Resistance is futile, but my stubborn mind will eat you and pessimistical thoughts for breakfast! (lol, jk)
May starve I'm rarely a pessimist except for when it comes to stopping the natural progress of things, it's like trying to tell people CRT is better than (current) LCD, Plasma, DLP, LCOS for qualiyy imaging. Doesn't matter how correct that statement may be, people are buying LCD, and soon even quality CRT at ridiculous premiums will be hard to come by and the imaging proffesionals can only hope that white LED backlight LCDs, OLEDs and flat EMDs can do as good a job. Eventually they will surpass any CRT we have now, but they need to get a move on for some of us since even Sony has stopped making their professional series tubes and they are rebranding lessr quality now.
Of course people aren't buying AGP anymore, it has fallen into the deep depths of "old" technology. I am just claiming that the AGP-to-PCI-e change had no practical reason, except SLi which I do not believe too many people have...You may have SLi boards, yet how many actually have two cards running in tandem?
There are a few other practical reasons, like ease of design, flexibility of power useage (1X one moment 16X the next), future scalability (while right now 8X may be fine, what about next gneration), and would that have meant an AGP 12X or 16X anyways? What about HyperMemory and Turbo cache, no way AGP could've handled that.
Also depending on the application 8X was already nearing saturation point in workstation apps according to Sudhain's investigation just before the launch of PCIe.
There is a smidgen of use in it under the theory that those with one card would use AGP 8x while those opting for SLi would be forced over to PCI-e x16 and the confusion may be too much for some people, or the compromises that might have to be made.
But like you say that's a minority of the PCIe users, however even still to the MFRs those users are far more valuable than those still wishing AGP were around and refusing to upgrade. Look at it from a marketing perspective of someone in the middle. Do you understand the MFR's desire to move to a single production line? Do you understand some people's resistance? Do you then take that into account and buy transition products like the AsRock S939 Mobo or the myriad of other socket PCIe mobos to allow people to keep the rest of their rigs when they finally decide to move?
Those few who are "true" enthusiasts, SLi was created for them. Only recently has the mainstream even thought about using it, and the irony is that it is so much more practical for them.
Not really, more on that in a moment.
A 7900GTX is a pretty fast card alone, no doubt about it, and two are hardly nessecary except for playing at extreme resolutions with high AA/AF, while on the other hand a 6600GT is a mainstream, reasonable card but with the resolution turned up and a little AA/AF it'll start to lag on newer games and a second would certianly come in handy.
Except that it shares that 128MB, therefore still is both resolution and AA limited. Also if you sell that old GF6600GT and put the money you would've spent on another GT and the SLi MoBo, you can probably get a far faster single card. That is similar to the situation confronting most AGP users, instead of wasting money on that terrible plain GF7800GS for $300, spend $75-100 on a new PCIe MoBo for your socket (Socket A/S423 need not apply) and then spend the remaining money on a PCIe card that will far outperform, and give you the future upgrade path you have to bite the bullet on anyways. And then when time to upgrade your card again, the future value of your true GF7800GT will be more than the GF7800GS that no one really needs anymore.
BTW, PCI-e was created for SATA ports, dual GB ethernet, etc. to be able to be made on a card so that if you needed extra SATA ports, you would not be forced to upgrade to a new motherboard.
Nope, PCI-X already did that long before PCie.
GPU's were more of an extra idea as AGP is certainly not a bottle neck.
Nope, from it's early days, shortly after inception/conception 3GIO meant to be the unifiying replacement for all things, only AMD's hypertransport concept would've split things again, which IMO is part of the reason they lost the race.
PCIe for GPU's would have better implemented if there was a true need for it, when we actually needed that much bandwidth.
no it would've been far better implemented if like so many things before it you didn't have these weak fallbacks. The failure of the prescott as the launch vehicle for PCIe didn't help much either. Without an S939 mobo with PCIe there was no drive to adopt since the AMD64s were kicking the Prescott's but, and without that push everyone's left wondering what to do. Keep buying the better performing AMD64 and the current AGP cards, or buy an underperforming socket to anticipate the video card upgrade path. Neither Intel not AMD make their dual core or 64 bit parts in socket A or socket 423/478 format, but you don't seem to see people screaming about that? Why, because you know they don't care, you know they are moving and so should you. On a socket 754? Expecting duo core are you? Good lluck with that!
With that, I can gurantee that PCI-e x16 motherboards would have sold so many more units when it was proven that PCI-e x16 was viable.
Viability had nothing to do withit, they would have sold many more if prescott had thumped the AMD64s, and had PCIe been brought to S939 at the same time. If the performance and heat situation were reversed I doubt we'd have seen the GF7800Gs, nor even the X850XT-AGP, and we likely wouldn't be having this conversation either. It's mainly because of that one mis-step that PCIe failed to launch properly. Without that I doubt ATi would've even bothered to implement Rialto in their X700s, and nV would've move on as well. Instead the based for S939+AGP got slightly larger, and thus created many more people who had just upgraded and didn't want to again.
As much as people complain about he transition, at this point in time it's purely economics, you get a better bang for your buck with PCIe and it's cheaper for the MFRs to produce just PCIe. You can still stick with AGP in the middle/muddle performance range, but there's no incentive for AGP development anymore, not for MFRs nor the consumer, and likely extending more lifesavers like the GF7800GS just makes the situation worse, not improve it. No wonder it was nV's OEM partners that pushed for it, they are looking for the quick buck whether it's ATi, nV, XGi/S3, Via.
If you can give me one logical reason why AGP > PCIe then perhaps your argument would have a leg to stand on, but as I doubt there is that reason (other than reticence/resistance to upgrade) then the argument's more about people than about the technology, and like I said before, the rest of us don't care if they get passed by, we just want to stop hearing the constant whining and bitching from the admitted peanut gallery, and for the MFR to keep rewarding those who buy the most new products and pay for the future R&D. AGP is the past in both sales and technology, nothing to gain by focusing on that.
"
Hello AGP laggards, let me introduce you to the ISA fan club, they will be able to help you through your transition as they've been there and done that. They're likely also the only people who really empathize with your situation."
Once again, it's done, move on, upgrade and be done with it, the AsRock S939 SATA2 board is perfect AM2 support, and AGP+PCIe, what more could you want?
Facebook
Twitter
Instagram