Which power supply for this system?

[peterb]

Price != quality. Cheap != bad.

TBH all PSUs tend to be built to a price and attacked by the bean counters. There simply isn't the market for quality as it's all done on price.

Your equations are true, but not complete. I have seen some reasonable cheap PSU, but I have also seen some truly nasty ones that look as if they have been thrown together. One in particular had a connector miss placed so it was shorted out, at which point the wires glowed and the insulation caught fire.

It was replaced with a not quite so cheap but fairly well known branded device (cost about £25 instead of £15) as luck would have it, the connector again shorted (later investigation showed that the receptacle was damaged) but this PSU just shut itself down without any drama.

You can pay a lot for some rubbish, but the chances of getting a poorly built expensive branded PSU are lower, because the company with the name on the badge has a reputation to protect.

So one manufacturer (as I said in an earlier post) may be producing several lines of branded PSUs, of variying quality because the commissioning company will have specified the electrical characteristics and the unit price. If the unit price is too low, something has to give, but if the spec is important the price will reflect that. A better unit will also have more comprehensive QA and QC checks, which add to the price.

It depends whether you want a PSU that is built up to a spec, or down to price.

But to complete your equations

High quality != cheap

[me-yeah]

I assumed you knew which is why you said it'd take no time at all.

Specs as per the original post, 650W bronze vs gold, 2 hours of gaming a day (at 300W load).

edit: Let's work through it then.

Electricity is 10p/kWh. 2 hours of gaming at 300W at 100% efficiency would be 6p. Bronze efficiency is 85% at 50% load so 7.059p. Gold is 90% @ 50% load, so 6.667p

So if using the PC for gaming like that, you'd save 0.392p a day, or 11.92p a month.

An Antec 650W bronze costs £51.18 at scan.
An Antec 650W gold costs £73.00 at scan.

Difference of £21.82.

Or 183 months.

Or 15.25 years.

Not quite 'no time at all'.

games use graphic cards and cpu at full load 90% of the time, especially if your rocking 1080p or 4k

if you don`t use the tdp when working out the watts you need, your PC will more then likely shutdown after 10 minutes of playing a game, because it does`nt have enough watts to use

[kalniel]

games use graphic cards and cpu at full load 90% of the time, especially if your rocking 1080p or 4k

Exactly. Which is why I calculated 2hrs at 300W.

if you don`t use the tdp when working out the watts you need, your PC will more then likely shutdown after 10 minutes of playing a game, because it does`nt have enough watts to use

That's not true at all. TDP is not a comparable measure of the power consumption.

But say we double the power consumption for no reason at all, that's still over 7 years to make back the difference in price going for a gold rated supply over bronze. Still not 'no time at all'.

[me-yeah]

Exactly. Which is why I calculated 2hrs at 300W.

That's not true at all. TDP is not a comparable measure of the power consumption.

But say we double the power consumption for no reason at all, that's still over 7 years to make back the difference in price going for a gold rated supply over bronze. Still not 'no time at all'.

TDP is watts for watts

https://en.wikipedia.org/wiki/Thermal_design_power

you can put some thermoelectric peltiers on your heatsink and you will probably generate the same amount of watts your card has just used

[kalniel]

TDP is watts for watts

https://en.wikipedia.org/wiki/Thermal_design_power

TDP is not power consumption. It even says it in your link:

"Rather than specifying CPU's real power dissipation, TDP serves as the nominal value for designing CPU cooling systems."

you can put some thermoelectric peltiers on your heatsink and you will probably generate the same amount of watts your card has just used

Just.. no.

[me-yeah]

TDP is not power consumption. It even says it in your link:

"Rather than specifying CPU's real power dissipation, TDP serves as the nominal value for designing CPU cooling systems."

the public does`nt need to know the cooling design TDP, besides that, find a old CPU and put the watts up above the tdp in the bios and then it will die

[kalniel]

besides that, find a old CPU and put the watts up above the tdp in the bios and then it will die

Put the watts... what? I've never seen a BIOS where you directly adjust the watts on a CPU, can you explain how you'd do it? Surely you can't know the amperage?

[me-yeah]

Put the watts... what? I've never seen a BIOS where you directly adjust the watts on a CPU, can you explain how you'd do it? Surely you can't know the amperage?

get a AMD black edition, they are unlocked for overclockers, in a perfect world, TDP is watts for watts

then put on be-quiet dark rock pro 3 heatsink

[kalniel]

get a AMD black edition, they are unlocked for overclockers, in a perfect world, TDP is watts for watts

then put on be-quiet dark rock pro 3 heatsink

OK.. but what's the option called for directly adjusting the watts in BIOS?

[ik9000]

OK.. but what's the option called for directly adjusting the watts in BIOS?

It's not in the bios silly. You need to take the CPU out and cut off two of the pins.

[Firejack]

Perhaps he is referring to the configurable TDP for AMD Kaveri APU's? The option in the BIOS is often mislabelled in broken English. I can see how people would get confused by its meaning.

[kalniel]

Perhaps he is referring to the configurable TDP for AMD Kaveri APU's? The option in the BIOS is often mislabelled in broken English. I can see how people would get confused by its meaning.

Could be. AFAIK that's a throttling limit, not actually a means to overclock the CPU, ie you can lower the limit to fixed values to reduce heat output.

[peterb]

Interesting (depending on your definition of the word) though this is, the OPs question was effectively answered by post 10 on the front page...

Me-yeah, TDP in this context is a notional value used to design the complete thermal management system of the PC. At the notional TDP the system temperatures will be stable, with enough margin to cope for the odd occasions when the actual dissipation exceeds the TDP. Under those conditions, the temperature will rise until a new equilibrium is reached, that is when the heat energy is being removed as quickly as it is being generated.

If that margin isn't great enough, or the thermal design is poor, the CPU may throttle itself or even shut down.

When the processor is running at less than the TDP, system temperatures will be lower.

However, TDP has little relationship to the energy the CPU is actually using and the current it is taking. it is a notional value to aid system designers.

[me-yeah]

in a perfect world TDP is watt for watt, anyway, and carbon heat shields and radiowaves if you read any NASA stuff

so if you put some thermoelectric peltiers on your heatsink, you can plug you heatsink back into your PSU and have a few free 200watts or something instead of having waste

[peterb]

in a perfect world TDP is watt for watt, anyway, and carbon heat shields and radiowaves if you read any NASA stuff

so if you put some thermoelectric peltiers on your heatsink, you can plug you heatsink back into your PSU and have a few free 200watts or something instead of having waste

We don't live in. Perfect world, and it is clear your grasp of thermodynamics is poor. I'm not quite sure what radio waves and carbon shields have to do with it.

Peltier devices are solid state heat pumps, with a hot and cold face, the heat extracted from the cold side appears on the hot side. So the hot side also has to be fitted with a heat sink, which should be on the outside of the case.

They are not 100% efficient either, so will add to the PSU load.

[ik9000]

Peterb is right. It's basic thermodynamics. Look up 2nd law of thermodynamics and Carnot cycles. See also perpetual motion and why it's not possible.

If you're proposing to use the heat from the CPU to somehow harvest power then you're in essence asking for a Carnot heat source. See wiki page:

https://en.wikipedia.org/wiki/Carnot_heat_engine

In the case of a peltier semiconductor the cooling effect is driven by application of a voltage across the wafer - so you're doing work to drive a temperature gradient across the two surfaces. This can cool the CPU on the cold side, but the hot side still needs cooling as Peterb says.

Worse - you've done more work to achieve this, so you've generated additional heat in the system that has to be dealt with (in the PSU) and you've consumed more electrical power in getting there.

Peltier can be useful if you've no other option, but it is not the best option. That's why we don't use it in fridges, coolers, CPU blocks or much else to be honest.

Your proposal above that the heat from the CPU - cooled by the peltier - could be used to increase power from the PSU is plain wrong. The same PSU would have to drive the peltier wafer - and you can't get out more than you put in. Since it takes work (energy consumption) to drive a heat pump the net result will be the energy potentially available for extraction would always be net loss overall. And that's setting aside the question of just how that spare heat could be recycled into electrical power inside your machine... It would more likely increase the resistance and losses within the PSU than help it!