Intel xeon or Ryzen 5?

[Tech1diot]

Hello,
I have an Intel Xeon 2650-v3, which is a 10-core 20-thread CPU. I also have a spare 2400g, a 4-core 8-thread. I'm going to be adding another system to my homelab, but I'm trying to decide which to use. Motherboards for both (2011-3/x99 and am4) are about the same price (~60$), both use DDR4 ram, which I have on hand (4x 8gb 2666mhz cl19). Both use NVME and data, PCIe Gen 3, etc. The only difference between them is power draw and performance; from what I've seen on Geekbench, the 2400g is about 65% slower in most tasks (I know it's a benchmark, but I use it as a baseline). The 2650-v3 draws like 105 watts and the 2400g draws about 65. I'm trying to decide whether I want more cores and multithreading abilities at the cost of potentially a less stable environment (the Chinese-hacker-style X99 boards aren't always 100% stable, especially using non-ECC memory), or if I want a more efficient CPU with a more stable environment at the cost of performance.
Task-wise its all over the place. I'm planning on running stuff like Home Assistant, language models, Opnsense, VMs, stuff like that.
Any advice would be greatly appreciated.
(I have a compute GPU on hand already)

[Tetras]

To be brutal, I think you'd be better off selling them and getting a 12100F or 12400F, along with a cheap H610 DDR4 board.

[Timme]

edit: a good comparison 

[porina]

The power you mention is the TDP, and doesn't necessarily represent a particular load. Unless you're running them 100% all the time the actual power draw for both will be much lower. Ryzen 65W TDP CPUs typically limit at 88W. Intel CPUs, depends on mobo settings, but I'd expect server boards to be closer to TDP than on enthusiast ones. If it wasn't for needing to gamble on a Chinese X99 board that would have been the way I'd go of the two. 2400G, I'm not even sure without looking it up if it is Zen or Zen+, but either way they were awful if you ever need the FPU. Cinebench R15 was about all they were good at.

[Timme]

17 minutes ago, Tech1diot said:

Hello,
I have an Intel Xeon 2650-v3, which is a 10-core 20-thread CPU. I also have a spare 2400g, a 4-core 8-thread. I'm going to be adding another system to my homelab, but I'm trying to decide which to use. Motherboards for both (2011-3/x99 and am4) are about the same price (~217.44₪ (60$)), both use DDR4 ram, which I have on hand (4x 8gb 2666mhz cl19). Both use NVME and data, PCIe Gen 3, etc. The only difference between them is power draw and performance; from what I've seen on Geekbench, the 2400g is about 65% slower in most tasks (I know it's a benchmark, but I use it as a baseline). The 2650-v3 draws like 105 watts and the 2400g draws about 65. I'm trying to decide whether I want more cores and multithreading abilities at the cost of potentially a less stable environment (the Chinese-hacker-style X99 boards aren't always 100% stable, especially using non-ECC memory), or if I want a more efficient CPU with a more stable environment at the cost of performance.
Task-wise its all over the place. I'm planning on running stuff like Home Assistant, language models, Opnsense, VMs, stuff like that.
Any advice would be greatly appreciated.
(I have a compute GPU on hand already)

Please, remove the solution tag before you've read the edited post:)

[Tetras]

20 minutes ago, Timme said:

edit: a good comparison 

I generally use PassMark for comparing older CPUs, pretty rough, but eh.

 

(CPU, single thread, multithread)

E5-2650 v3: 1640, 11747
Ryzen 5 2400G: 2145, 8737
i3-12100 (F would be the same): 3272, 12743
i5-12400F: 3485, 19468

 

Soo, basically, it suggests the 12100F would have the same multithread as the Xeon, but almost double the single thread, while the 12400F would pretty much double both.

 

Usually there's a good spread of single and multi threaded apps on a server, but it depends. If need to assign individual cores to a VM and there's little workload, the Xeon obviously has more of them.

[Timme]

7 minutes ago, Tetras said:

I generally use PassMark for comparing older CPUs, pretty rough, but eh.

 

(CPU, single thread, multithread)

E5-2650 v3: 1640, 11747
Ryzen 5 2400G: 2145, 8737
i3-12100 (F would be the same): 3272, 12743
i5-12400F: 3485, 19468

 

Soo, basically, it suggests the 12100F would have the same multithread as the Xeon, but almost double the single thread, while the 12400F would pretty much double both.

 

Usually there's a good spread of single and multi threaded apps on a server, but it depends. If need to assign individual cores to a VM and there's little workload, the Xeon obviously has more of them.

This is what 10 years of tech advancement does.