Question Which MB and why?

[instawookie]

I've been looking at these two MB's for a hot minute to pair with my Meg 700R case. But I can't decide on which, I'm leaning towards Carbon Wifi 2 judging by the way the PCI-E lanes work. Price wise they're pretty close currently.

https://us.msi.com/Motherboard/MPG-Z790-CARBON-WIFI-II

https://us.msi.com/Motherboard/MEG-Z790-ACE

 

[Aeacus]

Just compare the two,
link: https://www.msi.com/Motherboards/Pr...LVo3OTAtQUNF,TVBHLVo3OTAtQ0FSQk9OLVdJRkktSUk=

I'm leaning towards Carbon Wifi 2 judging by the way the PCI-E lanes work.

If you want less lanes on PCI-E x16 slots and slower bandwidth, then sure, go for Carbon Wifi II.

 

[instawookie]

Just compare the two,
link: https://www.msi.com/Motherboards/Pr...LVo3OTAtQUNF,TVBHLVo3OTAtQ0FSQk9OLVdJRkktSUk=


If you want less lanes on PCI-E x16 slots and slower bandwidth, then sure, go for Carbon Wifi II.

Bwhahaha , woops... I think when I did my comparison, I forgot which board was left and right. But none the less that's exactly what I was looking at as far as differences that stood out.

I think this also threw me for a loop too on the ace board.

"PCI_E1 slot will run at x8 speed when installing device in the PCI_E2 slot or M2_4 slot.
**M2_4 slot will be unavailable when installing device in the PCI_E2 slot."

 

[Aeacus]

I think this also threw me for a loop too on the ace board.

"PCI_E1 slot will run at x8 speed when installing device in the PCI_E2 slot or M2_4 slot.
**M2_4 slot will be unavailable when installing device in the PCI_E2 slot."

Well, since both PCI_E1 and PCI_E2 are using CPU PCI-E lanes, thus providing PCI-E 5.0 speeds, it is normal that some of those PCI-E lanes are shared with M.2 slot. Since some M.2 slots also use CPU PCI-E lanes and CPUs have limited number of lanes. Now, if you'd run Xeon CPU, with plethora of PCI-E lanes, it would be different story, but consumer CPUs have limited number of PCI-E lanes.

Then, there is the question if you'd be using more than one PCI-E x16 slot? One slot for sure is reserved for GPU. But other two? I doubt you'd run dual-GPU, since SLI/CrossFire is long dead. This would leave M.2_4 free to use and utilize fast CPU PCI-E lanes.

 

[instawookie]

Bwhahaha , woops... I think when I did my comparison, I forgot which board was left and right. But none the less that's exactly what I was looking at as far as differences that stood out.

Well, since both PCI_E1 and PCI_E2 are using CPU PCI-E lanes, thus providing PCI-E 5.0 speeds, it is normal that some of those PCI-E lanes are shared with M.2 slot. Since some M.2 slots also use CPU PCI-E lanes and CPUs have limited number of lanes. Now, if you'd run Xeon CPU, with plethora of PCI-E lanes, it would be different story, but consumer CPUs have limited number of PCI-E lanes.

Then, there is the question if you'd be using more than one PCI-E x16 slot? One slot for sure is reserved for GPU. But other two? I doubt you'd run dual-GPU, since SLI/CrossFire is long dead. This would leave M.2_4 free to use and utilize fast CPU PCI-E lanes.

Yea, I'm just using a singular 4090 w/ a 14700k and currently got 4 M.2 drives. 970 Evo for windows and 3x 980 Pro M.2s for games and movies. I think I was misunderstanding this mostly PCI_E1 slot will run at x8 speed when installing device in the PCI_E2 slot or M2_4 slot, In the scenario I use all of my current M.2's I would reduce speed of the GPU because I would end up utilizing slot #4 for the m.2's.

 

[Aeacus]

In the scenario I use all of my current M.2's I would reduce speed of the GPU because I would end up utilizing slot #4 for the m.2's.

Even when RTX 4090 runs in x8 mode, there is negligible diff in performance.

TechPowerUp did in-depth testing and RTX 4090, which usually runs PCI-E 4.0 x16, looses, at most 3% performance (1080p) or 2% (1440p/4K), when using it either PCI-E 3.0 x16 or PCI-E 4.0 x8 (since bandwidth is same between both),
testing: https://www.techpowerup.com/review/nvidia-geforce-rtx-4090-pci-express-scaling/28.html

As of why it only looses so little, well, RTX 4090 does not utilize the full bandwidth offered by PCI-E 4.0 x16. There is a lot of bandwidth headroom, which, one day, is used by better GPUs (e.g RTX 50- and 60-series).

 

[instawookie]

Even when RTX 4090 runs in x8 mode, there is negligible diff in performance.

TechPowerUp did in-depth testing and RTX 4090, which usually runs PCI-E 4.0 x16, looses, at most 3% performance (1080p) or 2% (1440p/4K), when using it either PCI-E 3.0 x16 or PCI-E 4.0 x8 (since bandwidth is same between both),
testing: https://www.techpowerup.com/review/nvidia-geforce-rtx-4090-pci-express-scaling/28.html

As of why it only looses so little, well, RTX 4090 does not utilize the full bandwidth offered by PCI-E 4.0 x16. There is a lot of bandwidth headroom, which, one day, is used by better GPUs (e.g RTX 50- and 60-series).

Thanks for the help good to know, I'll mostly just discard one or two of the M.2's and replace with a larger.