6800GS AGP vs 6800GT PCI-E

[Rhezuss]

Hey,

I just tried my new 6800GT with 3Dmark05 to see the difference with the 6800GS AGP I just RMA. My 6800GT scores lower than the 6800GS???

Here's the spec:

Formerly - A8V/A64 3200+ (stock) + BFG 6800GS OC 256mb AGP (370/1050, unlocked)
3Dmark05 score: 5276

Now - ASRock 939Dual-SATA2/A64 3200+ (stock) + ASUS Extreme 6800GT 256mb PCI-E (350/1000 stock)
3Dmark05 score: 4875

Is it that the GS was clocked a little higer? Is it the new motherboard (ASRock) that is not as performant as the ASUS, even with PCI-E?

I really don't understand the 400pts score difference.

 

[JPB]

Same thing here.... http://www.gpureview.com/show_cards.php?card1=354&card2=180

 

[TheCanuck]

I've actually heard that PCI-E is a little slower than AGP if you're looking at the same clocks speeds for the same card. Also, since the BFG 6800gs is overclocked slightly and you unlocked the pipes you'd expect higher results in 3d2005 than a stock 6800GT.

For the record, I've got a BFG 6800gs unlocked at stock mhz with an Athlon XP 3200+ and I get 5200 pts in 3dmark2005 as well. 3dmark2005 is more of a graphics test, so processor speed / motherboard type won't affect your score that much.

 

[Rhezuss]

Originally posted by: TheCanuck
I've actually heard that PCI-E is a little slower than AGP if you're looking at the same clocks speeds for the same card. Also, since the BFG 6800gs is overclocked slightly and you unlocked the pipes you'd expect higher results in 3d2005 than a stock 6800GT.

For the record, I've got a BFG 6800gs unlocked at stock mhz with an Athlon XP 3200+ and I get 5200 pts in 3dmark2005 as well. 3dmark2005 is more of a graphics test, so processor speed / motherboard type won't affect your score that much.

Thanks for the info. It's just disappointing, I thought that PCI-E was a little faster than AGP even with lower clocks. I'll try to OC the GT a little to at least equal my GS scores.

 

[Rhezuss]

Ok, I just OCed the 6800GT from 350/1000 and now at 420/1120. It's scoring 5667 at 3Dmark05 instead of 4875 stock. All that with the stock cooling and the same temps when running 3DM05 with stock clocks.

I'm pretty impressed with the results