Hi all, I realize this question has likely been posed before, but I've found conflicting answers so far and am wonder if anyone else has any experience relating to this specific topic.
Now, the current generation of PCIe is 3.0, and it is usually recommended that you install it in the x16 slot. Most motherboards nowadays come with at least 2 PCIe 3.0 x16 slots and support crossfire at the very least, the more expensive ones supporting SLI.
Now for my question/theory. It seems most of the inexpensive motherboards come with a PCIe 3.0 x16 slot, and a PCIe 3.0 x4 slot (Physically x16, but not fully utilized). This x4 slot is what has me worried, because that's 1/4th of the bandwidth of a standard speed slot. If you were to utilize two high-end GPU's in a motherboard such as this in crossfire, would you see a performance drop? If so how much?
If my understanding of PCIe generations is correct, the revisions have doubled the bandwidth rate every time, therefore:
PCIe 3.0 x16 = PCIe 2.0 x 32 (theoretical)
PCIe 3.0 x8 = PCIe 2.0 x16
PCIe 3.0 x4 = PCIe 1.0 x16
Most of the benchmarks I've found comparing these two showed very little performance drop between the differing speeds, but most of the data was collected years ago (2010 - 2012). Have the recent improvements to graphics cards in recent years, such as Polaris and Pascal, made these slower PCI slots useless for crossfire utilization?
Now, the current generation of PCIe is 3.0, and it is usually recommended that you install it in the x16 slot. Most motherboards nowadays come with at least 2 PCIe 3.0 x16 slots and support crossfire at the very least, the more expensive ones supporting SLI.
Now for my question/theory. It seems most of the inexpensive motherboards come with a PCIe 3.0 x16 slot, and a PCIe 3.0 x4 slot (Physically x16, but not fully utilized). This x4 slot is what has me worried, because that's 1/4th of the bandwidth of a standard speed slot. If you were to utilize two high-end GPU's in a motherboard such as this in crossfire, would you see a performance drop? If so how much?
If my understanding of PCIe generations is correct, the revisions have doubled the bandwidth rate every time, therefore:
PCIe 3.0 x16 = PCIe 2.0 x 32 (theoretical)
PCIe 3.0 x8 = PCIe 2.0 x16
PCIe 3.0 x4 = PCIe 1.0 x16
Most of the benchmarks I've found comparing these two showed very little performance drop between the differing speeds, but most of the data was collected years ago (2010 - 2012). Have the recent improvements to graphics cards in recent years, such as Polaris and Pascal, made these slower PCI slots useless for crossfire utilization?