1 |
> > Even my rig - hotter, doesn't reach 300W when I artificially torture the |
2 |
> > system. Normal 'max' load is in 200W range. An normal desktop? Under 100. |
3 |
> > |
4 |
> |
5 |
> OK -- just to find out the truth I've attached a kill-a-watt to my current |
6 |
> workstation which is ~4yrs old w/ slow cpu and ancient video card but |
7 |
> has been upgraded w/ 7 SATA Drives: |
8 |
> |
9 |
|
10 |
So the figures below are AC in, not DC out. Pretty sure the figure everyone |
11 |
uses for comparison is DC out. |
12 |
|
13 |
|
14 |
> Idle - ~285W |
15 |
> Light Use (emerge --sync) - ~310W |
16 |
> Kernel Compile w/ video app running and minor torture- ~340W |
17 |
> |
18 |
> This is definitely much higher than 100-200W stated above. |
19 |
> |
20 |
> Anyhow, given that the discussion was about a system lasting ~8yrs, which |
21 |
> is |
22 |
> twice the current age of my system, I don't think it's unfeasible that |
23 |
> future |
24 |
> upgrades (especially if video card related or if moving cpu from 2 core to |
25 |
> 8 |
26 |
> core) could get normal power util 20% higher to ~372W eventually. |
27 |
> |
28 |
> If you conservatively state that PSU wattage should be 1.66 * normal util |
29 |
> (so |
30 |
> that PSU is normally running at 60% of peak) then: |
31 |
> |
32 |
> 1.66 * 372 = 617 |
33 |
> |
34 |
|
35 |
I think you've double dipped there....see above comment. |
36 |
|
37 |
My 2c WRT power supplies - buy a quality brand as they are one of the least |
38 |
reliable components in a PC. |