So, I am on the fence of upgrading from 11th gen to 13th gen Intel. But the sales keep looking better and better. My needs are shoving 5 m.2 into the mobo whether it be via add-in card or onboard, with thunderbolt, onboard or headers for add-in. Most of the motherboards decently priced run the secondary/tertiary pci slots at x4 which limits me on nvme add-in cards if there are not enough on-board. At least what I have seen. The boards that check all the boxes are +$500, maybe there was one that was $400. Not worried about the z790 chipset bandwidth. I also have a 3080 TI.
Stumbled across this board, it checks all the boxes and was very surprised by the price. $300 is still pretty steep. I feel like there is a catch here. What do you y’all think chat.
GIGABYTE Z790 AERO G LGA 1700 Intel Z790 ATX Motherboard with DDR5, 5* 5.0 M.2, PCIe 5.0, USB 3.2 Gen2X2 Type-C, WiFi 6E, Intel 2.5GbE LAN
I see 3x physical 16 lane PCIe slots, 5x m.2 slots. I know that the CPU only supports 20 lanes of PCIe connectivity and immediately realize that not all of these can be connected to the CPU. What that means is that there will be compromises:
not full connectivity (e.g. only 4 lanes of PCIe electrically connected on the 16 lane slot),
switched connectivity (if one devices is connected another is turned off), or
shared bandwidth (typically through the chipset connection to the CPU).
Now, I can look for these compromises. Some are easy to spot (on the specs website), others - like switched connectivity - are often only documented in the manual and in footnotes.
The top slot is attached from the CPU at x16 (or x8 if M2C_CPU is populated); M2A_CPU and M2C_CPU are attached from the CPU. So you have either a x16/x4 or a x8/x4/x4 configuration.
Everything else attached from the PCH, which is less of a problem on H670/H770/Z690/Z790 than it is on prior generations and AMD platforms because the PCH gets [the equivalent of] a Gen4 x8 uplink.
Nope, I think you’re good to go. The only potential “catch” I can see is that the top slot drops down from x16 to x8 when M2C_CPU is populated, but M2C_CPU only consumes four lanes. So there are an additional four lanes that simply route nowhere in this configuration, which is slightly irksome.
Consumer mobos all work in similar fashion. It’s good to understand the limitations of mobo, especially since op’s needs are shoving 5 m.2 into the mobo.
