Hi Guys and Girls
I have a problem which might be nothing, not sure. This is the first crossfire machine I have built, I have a gigabyte z77m-d3h motherboard and have x2 rx470's running in crossfire in it - when checking with gpu-z I see the second card is reporting PCIex16@x4 whilst the first is reporting @x16?
Is there something not working right? Do I lose performance?
Thank you for the help, below is image of stress test done showing clearly 1 at x 16 and 1 at x 4:
I appreciate your time and sorry if someone has asked this before I did do a search and nothing relevant came up.
Solved! Go to Solution.
Your motherboard is rated for x16/x4, so there is no problem.
Thank you, will this affect performance much? I am getting 2 rx580 8gigs cheap because of crypto crash and wondering if it makes big diff?
The problem is that it's 2.0 x4, which is really slow for a modern mainstream graphics card, alone much less in a multiple GPU setup, and not 3.0 x4. Even worse PCIe 2.0 x4 only provides 2GB/s bandwidth which isn't tested today by reputable review sites. I did find a forum post comparing a GTX 1060, a card of equal power to the RX 580, on PCIe 3.0 x16, Thunderbolt 3, 2, and 1. PCIe 2.0 x4 provides the same bandwidth as Thunderbolt 2, so you can make a judgement. Single card you're losing about 17% performance, but add in the multiple GPU crosstalk and the loss will be higher. It's not a direct apples to apples comparison, but it is something to look at.
Personally I would not do it, I'd get a single GTX 1070 Ti or a Vega 56, because your baseline performance will vary between 0 and 100% depending on the game, and multiple GPUs introduce artifacts not present in single GPU setups, such as microstuttering. Personally too I'd look to upgrade your system to something a bit more modern than 2012 before I looked at just updating the GPU, especially since you're only likely to see around 30% real performance improvement.
https://egpu.io/forums/mac-setup/pcie-slot-dgpu-vs-thunderbolt-3-egpu-internal-display-test/