And I said it reflected less badly on ATi as the 3DM03 attack did on nVidia. (I'll even ignore how much weight people placed on 3DM03 in the 6800's debut.
)
If you can't buy damage control BS like this interview, and the "real" trilinear nonsense, as contributing factors in my favoring nVidia of late I guess that's your perogative.
To each his own, but I don't think offering the option for full trilinear makes up for nV's behavior last year. Then again, it's quite probable ATi would have done with same with equally-underperforming hardware. I just don't see what ATi has done wrong and nV right to swing you back into the nV camp.
5. HS insult fixed, we'll leave Gabe out of it.
Cool. I think we ATers fling enough insults at each other that we don't need to drag professionals into this mudbath.
Insomniak,
$70 more right off the bat seems a bit much to me, but you've got a good point about how attractive SLI can be for the midrange, and the 6600GT in particular. I'm still a little unsure about only 128MB RAM, though.
But I do think SLI is targetted as much at the very high end as at anything else. If you had the money, don't tell me that you wouldn't want 2x6900U performance this winter when everyone else is bragging about "just" 1x6900U performance.
VIAN, thanks for sticking a "good" in there.
There, I've filled my smiley quota for the week. Back to frowning fope me!
Facebook
Twitter