Log Crazy SFF NAS | 29 SSD inside Fractal Design Ridge | 16C 32T EPYC | 128 RAM

huday

Minimal Tinkerer
Original poster
New User
Jun 13, 2024
3
7
Hello everyone,

I just finished build it, thoughts?

Hardware:
SFF: Fractal Design Ridge
CPU: AMD EPYC 7282 16C 32T
RAM: A-Tech 128GB (4x32GB) DDR4 3200MHz PC4-25600 ECC RDIMM
MB: ASRock Rack ROMED4ID-2T (Deep Mini-ITX)
PSU: FX600 Platinum Cybenetics Platinum 600W (Flex ATX)
STORAGE: 16x SATA SSD Patriot P210 128GB SATA, 13x NVMe M.2 SSD Patriot P300

Future plans:
1TB RAM
8TB each SSD*29 = 232TB
AMD EPYC 7773X 64C 128T

Thanks for your time.

 

huday

Minimal Tinkerer
Original poster
New User
Jun 13, 2024
3
7
btw, interesting card has been released NVIDIA RTX 4000 SFF Ada Generation, length is only 168 mm, I am thinking maybe to put it here instead of four M.2 PCIe x16 extension card?
 

huday

Minimal Tinkerer
Original poster
New User
Jun 13, 2024
3
7
This is madness lol I got competition!
Amazing work 🔥
Thank you both, I appreciate it! 😊 mknrls, I saw your build - you're using the same motherboard as I am. Fascinating work. Initially, I considered building a liquid-cooled setup for both the CPU and GPU, using something like the RTX A5000 or A6000. I've done something similar in the past inside Fractal Design Node 202 with less powerful hardware (you can see it here: link). However, I realized that with all these PCI lanes, it's possible to install bunch of disks. My current idea is to configure a TrueNAS or maybe an Unraid server, and then build a second server with liquid cooling for the CPU and GPU. This second server will have a hypervisor installed and host virtual machines with vGPUs from the A5000 or A6000, without any physical disks attached, as it will utilize all necessary storage from the NAS.
 
  • Like
Reactions: mknrls

mknrls

Caliper Novice
Oct 12, 2019
32
51
btw, interesting card has been released NVIDIA RTX 4000 SFF Ada Generation, length is only 168 mm, I am thinking maybe to put it here instead of four M.2 PCIe x16 extension card?
not a bad idea, you might be able to fit both depending on spaces