First PCI-Express Graphics card

IDE devices do not use PCI bandwidth in chipsets that are not PCI devices themselves. That's the whole point of using a faster-than-PCI chipset interconnect such as in iNTELs i8xx series. VIA and SiS have their own proprietary connections too, while Nvidia uses HyperTransport.

What use would a 400MB/s bidirectional connection be if all your devices resided on the not-very-efficient 133MB/s PCI bus?!?!?

*G*
 
http://www.bit-tech.net/feature/16/4
PCI Express. This is a new initiative, being led by Intel, to bring to market the next generation of the PCI bus. Much as PCI phased out ISA, it is hoped that PCIE will phase out PCI as we know it. PCI currently, like ATA, runs on a parallel bus. This creates problems with error control and scaling when we get up to faster and faster bus speeds. PCIE is based on a serial bus (much as serial ATA is being phased in now) and can go faster, with more bandwidth, than we currently see in PCI. It was announced that 8x will be the last AGP interface, and PCIE will be the de facto graphics interface. We will be seeing PCIE graphics cards from both Ati and nVidia in time for launch of PCIE in the middle of 2004. This struck me as a little harsh : at least when graphics cards made the move from PCI to AGP we still had the legacy PCI interface to allow users to use their existing graphics cards as they upgraded motherboards. When I queried this, I was simply told that AGP was being replaced and board partners would be ready to go. If AGP is not even going to be kept on to facilitate initial motherboard upgrades, there are going to be a heck of a lot of cheesed off graphics cards owners, I would imagine.

Anyone have any idea what graphics card might have plans on using pci-exp (ie r500,nv50) or will we see current and near future products updated to use pci-exp.

later,
 
Althornin said:
Dave H said:
The Southbridge consists of an IDE controller, which connects to the hard drives, optical drives, etc. over the IDE bus;

There is no "IDE bus" per se. Those integrated IDE controllers reside on the PCI bus, and communicate with drives via IDE channels. Accessing the HD uses some of your PCI bus bandwidth.

Just nitpicking :)
More nitpicking: the IDE bus is the bus that sits between the IDE controllers and the drives.
 
I'm sure there will still be 'legacy' AGP slots for at least a year after PCIE is available... and possibly also legacy AGP compatible versions of cards.
 
I guess I could go look this up myself, but I figure one of you guys already know so.. : will PCI-Express slots provide enough power to keep out R500s and NV50s going, or will we still need to connect them directly to our PSUs?
 
They still have PCI graphics cards out there, and there's a question as to whether there will be legacy AGP cards????? ;)
 
Answers to these questions from Anand's IDF coverage:

Use of AGP cards in PCIE mobos will be facilitated with PCIE-to-AGP bridges (presumably you plug a thingy into your PCIE slot and your AGP card into that thingy). ATI has already demonstrated a 9700 Pro working with one, apparently.

PCIE spec delivers 60W to the card, a nice improvement on AGP. And which probably seemed like a lot when the spec was set, but won't be enough to keep extrenal power connectors off high-end cards. (The measurement I've seen on GFfx Ultra power draw is 75W.)
 
Dave H said:
Use of AGP cards in PCIE mobos will be facilitated with PCIE-to-AGP bridges (presumably you plug a thingy into your PCIE slot and your AGP card into that thingy). ATI has already demonstrated a 9700 Pro working with one, apparently.

Maybe a plug in thingy for the lab, but this bridge will be part of the motherboard, if not part of the chipset in a retail motherboard. We already have similar things for ISA to PCI bridges and PCI/PCI bridges. (Check your device manager info in windows and you'll see these things listed)
 
Back
Top