mooch49

i have a p45 board as well. i am planning to crossfire 2 4870's. would this chipset's bottleneck effect that configuration as well.

Brutuz

I doubt it.

I know how you can test without getting an x48, just overclock the PCIe to 105Mhz and 110Mhz if its stable, then see if there's a noticeable difference.

__________________

Phil_Boggild

Well LET'S FIND OUT! I just ran the Crysis benchmark at PCIe x16 and PCIe x8 as verified by GPUz. As you can see from the screens, PCIe x8 was actually a couple of FPS faster, though this is obviously just margin-for-error stuff. I was pretty satisfied that the card was performing identically at x16 and x8 in the Crysis benchmark, but obviously I'm only running at 1280 x 1024. Things could very easily be different at 1920 x 1200.

In light of this, I decided to increase the burden on the card to see if I could reveal a bottleneck. I did this by applying 8xAA in the benchmark, and forcing 16xAF in the CCC. Once again, the performance of the card in the 16x slot and the 8x slot was identical. I realise now that AA was actually at 4x and not 8x.

The question is, does 1280 x 1024 with 4xAA and 16xAF create the same sort of demand as, say, 1920 x 1200 with 2xAA and 8xAF. I don't know. I'm not terribly clever.

All the same, I found this very encouraging. I'm also really impressed with the performance of my 4870, given how well it dealt with the hefty AA.

Attached Thumbnails

       

rico2001

@Phil_Boggild
Good stuff man. Pci-e works on bandwidth so I believe you would see a bottleneck at any resolution if there was one. Thanks for the test and pics.

__________________
My reviews:
4870, 4870 Tri-fire, 4850 X2 Quad-fire, 4850 X2 vs. 4870 X2,
4850 X2 + 4870 Tri-fire, 4850 X2 + 4870 X2 Quad-fire, 4850 + 4870 Crossfire,
(NEW) 4890 vs. 4850 X2

Steam nick: rico2001

"The ability to learn faster than your competitors may be only sustainable competitive advantage." ~Arie de Geus

almighty15

The bottleneck gets bigger as resolution get higher

__________________
Core 2 Duo E6420 @ 3.9Ghz : http://valid.canardpc.com/show_oc.php?id=558967

Pentium Dual Core E6300 - 4.2Ghz : http://valid.canardpc.com/show_oc.php?id=646455

Gigabyte EP45 Extreme FOR SALE.....PM for details

budgetgamer

And your worried about a little 1-2FPS bottleneck.

__________________

Phil_Boggild

I didn't claim at any point to be anything other than a lunatic

Phil_Boggild

From everything I've seen so far, I'm not sure I agree with you.

THERE'S ONLY ONE WAY TO FIND OUT.

I'll order my 1920 x 1200 monitor and run the same test again, and on THAT basis, I'll decide whether or not to run Crossfire.

rico2001

@Phil
From almighty15's idea and your testing, I could do the same since I have the same. I'll redo my 4870 review over at 8x and see what I get.

__________________
My reviews:
4870, 4870 Tri-fire, 4850 X2 Quad-fire, 4850 X2 vs. 4870 X2,
4850 X2 + 4870 Tri-fire, 4850 X2 + 4870 X2 Quad-fire, 4850 + 4870 Crossfire,
(NEW) 4890 vs. 4850 X2

Steam nick: rico2001

"The ability to learn faster than your competitors may be only sustainable competitive advantage." ~Arie de Geus

almighty15

Why don't you run the tests again only run Rygels texture pack and see what happens...

__________________
Core 2 Duo E6420 @ 3.9Ghz : http://valid.canardpc.com/show_oc.php?id=558967

Pentium Dual Core E6300 - 4.2Ghz : http://valid.canardpc.com/show_oc.php?id=646455

Gigabyte EP45 Extreme FOR SALE.....PM for details