Power Preservation

[XZDX]

Maybe LTT looks at this, but for literally years we've had 80W CPU's.  What if you tried different age PSU's on one machine?  Would it get more efficient with newer PSU?

 

Thinking like....  P2 vs P4 vs PD VS E6600 or something

 

Edit:  If you were to update a board from electrolitics to all CIC, aside from CPU TDP ETC, could you make a machine that much more efficient?

[flibberdipper]

I mean technically it would. I had my dual P3 system running off an older Rosewill RV350 and that PSU would vomit out heat like nobody's business (aka a looooot of energy wasted in the form of heat). Threw the same system on a CX430M and the air that came out of that PSU was always ice cold, which would be less energy wasted in the form of heat.

[manikyath]

11 minutes ago, XZDX said:

What if you tried different age PSU's on one machine?

if none of the PSU's are faulty, the machine works the exact same way, and efficiency is as quoted on the PSU's efficiency graph. this is nothing different from comparing PSU's using a dummy load, except it's less precise than a dummy load.

[tkitch]

It would be more efficient, if the newer PSU is more efficient.

 

This isn't really something that would make an interesting video.

 

If what you're trying to do is make a super low-powered PC, they did do a video on that:

 

 

[Needfuldoer]

That just sounds like testing power supply efficiency with extra steps. 

[XZDX]

37 minutes ago, Needfuldoer said:

That just sounds like testing power supply efficiency with extra steps. 

Running a pentium 2 at cost of CPU TDP only?  Yall need more engineering time.

[Needfuldoer]

39 minutes ago, XZDX said:

Running a pentium 2 at cost of CPU TDP only?  Yall need more engineering time.

Maybe I'm missing your point, but if the load doesn't change (you run all the same components), and the power supply is the only variable, how is that a test of anything but various power supplies' efficiency when running a Pentium II computer?

 

A Pentium II won't be any more or less efficient regardless of whether it's running off a 30 year old power supply or a new 80 Plus Titanium rated power supply. The system as a whole will draw less power from the wall, but that's only because the newer power supply is more efficient at converting line voltage AC to low voltage DC. The CPU doesn't care where those watts come from, it's always going to draw as much as it wants.

 

Processor efficiency comes from architectural improvements and process node shrinks, not from the power supply.

[starsmine]

5 hours ago, XZDX said:

Running a pentium 2 at cost of CPU TDP only?  Yall need more engineering time.

Im lost to what you are asking about.

80W is 80W

A PSU that is 80% @80W efficient requires 100W to provide that 80W.
A PSU that is 50% @80W efficient requires 160W to provide that 80W
A PSU that is 94% @80W efficient requires 85W to proved that 80W


but they all provide 80W. 
no mater what generation of CPU is asking for 80W its 80W. How the CPU uses that 80W will change from CPU to CPU sure. 

[Godlygamer23]

7 hours ago, XZDX said:

Maybe LTT looks at this, but for literally years we've had 80W CPU's.  What if you tried different age PSU's on one machine?  Would it get more efficient with newer PSU?

 

Thinking like....  P2 vs P4 vs PD VS E6600 or something

 

Edit:  If you were to update a board from electrolitics to all CIC, aside from CPU TDP ETC, could you make a machine that much more efficient?

Depends on how you define efficiency, and the scope of how you're measuring it. If you're looking at just the CPU, no it wouldn't get more efficient. But if you're looking at the system as a whole, then yes it would be more efficient. But the CPU's own efficiency remains the same. It is getting 80W of energy. That's it. You're just making the front end more efficient.

[cmndr]

In terms of compatabilty:
Really old PSUs used the AT standard designed by Intel. 
After that Intel came out with ATX... and then different revisions of ATX. 

I haven't tried it, but I suspect that the AT -> ATX transition of parts would not work particularly well, the physical connectors are different. 
Older ATX PSUs have a different pin out for providing CPU power and main power to the board. 

On top of this older systems used to rely much more heavily on 5V power delivery while newer systems mostly rely on the 12V rail. 

In terms of efficiency:
Newer PSUs are around 90% efficient. The efficiency is around 90% between 10-90% load. This is under modern power draw conditions.
Old PSUs are often 70%ish or less. Usually at around 50% load with it dropping off at lower or higher loads.

 

 

I suspect that modern PSUs will be more efficient in modern systems and older PSUs will struggle with compatibility. 

This is a thing you'd want to test with VERY VERY cheap motherboards and CPUs. There's stuff that'll die. 

[manikyath]

3 hours ago, cmndr said:

This is a thing you'd want to test with VERY VERY cheap motherboards and CPUs. There's stuff that'll die. 

that's why you test with a dummy load that can deal with this.

[cmndr]

13 hours ago, manikyath said:

that's why you test with a dummy load that can deal with this.

Yeah... but "exploding PSUs" is a thing that definitely was MORE of a thing with older PSUs. 

Also a lot of the older motherboards need to be recapped. Especially the ones from the early 2000s during the great capacitor plague. 
https://en.wikipedia.org/wiki/Capacitor_plague

[Needfuldoer]

16 hours ago, cmndr said:

Yeah... but "exploding PSUs" is a thing that definitely was MORE of a thing with older PSUs. 

That's usually caused by a failed metallic foil line filtering capacitor. Most of the time there's no harm to the electronics, they just fill the room with acrid smoke.

 

https://hackaday.com/2023/04/01/why-do-rifa-capacitors-fail/

[cmndr]

17 hours ago, Needfuldoer said:

That's usually caused by a failed metallic foil line filtering capacitor. Most of the time there's no harm to the electronics, they just fill the room with acrid smoke.

 

https://hackaday.com/2023/04/01/why-do-rifa-capacitors-fail/

I'll need to look into that a bit. My oversimplified understanding of reality is that before 2008ish PSUs were kind of "ehh" in terms of quality and efficiency and that after that PSUs just got way better, more reliable and cheaper. Nearly 2x the wattage, half the cost and half the waste power. 

[Needfuldoer]

3 minutes ago, cmndr said:

I'll need to look into that a bit. My oversimplified understanding of reality is that before 2008ish PSUs were kind of "ehh" in terms of quality and efficiency and that after that PSUs just got way better, more reliable and cheaper. Nearly 2x the wattage, half the cost and half the waste power. 

Electrolytic caps from the late 90s through late 2000s suffer from the capacitor plague. Hot components and sketchy build quality led to a lot of premature failure.

 

Surface mount electrolytic caps from the late 80s and early 90s are failing now too, but mostly from age. I tinker with old computers from that era, and if they use these caps they're almost guaranteed to be bad by now. (The telltale sign is oily-looking wet residue on the board. That's liquid electrolyte.)