Hello, I want to standardize my home servers and reduce them to 3 Proxmox computers. 2x a Tiny server and a slightly more powerful one for AI (ollama/open webui and deepseek-r1-70b, CPU based only, no GPU).
For the more powerful server, I am wavering between 2 processors: i9-10940X vs. i9-14900KS.
i9-10940X
- 14 Cores (3,30-4,8 GHz == 67,2 GHz)
- 28 Threads
- Quadro-Channel DDR4-2933 (PC4-23466, 93.9GB/s)
i9-14900KS
- 24 Cores (8Power+16Economy - 2,40-6,2 GHz == 117,6GHz)
- 32 Threads
- Dual-Channel DDR5-5600 (PC5-44800, 89.6GB/s)
I don’t like the Idea of the Power/Economy-Cores… And the newer i9 has only dual-channel for RAM instead of quad. But it has double of GHz.
Which is better for my solution? I also want a relative low idle power consumption.
thank you all!
I can’t speak to AI performance, but given you’re stated goal of lower idle power consumption, I’d go with the 14900K, not the KS as you have listed.
Reason being the $250 price difference between the two, when the KS is just a slightly higher binning of the K with an additional 200MHz on the boost clocks. With that higher boost being something you’re unlikely to practically see without a substantial and robust cooling system, I don’t think it’s worth the extra money.
The reason I’d go with the K over the 10940X is the lower limit on it’s power consumption. The E cores are very efficient and can down clock substantially meaning it idles at really low power. The 10940X doesn’t have that benefit.
Beyond that, I’d say look at IPC, per thread, per max sustainable clock of each core, to get a general out look on performance.
Note: all of the above assumes we’re working within your listed options. My actual recommendation would be an AMD 7800x3d or 9800x3d.
Thank you! The AMD-Route sounds also promising, but I’m not sure about there idle power consumption. they say that intle is bettle in idle mode. But i’m not sure if proxmox can handle the E-cores properly.
Might be a bit late on this, but ProxMox doesn’t really handle assigning threads to the e/p cores. That’s handled by the kernel and as long you’re running kernel version 6.1 or greater you should be good on that front.
If you really need to, you can also pin specific VMs to specific cores. So that if you’ve got something that always needs the performance it can always run on the p-cores and things that aren’t as demanding can always run on e-cores.
That said, especially if you’re over provisioning, it’s probably better to let the scheduler in the kernel handle thread assignments.
“They say”, but they’re right. Ryzen chips do have worse idle power usage, but you’re talking about 10w or so, at most.
And uh, if you were looking at an X-series CPU, I can’t see how that 10w is a dealbreaker, because you were already looking at a shockingly inefficient chip.