How to Pick the Right Power Supply

[cytherian]

Great resource on PSU guys, thanks loads for providing it. I wish I'd seen this before I embarked on my replacement PSU quest for an old desktop. The old PSU couldn't handle the new graphics card, so I had no choice but to upgrade. But the PC was an aging HP Compaq D530 with an odd ATX case that forced the need for a proprietary PSU design. A regular ATX PSU just wouldn't do without case modifications.

Anyway... with the desktop slowly drifting into the background and mobile devices taking front and center, I guess the need for replacement PSU's is going to reduce significantly unless the desktop makes a comeback.

 

[NoN]

eXtreme Power Supply Calculator Lite
for personal, non-commercial use only!

Looks quite accurate and the most complete that i have seen around.

 

[ICIT2LOL]

Yep Non I have used this site beofre and had forgotten about it - nice pick!!

Was trying this afternoon to decide between the Corsair HX 650 V2 gold and the TX 750 V2 so am going to hunt through for the reviews

 

[NoN]

What i found is that this power generator doesn't calculate straight at full load as some does (tested two other sites). Therefore for me, this one is more accurate as i can manually add in % the amount it could eventually be loaded in Power.

I've got a power supply Cooler Master GX650W 80+ Bronze-V2 (Which is not anymore made by Seventeam), but i'm also concern about efficiency.

Bought this one to start my new rig as it was a good bargain and enough quality to stand amongst some others.

Oh well, i'm trying to look at some Gold label because me house have some bad electricity design in and i'm not enough confident with an APC Ups Backup.

Good luck in your search!

 

[Proximon]

When you actually do the math regarding efficiency you discover savings are normally fairly minor. In most cases you will have a hard time breaking even by picking an 80plus Gold PSU over a Bronze.
That's not to say you should avoid more expensive Gold or Platinum PSUs. They will make a difference in heat generated, and most importantly global power usage.
Take my current load: 120.9W at the wall. My PSU is a Seasonic X650, which is about 90% efficient at that load. So, I can calculate my actual DC draw to be 108W.
If I was using a Platinum Seasonic 1250W, my efficiency at that load would be about 85%, or 124W.
4W difference, despite the PSU being double the rating.

Your new CM PSU has a fairly flat efficiency curve according to Kitguru
http://www.kitguru.net/components/p...er-master-gx650-bronze-power-supply-review/5/
But they didn't test at a lower load that most modern machines will idle at.

This Greek site has the numbers you need
Coolermaster GX650W Bronze review

Now, an IT pro does need to know about temps vs. efficiency and might have to worry about that 4W difference if it was multiplied by a large number of users.

I thought I would address the question specifically as it comes up a lot and might help others as well.

 

[NoN]

When you actually do the math regarding efficiency you discover savings are normally fairly minor. In most cases you will have a hard time breaking even by picking an 80plus Gold PSU over a Bronze.
That's not to say you should avoid more expensive Gold or Platinum PSUs. They will make a difference in heat generated, and most importantly global power usage.
Take my current load: 120.9W at the wall. My PSU is a Seasonic X650, which is about 90% efficient at that load. So, I can calculate my actual DC draw to be 108W.
If I was using a Platinum Seasonic 1250W, my efficiency at that load would be about 85%, or 124W.
4W difference, despite the PSU being double the rating.

Your new CM PSU has a fairly flat efficiency curve according to Kitguru
http://www.kitguru.net/components/p...er-master-gx650-bronze-power-supply-review/5/
But they didn't test at a lower load that most modern machines will idle at.

This Greek site has the numbers you need
Coolermaster GX650W Bronze review

Now, an IT pro does need to know about temps vs. efficiency and might have to worry about that 4W difference if it was multiplied by a large number of users.

I thought I would address the question specifically as it comes up a lot and might help others as well.

@proximon, First thank you for the kitguru link and comments.

I knew about the lab.gr and it was one of the first review with the one Here i readed about this power-supply.

I know it is a basic power-supply but quite well rated as it is also a second version, improved from Cooler Master for that GX series.

Now for those around buying a new power supply, efficiency from 82% to 92% makes a huge difference in Kw/year.

Let's say for a computer working 8 hours a day, 24/7, 365 days a year, makes around: 2920 hours working computer/year

Gx650w 80+Bronze low load results:

at 107,50 DC/AC (Watts), efficiency 82,85%, gives a PF/AC (Volts) 0.837/232,6V.

Let's say the price i have to pay is around 0,10euro/$ Kw/h.

0.837 x 2920 = 2444,04 Kw/year
2444,04 x 0,10$/euro = 244,40$/euro to pay per year.


If i think about an 92,85% efficiency and make the same math about the cost for the results given, it will a difference of 24,40$/euro per year.

Then i will have to pay 220$/euro per year. Therefore, 2,03$/euro per month less to pay.

With those results, it'll start to be interesting for me, only after two years, and if i decide to upgrade/new buy a power-supply 80+ Bronze to a 80+ Platinum or Gold.

Of course i'll have to wait at least 2 years to have my money back for the price i paid for an upgrade/new buy!!!

Because of so many crisis all around the world, prices are increasing next month for electricity to pay/month.
Gas was last month but my computer ain't Gas working!

The price of the power-supply actually doesn't fluctuate that much, then i'll be temporarily a winner!!!

 

[ignatzatsonic]

I pay about 13 cents per killowatt hour and run my PC about 4000 hours per year.

My previous PC was a Core 2 Duo E6600 with a Seasonic 330 watt PSU. It used about 127 watts on average. That's about $66 per year, given my power rate and number of hours per year that the PC is on.

I upgraded to a Sandy Bridge i-5 2500 with a somewhat more efficient Seasonic 560 watt PSU. This new PC uses about 75 watts on average. That amounts to about $39 per year, given my power rate and number of hours per year that the PC is on.

These power measurements were made with a Killawatt meter and are not wild guesses.

So, the new PC is $27 per year less expensive to operate, BUT at least 90% of that savings is due to the new CPU rather than the increased efficiency of the new PSU.

 

[NoN]

I pay about 13 cents per killowatt hour and run my PC about 4000 hours per year.

My previous PC was a Core 2 Duo E6600 with a Seasonic 330 watt PSU. It used about 127 watts on average. That's about $66 per year, given my power rate and number of hours per year that the PC is on.

I upgraded to a Sandy Bridge i-5 2500 with a somewhat more efficient Seasonic 560 watt PSU. This new PC uses about 75 watts on average. That amounts to about $39 per year, given my power rate and number of hours per year that the PC is on.

These power measurements were made with a Killawatt meter and are not wild guesses.

So, the new PC is $27 per year less expensive to operate, BUT at least 90% of that savings is due to the new CPU rather than the increased efficiency of the new PSU.

You won't make operate anyway your PSU without a CPU ....

They're tightened all together and the results are at the wall and pocket.

That's the reason of How to pick the right power supply!!

 

[ignatzatsonic]

You won't make operate anyway your PSU without a CPU ....

They're tightened all together and the results are at the wall and pocket.

That's the reason of How to pick the right power supply!!

What???

You may have a point, but it escapes me.

 

[NoN]

You won't make operate anyway your PSU without a CPU ....

They're tightened all together and the results are at the wall and pocket.

That's the reason of How to pick the right power supply!!

What???

You may have a point, but it escapes me.

I meant, if you ever want to have the results of that power supply you're looking for, then the guy that did benchmark the low load for you, sure he had mounted a computer to operate the power supply, no matter the cpu he used for that purpose!

We are talking about the load in %!

Edit:
I gave a result of 24,40$/euro per year, then you gave me a result of 27$/year less expensive for your new PC...pretty similar, no?

finally not so...

 

[ignatzatsonic]

I meant, if you ever want to have the results of that power supply you're looking for, then the guy that did benchmark the low load for you, sure he had mounted a computer to operate the power supply, no matter the cpu he used for that purpose!

We are talking about the load in %!

I have no idea what you are talking about.

I'm not looking for a power supply. As I said in my post, I already upgraded it.

No one did a benchmark for me.

I measured the actual power consumption on my old PC and on my new PC, using a Killawatt meter.

Obviously, a computer needs both a CPU and a PSU. Who would think otherwise?

 

[NoN]

Doesn't matter...i surely must be out of my mind...not knowing any!

 

[ignatzatsonic]

Doesn't matter...i surely must be out of my mind...not knowing any!

That could be. I don't know you, but I'll take your word for it.

My new PC uses about $27 less power per year than the old one.

But less than $5 of that is due to the more efficient power supply. Most of the difference is because of a more power-efficient CPU.

My point is that the average person will likely save more by moving to a more power-efficient CPU than by moving to a more power-efficient PSU.

How many dollars will of course depend on power rates (13 cents per KWH for me) and hours powered on (4000 per year for me).

 

[NoN]

When you actually do the math regarding efficiency you discover savings are normally fairly minor. In most cases you will have a hard time breaking even by picking an 80plus Gold PSU over a Bronze.
That's not to say you should avoid more expensive Gold or Platinum PSUs. They will make a difference in heat generated, and most importantly global power usage.
Take my current load: 120.9W at the wall. My PSU is a Seasonic X650, which is about 90% efficient at that load. So, I can calculate my actual DC draw to be 108W.
If I was using a Platinum Seasonic 1250W, my efficiency at that load would be about 85%, or 124W.
4W difference, despite the PSU being double the rating.

Your new CM PSU has a fairly flat efficiency curve according to Kitguru
http://www.kitguru.net/components/p...er-master-gx650-bronze-power-supply-review/5/
But they didn't test at a lower load that most modern machines will idle at.

This Greek site has the numbers you need
Coolermaster GX650W Bronze review

Now, an IT pro does need to know about temps vs. efficiency and might have to worry about that 4W difference if it was multiplied by a large number of users.

I thought I would address the question specifically as it comes up a lot and might help others as well.

@proximon, First thank you for the kitguru link and comments.

I knew about the lab.gr and it was one of the first review with the one Here i readed about this power-supply.

I know it is a basic power-supply but quite well rated as it is also a second version, improved from Cooler Master for that GX series.

Now for those around buying a new power supply, efficiency from 82% to 92% makes a huge difference in Kw/year.

Let's say for a computer working 8 hours a day, 24/7, 365 days a year, makes around: 2920 hours working computer/year

Gx650w 80+Bronze low load results:

at 107,50 DC/AC (Watts), efficiency 82,85%, gives a PF/AC (Volts) 0.837/232,6V.

Let's say the price i have to pay is around 0,10euro/$ Kw/h.

0.837 x 2920 = 2444,04 Kw/year
2444,04 x 0,10$/euro = 244,40$/euro to pay per year.


If i think about an 92,85% efficiency and make the same math about the cost for the results given, it will a difference of 24,40$/euro per year.

Then i will have to pay 220$/euro per year. Therefore, 2,03$/euro per month less to pay.

With those results, it'll start to be interesting for me, only after two years, and if i decide to upgrade/new buy a power-supply 80+ Bronze to a 80+ Platinum or Gold.

Of course i'll have to wait at least 2 years to have my money back for the price i paid for an upgrade/new buy!!!

Because of so many crisis all around the world, prices are increasing next month for electricity to pay/month.
Gas was last month but my computer ain't Gas working!

The price of the power-supply actually doesn't fluctuate that much, then i'll be temporarily a winner!!!

So i should adjust my thinking...

Guys think about it!
Take a more efficient CPU aswell a more efficient PSU and you'll be the winner!!

 

[NoN]

Doesn't matter...i surely must be out of my mind...not knowing any!

That could be. I don't know you, but I'll take your word for it.

My new PC uses about $27 less power per year than the old one.

But less than $5 of that is due to the more efficient power supply. Most of the difference is because of a more power-efficient CPU.

My point is that the average person will likely save more by moving to a more power-efficient CPU than by moving to a more power-efficient PSU.

How many dollars will of course depend on power rates (13 cents per KWH for me) and hours powered on (4000 per year for me).

0,13cents per hour x 4000 hours powered = $520 per year...$43,33/month

I pay about 13 cents per killowatt hour and run my PC about 4000 hours per year.

My previous PC was a Core 2 Duo E6600 with a Seasonic 330 watt PSU. It used about 127 watts on average. That's about $66 per year, given my power rate and number of hours per year that the PC is on.

I upgraded to a Sandy Bridge i-5 2500 with a somewhat more efficient Seasonic 560 watt PSU. This new PC uses about 75 watts on average. That amounts to about $39 per year, given my power rate and number of hours per year that the PC is on.

You confused me to say that your new PC is at $39 per year.

You should be max at 2948Kw/per year

2948Kw/year x $0,13 =$383,24

 

[ignatzatsonic]

0,13cents per hour x 4000 hours powered = $520 per year...$43,33/month

You confused me to say that your new PC is at $39 per year.

Apparently, you remain confused.

That is 13 cents per KWH for my cost of power.

Not 13 cents per hour of operation of my PC.

It would cost me 13 cents per hour of operation of my PC IF AND ONLY IF my PC used 1000 watts of power. It does not.

I hope you can understand the distinction. Maybe not.

 

[NoN]

0,13cents per hour x 4000 hours powered = $520 per year...$43,33/month

You confused me to say that your new PC is at $39 per year.

Apparently, you remain confused.

That is 13 cents per KWH for my cost of power.

Not 13 cents per hour of operation of my PC.

It would cost me 13 cents per hour of operation of my PC IF AND ONLY IF my PC used 1000 watts of power. It does not.

I hope you can understand the distinction. Maybe not.

Sorry, i find just impossible to pay that, for powered your PC a year!
I'm not arguing either...

You should be max loads at 2948Kw/per year for an average user.
2948Kw/year x $0,13Kw/hour = $383,24/year

but at low loads doesn't goes that cheaper.

The point saying by proximon is that even if the PSU is double rating in power it doesn't make a huge difference between the two of them in Watt output. therefore doesn't help me that much to save money for an upgrade.

That's not to say you should avoid more expensive Gold or Platinum PSUs. They will make a difference in heat generated, and most importantly global power usage.
Take my current load: 120.9W at the wall. My PSU is a Seasonic X650, which is about 90% efficient at that load. So, I can calculate my actual DC draw to be 108W.
If I was using a Platinum Seasonic 1250W, my efficiency at that load would be about 85%, or 124W.

 

[ICIT2LOL]

I pay about 13 cents per killowatt hour and run my PC about 4000 hours per year.

My previous PC was a Core 2 Duo E6600 with a Seasonic 330 watt PSU. It used about 127 watts on average. That's about $66 per year, given my power rate and number of hours per year that the PC is on.

I upgraded to a Sandy Bridge i-5 2500 with a somewhat more efficient Seasonic 560 watt PSU. This new PC uses about 75 watts on average. That amounts to about $39 per year, given my power rate and number of hours per year that the PC is on.

These power measurements were made with a Killawatt meter and are not wild guesses.

So, the new PC is $27 per year less expensive to operate, BUT at least 90% of that savings is due to the new CPU rather than the increased efficiency of the new PSU.

My giddy aunt iganatzatsonic I want to go on your power grid my power costs me 38c per KWhr my UPS is showing up 183Watts at the moment but I'd have other stuff besides this machine on stand by modem etc so 8 hours a day at 7c an hour gives me about $210 a year (365 days) about $51 a quarter.

What (no pun intended) I don't get though is that because a PSU is rated at say 450 watts does mean the primary circuit is dissipating that many watts does it?

 

[ignatzatsonic]

Sorry, i find just impossible to pay that, for powered your PC a year!
I'm not arguing either...

You should be max loads at 2948Kw/per year for an average user.
2948Kw/year x $0,13 = $383,24/year

Have you considered the possibility that power rates vary around the world and that some PCs use more power than others?

I live in the USA. I have no idea where you live.

I pay 13 cents per kilowatt hour. I have no idea what you pay.

My PC typically uses less than 100 watts. It runs 10 or 12 hours a day--about 4000 hours per year. Ordinary multiplication says that amounts to a little over a penny a day. I'm sure you can do the arithmetic.

Sorry about your power rates and your power consumption.

My PC specs are in plain sight.

Since you say you find that impossible to believe, your alternative is to believe I am making these numbers up out of thin air. Feel free to do that.

 

[ICIT2LOL]

Sorry, i find just impossible to pay that, for powered your PC a year!
I'm not arguing either...

You should be max loads at 2948Kw/per year for an average user.
2948Kw/year x $0,13 = $383,24/year

Have you considered the possibility that power rates vary around the world and that some PCs use more power than others?

I live in the USA. I have no idea where you live.

I pay 13 cents per kilowatt hour. I have no idea what you pay.

My PC typically uses less than 100 watts. It runs 10 or 12 hours a day--about 4000 hours per year. Ordinary multiplication says that amounts to a little over a penny a day. I'm sure you can do the arithmetic.

Sorry about your power rates and your power consumption.

My PC specs are in plain sight.

Since you say you find that impossible to believe, your alternative is to believe I am making these numbers up out of thin air. Feel free to do that.

Yeah you are quite right igantazatsonic for example the rates down under are through the roof at the moment mate - wait til they start bunging on the carbon tax like our wretched mob of pollies did here. My quarterly bill last quarter was $1204! that equates to $1208 US! = $4832 per annum.
In my case thats a large % of my salary