[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: how many W a PSU for non-gaming Debian?



Emanuel Berg writes:

Linux-Fan wrote:

> Please keep the following points in mind when doing PSU
> wattage sizing for modern PCs:
>
> - Judging a CPU by its thermal design power is no longer
>   feasible due to some CPUs permanently overclocking while
>   the actually available cooling power permits it. On some
>   Intel CPUs this can mean about twice the power than you
>   would have expected. If we were to apply this logic
>   directly to the unspecified (?) AMD CPU from the OP's
>   config

Oh, the OP has the AMD4 x86_64 CPU that comes with/in the Asus
ROG Strix B450-F Gaming motherboard!

By default, a motherboard is just that, a motherboard. Unless you have some specific "bundle" package, there is no CPU included with it. According to
https://rog.asus.com/motherboards/rog-strix/rog-strix-b450-f-gaming-model/
the board has "AM4 socket: Ready for AMD Ryzen(TM) processors". And according to https://en.wikipedia.org/wiki/Socket_AM4 this socket can accomodate for a wide variety of different processors.

> - 80+ certified PSUs are rated in terms of their performance
>   at certain load percentages. If you choose a high-power
>   PSU (e.g. 600W) then even if it has a high efficiency
>   according to 80+ it will not necessarily be more efficient
>   than a less highly rated 300W model.

Not following?

You've snipped the part by Andy Cater:

| A larger PSU in wattage terms may have better capacitors, more capacity to
| withstand dips and spikes in mains voltage and may have a better power factor
| so be more effective overall.
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
| the cost differential between 300 and 600W should be relatively small.
|
| Easier to overspecify: the other thing is that larger PSU wattages may have
| quieter / better quality fans. I love almost silent PCs.

I just wanted to point out that larger PSU can be more efficient, but smaller PSU can also be more efficient. Even when energy efficiency labels are compared (cf. https://en.wikipedia.org/wiki/80_Plus), a better rating (e.g. gold over silver or such) may not always indicate better efficiency.

Say, for example, you have two PSUs under consideration:

(a) PSU with 450W and 80+ silver rating
(b) PSU with 600W and 80+ gold rating.

Then at 20% load, 80+ specifies (a) to have efficiency 87% and (b) to have efficiency 90%. In absolute numbers:

(a) PSU with 450W will have specified 87% at 90W i.e. draw 90W/0.87 = 103W
(b) PSU with 600W will have specified 90% at 120W i.e. draw 120W/0.9 = 133W

As 20% is the lowest load specified for the rating (silver, gold etc.) we cannot tell how the respective PSUs operate if less than 20% load is requested. From the rating we only know that (a) will take at most 103W in idle loads and (b) at most 133W, hence the power consumption of (b) could potentially be higher in very-low-load idle scenarios which are not uncommon to be the dominating factor for typical PC worksloads.

HTH
Linux-Fan

öö

Attachment: pgpEEAkKPX6RC.pgp
Description: PGP signature


Reply to: