Will my old motherboard hold back a graphics card upgrade?

[Athawolfus]

Hey all,

I am new on this forum, so I decided to let things start with a question:

I'm thinking about getting a new high-end graphics card, something from the Radeon HD 6000 or 5000 or Geforce 500 series perhaps...

So I have a Gigabyte GA-P35-DS3R motherboard and I know that it doesn't support PCI-E 2.0 x16 bus interface. All these new cards specify that a PCI-E 2.0 port is required and I've even seen a specification for a PCI-E v2.1 port.

So does that mean that if I get such a new card and plug it into my motherboard, it will just not run? Or will it be recognized and run but with a significant drop in overall performance?

Please don't suggest that I get a new motherboard. I'm planning to do that sometime later but now have money saved up for a good graphics card upgrade. What should I expect given my current motherboard? My processor is a Core 2 Duo E8400 and I currently have a Geforce 8800 GTX.

Thanks.

 

[bikeracer4487]

Well, physically, a PCI-E 2.0 card will work fine in a 1.1 slot. Unfortunately, there haven't been any recent articles I could find detailing the performance hit of using a 1.1 slot. Tom's did an article a while back that you can find here: http://www.tomshardware.com/reviews/pci-express-2.0,1915-10.html

If you look at the results, the x8 results would represent a PCIE 1.1 slot. As you can see, there is almost no performance hit by halving the bandwidth. The problem with this article however is that the 9800 GX2 being used to test represents the flagship card at the time, which is about equivalent to a GTX 460, so I can't say with any real certainty what kind of performance hit a higher end card like a 570 or 6950 would experience, but if we extrapolate I wouldn't expect it to be any higher than 5 fps or so...but again, that's just an educated guess.

EDIT: Remembered an article that wasn't about PCI-Express 1.1 but was rather about the different lane-dividing strategies of motherboard for multi-GPU setups, but it directly applies to your question, you can find it here: http://www.tomshardware.com/reviews/pcie-geforce-gtx-480-x16-x8-x4,2696.html

Again, the x8 lanes would represent the PCI-E 1.1 performance. Anyways, this article uses a GTX 480 which is MUCH closer to the kind of performance you'd get with a high-end upgrade...in fact, you shouldn't get much higher unless you go with a 580 (which is still only like a 10% increase) or a multi-GPU setup. Anyways, the only game in the article that demonstrated any real performance difference was Modern Warfare 2, and that was at framerates well above 100 fps for the x8 lanes. So, all that said, I think I can safely say that the performance hit by using a 1.1 interface would be marginal at best, happy upgrading!

 

[Athawolfus]

OK, thank you very much, friend. Under the present circumstances, I think I'll just go for the upgrade. I was afraid of a big performance hit because of all the "specifications" but now this doesn't seem to be the case.

I'm convinced now but to get one thing certain: you said to check under PCI x8 results. So would these be the same as the results from a PCIE v1.0 x16 slot? Thanks.

 

[bikeracer4487]

OK, thank you very much, friend. Under the present circumstances, I think I'll just go for the upgrade. I was afraid of a big performance hit because of all the "specifications" but now this doesn't seem to be the case.

I'm convinced now but to get one thing certain: you said to check under PCI x8 results. So would these be the same as the results from a PCIE v1.0 x16 slot? Thanks.

Yeah, the thing with PCI-Express 2.0 is that it exactly doubled the bandwith of PCI-E 1. So by halving the lanes in PCI-E we end up with the same bandwidth as PCI-E 1.0, so it's possible to pretty much exactly duplicate PCI-E 1.0 performance. So, in conclusion: PCIE 1.0 x16 = PCIE 2.0 x8

 

[Athawolfus]

Best answer selected by Athawolfus.

 

[Maziar]

This topic has been closed by Maziar