x16 is theoretically twice as fast as x8. At PCIe 3.0, each lane is a little under 1,000 MB/s. So at x8 (3.0) you'd get maximum speeds a little under 8,000 MB/s, versus maximum speeds a little under 16,000 MB/s at x16 (3.0). In a real life scenario, I'm not sure it would make a large or noticeable difference in performance.
Yeah, and welcome to the age of 4th generation PCIe (PCIe 4.0). The Aurora R7/8 support up to PCIe 3.0. The limiting factor at issue is that the two GPU slots only have 8 lanes. If you were to buy a 16x PCIe 2.0 GPU (and for high end GPU's, I think that would be circa the time of the GTX 590 or 690) the bandwidth is still limited to the 8 lanes supplied by the R7/8 motherboard. So in that scenario, the speed of (8x PCIe 2.0) is the same as (4x PCIe 3.0), in other words, 25% the theoretical limit of a non-neutered 16x PCIe 3.0 2080ti. Of course, in a real life scenario, the theoretical speed limit of 16x vs 8x is unlikely to be the limiting factor on performance.
r72019
6 Professor
•
5.3K Posts
1
August 6th, 2019 12:00
speedstep
9 Legend
•
47K Posts
0
August 6th, 2019 17:00
X8 pci-e 3.0 is the same speed
as X16 pci-e 2.0 so it depends on the card.
r72019
6 Professor
•
5.3K Posts
0
August 6th, 2019 20:00
Yeah, and welcome to the age of 4th generation PCIe (PCIe 4.0). The Aurora R7/8 support up to PCIe 3.0. The limiting factor at issue is that the two GPU slots only have 8 lanes. If you were to buy a 16x PCIe 2.0 GPU (and for high end GPU's, I think that would be circa the time of the GTX 590 or 690) the bandwidth is still limited to the 8 lanes supplied by the R7/8 motherboard. So in that scenario, the speed of (8x PCIe 2.0) is the same as (4x PCIe 3.0), in other words, 25% the theoretical limit of a non-neutered 16x PCIe 3.0 2080ti. Of course, in a real life scenario, the theoretical speed limit of 16x vs 8x is unlikely to be the limiting factor on performance.