![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
superkdogg
Gawd
- Joined
- Oct 14, 2004
- Messages
- 743
I honestly wasn't trying to flame before, and I'm not now either but here are some reiterations of things that people are overlooking.
1) PCI-e remains unnecessary except for those who want and use SLI/X-Fire. When NF4 came out for 939, people were saying (truthfully) that it is basically the same as NF3, but using PCI-e instead of AGP. The AGP bus allows the same bandwidth (forgetting about encoding/watching realtime HD) as PCI-e 8x. That's what all but the latest SLI boards "limit" cards to when running in SLI anyway. If you want to convince me that AGP was dead 2 years ago, show me a graphics card that is limited by the lack of bandwidth. I never said that PCI-e wasn't better, I just said that all of a sudden PCI-e forced (or encouraged) motherboard upgrades for no real reason.
*On a side note: I'm open to the fact that I'm wrong on this, but I bet that the 2x16 boards don't perform any better than the 2x8 boards in SLI because the interface is not the bottleneck. Anybody seen any benches that make me eat crow?
2) BTX- A better spec? I'm not sure. As some have noted, it is a lot like AT. So, either AT/BTX is better or ATX is. No matter which way you go, it proves my point that sometimes Intel will try to pull "innovations" just to encourage more new hardware purchases. Is it good business? Very much so. It sells chips/boards/cases, etc. Moving to whether or not it benefits consumers is something else entirely.
3) Summarizing my stance on the innovations is basically like this: If the latest video cards were still available for AGP, there's no reason that an ATX, pci (no e), socket 939 DDR1 system would be outperformed by a brand-new top of the line system with the exception of SLI. The system in question would be nearly two years old and would still stack up just fine if the graphics cards were still made for AGP.
4) DDR2 is only faster if the architecture in question needs more bandwidth. Which is precisely why Intel implemented it. Don't forget that the latencies also double. Which memory type is faster is very dependent on CPU workings and saying one or the other is faster is inaccurate. Also, can you point me in the direction of a benchmark that demonstrates the benefit of running DDR2 instead of DDR1 on an otherwise identical system? I haven't seen one like that. ----DDR2 is obviously the future. It is cooler and hopefully someday being cheaper to produce is helpful for consumers rather than facilitating lining the pockets of the manufacuturers.
Simply put: Most of these innovations were valuable and will continue to mature into strong, mainstay-type materials. The timing is the part that I find questionable, as most were released just when Intel realized there was a valid threat from competition, and even years later the new technologies have yet to be fully embraced (AGP cards still outsell PCI-e). I think it's fair to judge innovation by the rate and scale in which it is adopted. If new technology is wanted and needed people will snatch it up immediately. If it is merely incremental change or if there are other factors which make it less necessary and more obligatory, the new technology is adopted at a much slower rate-which is what is happenning to all the technologies being addressed.
1) PCI-e remains unnecessary except for those who want and use SLI/X-Fire. When NF4 came out for 939, people were saying (truthfully) that it is basically the same as NF3, but using PCI-e instead of AGP. The AGP bus allows the same bandwidth (forgetting about encoding/watching realtime HD) as PCI-e 8x. That's what all but the latest SLI boards "limit" cards to when running in SLI anyway. If you want to convince me that AGP was dead 2 years ago, show me a graphics card that is limited by the lack of bandwidth. I never said that PCI-e wasn't better, I just said that all of a sudden PCI-e forced (or encouraged) motherboard upgrades for no real reason.
*On a side note: I'm open to the fact that I'm wrong on this, but I bet that the 2x16 boards don't perform any better than the 2x8 boards in SLI because the interface is not the bottleneck. Anybody seen any benches that make me eat crow?
2) BTX- A better spec? I'm not sure. As some have noted, it is a lot like AT. So, either AT/BTX is better or ATX is. No matter which way you go, it proves my point that sometimes Intel will try to pull "innovations" just to encourage more new hardware purchases. Is it good business? Very much so. It sells chips/boards/cases, etc. Moving to whether or not it benefits consumers is something else entirely.
3) Summarizing my stance on the innovations is basically like this: If the latest video cards were still available for AGP, there's no reason that an ATX, pci (no e), socket 939 DDR1 system would be outperformed by a brand-new top of the line system with the exception of SLI. The system in question would be nearly two years old and would still stack up just fine if the graphics cards were still made for AGP.
4) DDR2 is only faster if the architecture in question needs more bandwidth. Which is precisely why Intel implemented it. Don't forget that the latencies also double. Which memory type is faster is very dependent on CPU workings and saying one or the other is faster is inaccurate. Also, can you point me in the direction of a benchmark that demonstrates the benefit of running DDR2 instead of DDR1 on an otherwise identical system? I haven't seen one like that. ----DDR2 is obviously the future. It is cooler and hopefully someday being cheaper to produce is helpful for consumers rather than facilitating lining the pockets of the manufacuturers.
Simply put: Most of these innovations were valuable and will continue to mature into strong, mainstay-type materials. The timing is the part that I find questionable, as most were released just when Intel realized there was a valid threat from competition, and even years later the new technologies have yet to be fully embraced (AGP cards still outsell PCI-e). I think it's fair to judge innovation by the rate and scale in which it is adopted. If new technology is wanted and needed people will snatch it up immediately. If it is merely incremental change or if there are other factors which make it less necessary and more obligatory, the new technology is adopted at a much slower rate-which is what is happenning to all the technologies being addressed.
And then you'll be glad to have stratospheric RAM speeds, 'cause those suckers are gonna need all the bandwidth they can get! 