PCIE v.2.0 vs PCIE v 1.0

[dynamicruss]

Is it possible that a GPU spec'ed at PCIE 2.0 would not perform as well if it were placed in a PCIE 1.1 slot?

 

[thescreensavers]

Yes,roughly kills 50% of a card performance , 2.0 is about 80% as fast as x16

 

[TJ Tom]

Last time I checked the HD5970 only performed 1-2% less in a PCI-e 1.0 slot. I think the bandwidth isn't a bottleneck yet for most cards, maybe soon it will be one though.

If you're curious, you can see how a certain card performs in an X8 slot, I was told that's close to a PCI-e 1.0 slot.

 

[veri745]

Like TJ Tom said, there is a minimal performance impact.

http://forums.anandtech.com/showthread.php?t=2045386&highlight=pci+scaling

 

[tcsenter]

Yes,roughly kills 50% of a card performance

Wrong decimal place. Should read:

"Yes, roughly kills 5.0% of a card's performance.."

 

[MrK6]

Is it possible that a GPU spec'ed at PCIE 2.0 would not perform as well if it were placed in a PCIE 1.1 slot?

It depends on the card. For cards that really need the bandwidth, such as the 5970, GTX295, and other multi-GPU, single card solutions, the performance will be hindered a bit. For something midrange like a 5770, the performance drop will be insignificant (~1%).

 

[RobsTV]

I think it also depends on the motherboard and the card used.

I bought a new cheap Foxconn 939 motherboard and an HD3870 (one of the first PCIe 2.0 cards I think), then was shocked to see how slow it scored in benchmarks. Much slower than the x1900 I usually run. Yes it was around 50% slower in some tests! Popped the HD3870 in a different old Jetway 939 I had running with the x1900, and the HD3870 then performed great, faster than the x1900, like I was expecting all along.. Both 939's were PCIe (1.0 or 1.1?). To confirm there was nothing else messed up in the system causing the slowdown, I tried the x1900 in the Foxconn, and it too worked great, with similar scores to when used in the Jetway. Bought a newer PCIe 2.0 motherboard, and the HD3870 scored fine, as originally expected.

 

[happy medium]

I think it also depends on the motherboard and the card used.

I bought a new cheap Foxconn 939 motherboard and an HD3870 (one of the first PCIe 2.0 cards I think), then was shocked to see how slow it scored in benchmarks. Much slower than the x1900 I usually run. Yes it was around 50% slower in some tests! Popped the HD3870 in a different old Jetway 939 I had running with the x1900, and the HD3870 then performed great, faster than the x1900, like I was expecting all along.. Both 939's were PCIe (1.0 or 1.1?). To confirm there was nothing else messed up in the system causing the slowdown, I tried the x1900 in the Foxconn, and it too worked great, with similar scores to when used in the Jetway. Bought a newer PCIe 2.0 motherboard, and the HD3870 scored fine, as originally expected.

If I remember corectly there are issues with pci-e 1.0 and later cards. Pci-e 1.1 is much more compatable.

 

[thevan]

Well if I remember correctly from back in the AGP days, the advancements from AGP4x to AGP8x did very little as the bus was only being used when the card ran out of memory. So I would say it doesn't matter too much, especially for higher-end cards with lots of memory. Plus you should be able to test it out easily enough by lowering the speed of the PCIe slot in the BIOS.

 

[soccerballtux]

Yes,roughly kills 50% of a card performance , 2.0 is about 80% as fast as x16

Wrong decimal place. Should read:

"Yes, roughly kills 5.0% of a card's performance.."

I don't think it's even that much. more like 3% maybe. Depends on the game.

 

[dynamicruss]

Thanks all for the responses, I originally had 2x Nvidia 9800GX2's in an Intel Skulltrail MB, I recently upgraded to a single Radeon 5870 and am still experiencing lagg during gaming. (Only during gaming, regardless of settings, and in some cases crash to windows.) However, if I understand correctly, having these cards in a PCIE v1.1 should not be the issue?