Folding Forum

I just bought this card for a folding rig: http://www.newegg.com/Product/Product.a ... =u00000687...and have an old PC around I was planning to use it with. It is a Core 2 Duo on a socket 775 motherboard. I believe the motherboard FSB speed is 1066mhz, and the max ram speed is DDR2 800mhz. I'm just wondering if this will negatively impact folding performance?

Will the card itself put out the same amount of work per day as it would with a faster PC? Or will a system with slower motherboard/RAM not be able to keep up with a newer DDR5 card? I would hate to have the card only producing a fraction of the PPD it is capable of due to being placed on an older/slower motherboard. If necessary, I will grab a budget i3 mobo/CPU/DDR3 combo, but would rather not waste the money if it won't make a difference. Any ideas?

Assuming that the motherboard has PCI-E 2.0 lane, the performance loss can vary from insignificant to ~10%. I would suggest that you try it out and post your results here so that others with similar GPU can provide some data for comparison. Please ensure that the client-type is set to advanced so that your GPU will be assigned FahCore_17 WUs which are very efficient on AMD GPUs when compared to FahCore_16 WUs.

I just did a similar swap from an i7 SB to C2 quad on 775 socket. Why ? Because the newer board was not providing enough PCIe slots for my two GPU.
The old board is an Asus P5Q deluxe with three PCIe slot. I have one GTX 780 in the 2.0 slot and a 660 Ti in second slot. Both PCIe 2.0. I have a CPU slot configured but paused to give enough CPU power to support two GPU and the system itself.
Solid 200kPPD out of the system with Ubuntu as OS. Very happy camper.
The i7 is now benched and sometimes started for some dedicated CPU folding.

Any motherboard with PCIe 2.0 x16 slots is not considered old for GPU folding. It performs just as fast in fah as anything you can buy today.

Less than 2.0, and or less than x8 slot can slow things down a little.

I have a GTX 760 in a x1 slot making 50 k PPD instead of 60 k. It takes a lot to slow down a GPU.

Thanks yall..That is just what I was hoping to hear.

I do GPU folding on two older machines, one on a Socket 939/nForce 4 motherboard (PCIe 1.0a), the other on a LGA775/Intel P35 motherboard (PCIe 1.1). The two GPUs I've used with them are a GTX 460 1 GB, and a GTX 570. I've recently switched the GPUs between the machines, and they do roughly the same PPD as before, with the GPUs at mostly 99% load. The PPD I get with them appears to be comparable to what others get on newer machines.

PCIe 1.0a and 1.1 both transfer data at up to 250 MB/s per lane. PCIe 2.x can transfer data at up to 500 MB/s per lane so you can mentally reduce the speed to half of what others are talking about. Sixteen lanes of PCIe 2 has been shown to be more than is needed. Reducing it to 8x is almost insignificant and even 4x or less still represents a rather small speed reduction.

I think the point here is that you should use what you have and not worry about it.

PCIe 2.0 at x8 (no speed reduction) equals PCIe 1.0 at x16 in bandwidth, so again, no speed reduction seen.

PCIe bandwidth is clearly not the gating factor in GPU performance until you make it very tight, like in my x1 slot.

bruce wrote:PCIe 1.0a and 1.1 both transfer data at up to 250 MB/s per lane. PCIe 2.x can transfer data at up to 500 MB/s per lane so you can mentally reduce the speed to half of what others are talking about. Sixteen lanes of PCIe 2 has been shown to be more than is needed. Reducing it to 8x is almost insignificant and even 4x or less still represents a rather small speed reduction.

I think the point here is that you should use what you have and not worry about it.

I'm aware of the speed differences between the versions, I listed them more as a compatibility note for the OP and others as compatibility issues occasionally crop up when newer video cards are mixed with older chipsets/PCIe revs.