Sign in with
Sign up | Sign in

GeForce 9800 GX2

PCI Express 2.0 Graphics Cards: How Much Extra Performance Do They Really Give?
By

pci express 2.0

pci express 2.0

pci express 2.0

pci express 2.0

The dual GPU GeForce 9800 GX2 card requires at least x8 PCI Express 2.0 to deliver high frame rates. Going to x4 link mode clearly bottlenecks the graphics flagship.

pci express 2.0

pci express 2.0

pci express 2.0

pci express 2.0

Crysis shows a noticeable impact when running on a reduced number of PCI Express 2.0 lanes, but only at very high quality. It’s amazing to see that you’ll get almost full performance even at x4 link mode.

pci express 2.0

These results confirm what we already saw on the ATI Radeon HD3850 card: Flight Simulator X heavily depends on PCI Express bandwidth. Reducing the x16 PCI Express 2.0 links to slower ones noticeably bottlenecks the simulator, as access to textures stored in the system main memory becomes far too slow.

pci express 2.0

pci express 2.0

pci express 2.0

pci express 2.0

Display all 6 comments.
This thread is closed for comments
  • 0 Hide
    mi1ez , 24 April 2008 17:28
    The x1 mark is wrong in the diagram- it's the gap right after the key, not ON the key!
  • -1 Hide
    Anonymous , 28 April 2008 18:15
    My nVidia GeForce 8800 GTS 512 is marketed as being PCIe 2.0. Why is this not listed?
  • 1 Hide
    spanner_razor , 5 May 2008 03:53
    Typo on the last page referring to 9800GX2 as 9900.
  • -1 Hide
    jeanpieterse , 7 August 2008 23:23
    My nVidia GeForce 8800GS 384MB is marketed as being PCIe 2.0. Why is this not listed?
  • 0 Hide
    jodrummersh , 18 October 2008 00:51
    This article is terribly misleading and does not test what it promises. It's only purpose is to test PCIe 2.0 x16 against PCIe 2.0 x8, x4 and x1.

    Woulda been nice if they actually tested PCIe 2.0 (x16) against PCIe 1.1 since this would actually be of some relevance and importance to us. Anyone else think so?
  • -1 Hide
    erdinger , 20 January 2009 22:09
    This article is terribly misleading and does not test what it promises. It's only purpose is to test PCIe 2.0 x16 against PCIe 2.0 x8, x4 and x1.

    Woulda been nice if they actually tested PCIe 2.0 (x16) against PCIe 1.1 since this would actually be of some relevance and importance to us. Anyone else think so?

    I agree