slower PCIE bus

[ajslave]

i have this::::

AMD Athlon 5200+(Brisbane)
Gigabyte nForce 405/6100
4GB DDR2 800Mhz
XFX 8800GT Zalman
Vista Ultimate x64

Crysis gives 20-23fps average on dX10, at 1280x960, no AA, everything very high.
STALKER gives ~100fps at 1280x960, all settings max, full AA.

the mobo supports on;y PCIe 8X, so will upgrading to a better PCIe 16x motherboard gimme a better frame rate??

[duplo83]

It will be a very small increase, if any, by switching to a board with a 16x bus speed. Crysis is just too demanding. In my opinion, I think Crytek dropped the ball on it's optimization. When three GTX 280 cards cannot get above 60 FPS @ 1280x1024, even with all the settings to 'Very High,' we should all be worried.

http://guru3d.com/article/geforce-gtx-2 ... ew-test/13

[martini161]

in terms of benifets resulting directly from the higher bus, they will be minalmal. but in terms of improvments resulting from a newer, better motherboard, there will be some, though it may not come in the form of fps

[GI-JOE]

One of the most beneficial of 16x is when your video card has only 256mb of memory or less(faster access to system ram). Other than that, there is slight difference.

[Zertz]

One of the most beneficial of 16x is when your video card has only 256mb of memory or less(faster access to system ram). Other than that, there is slight difference.

I'm pretty the bottleneck in that situation is not the PCI-E interface.

For single cards, there's like... no visible gain. You do have an entry-level board so like martini said, there will be some gains will a newer and better one.

[GI-JOE]

One of the most beneficial of 16x is when your video card has only 256mb of memory or less(faster access to system ram). Other than that, there is slight difference.

I'm pretty the bottleneck in that situation is not the PCI-E interface.

For single cards, there's like... no visible gain. You do have an entry-level board so like martini said, there will be some gains will a newer and better one.

Now that you mention it, I think I got some info mixed up that I read that had to do with pic-e 2.0. I'm pretty sure I read that the increased bandwidth would help graphics cards with less vRam?

[Zertz]

One of the most beneficial of 16x is when your video card has only 256mb of memory or less(faster access to system ram). Other than that, there is slight difference.

I'm pretty the bottleneck in that situation is not the PCI-E interface.

For single cards, there's like... no visible gain. You do have an entry-level board so like martini said, there will be some gains will a newer and better one.

Now that you mention it, I think I got some info mixed up that I read that had to do with pic-e 2.0. I'm pretty sure I read that the increased bandwidth would help graphics cards with less vRam?

I highly doubt it, the problem with shared video memory is the CPU -> RAM bandwidth which is already shared amongst whatever is running. Basically, PCI-E 2.0 doubles the bandwidth per lane compared to PCI-E 1.x along with other technical changes of course. The thing is, no card can saturate a 16x PCI-E 1.x, especially not the lower end cards that share memory so bandwidth from the GPU to the CPU isn't an issue.

It's just my theory based on what I know, if you have a link to confirm your info, I'll gladly read it!

[Methious]

I've been dealing with PCI-E 1.0/2.0 since it came out.

The latest round starts today with a new Gigabyte premium p45 board with a water cooled chip set.

Previously best I could get out of my 8800gt on PCIE 2.0 was 12500. Needed a ddr3 board and could only afford an Asus board with PCI-E 1.0, ended up swapping out the DDR3 for Reaper ddr2 800 for OCing reasons (NB was smokin hot, dual ram board). Long story shorter I hit 14999 on the 8800gt with PCIE 1.0 and ddr2 800 running at 800.

The new technology on a new generation board might be making a difference but I don't think the PCIE bus is making much difference yet. In the process of migrating to the new board for funsies I'm going to leave the whole system the same except the Mobo and rerun the tests with the Quad I got 14999 on and see what happens.

I think a lot of the results are dependent on the rigs setup. I noticed the largest increase in performance across the board when I spent a day and a half getting the holy grail of OCing hitting 3.2 on the q6600, with the FSB at 400 and Multiplier of 8 letting the ddr2 800 run at a 1:1 ratio with the FSB. That's when scores started jumping dramatically.

With the 8x bus you'll probably see more performance increase from the newer motherboard components than the 16x bus on the pcie. I'd still go PCIE 2.0 if I could because GPUs are making leaps and bounds and the newest ATI cards are going to GDDR5 and I don't know how that's affecting bandwidth yet.

[GI-JOE]

I'll have to find that article I read, can't right now since I am busy at work. Although it is very possible I'm confusing it with something else I read since I read a lot of reviews/articles etc.

Edit: Found it

But is PCIe 2.0 really necessary yet? As long as a graphics solution can operate with data that is stored within its local video frame buffer memory, both the reasonably mainstream Radeon HD 3850 and the hardcore GeForce 9900 GX2 will operate close to their maximum performance, even if the PCI Express link width is limited to x8 or x4. Once larger textures need to be accessed, as is the case in Crysis http://en.wikipedia.org/wiki/Crysis or Microsoft’s Flight Simulator http://en.wikipedia.org/wiki/Flight_simulator X, interface bandwidth becomes a crucial element. Any link width below x16 will noticeably limit these games’ playability.

Source: http://www.tomshardware.com/reviews/pci ... 15-13.html

Note the part of larger textures

[DMB2000uk]

Below 16x will hinder it, but going above doesn't make much of a difference (yet).

Dan