Hey everyone! I just upgraded to some CL30 RAM and a 4TB SN850X NVMe, plus a 9070 XT. I'm now in the market for a new motherboard to replace my old ROG Crosshair VI Hero, which I've had for 7 years. Here's what I'm looking for: I need at least two PCIe 16x slots that are spaced far enough apart to accommodate my current Sapphire Nitro+ 9070 XT and a second GPU down the line. Additionally, I want 4 RAM slots since I have 96GB on the way and plan to upgrade to 192GB in the future. I've heard that finding motherboards with the right PCIe spacing can be tricky, especially since my current GPU is quite large. I have a preference for Asus motherboards but I've also had good experiences with MSI in the past. I primarily use Linux (Ubuntu) for work, so overclocking isn't essential for me, though it seems most boards support it nowadays. One motherboard I found that seems suitable is the ASUS ROG Crosshair X870E Hero, but I'm open to other suggestions. Thanks!
4 Answers
Most boards these days don't truly support dual x16 PCIe for multi-GPU setups, which are pretty rare nowadays anyway. The Crosshair X870E Hero is a solid choice, albeit on the pricier side. You could spend more and get even fancy boards, but I'm not sure they offer much more value for your needs.
If you need good bandwidth for a second GPU, some Asus ProArt boards provide x8/x8 bifurcation, which could be what you're looking for. However, double-check the PCIe spacing; it can vary across models. The Crosshair Hero mentions having "2x PCIe 5.0 x16 slots," which sounds promising!
You're probably going to face some challenges with DDR5 if you plan to populate all 4 RAM slots, especially at speeds above 4800MHz. For heavy usage like multiple VMs, Threadripper might be a better option since it has more PCIe lanes and supports larger memory configurations better than standard boards.
I get that! Since I need the RAM mainly for running VMs and not gaming, do you think the speed drop would affect performance noticeably?
True, I'm not aiming for the highest-end setup. Just want to ensure my second GPU for local LLMs will run smoothly.