Alexandre Posted December 29, 2010 Posted December 29, 2010 (edited) I've noticed that about 90% of the people tend to get power supplies much larger than they actually need. This is because power supplies are most efficient (Efficiency = DC power supplied to computer/AC power drawn from wall socket) when under 40%-60% load so people get at least double the wattage they need. This has been tested and shown countless of times to be true in PSU reviews (even OCC ones). Even manufactures sometimes show this fact in their own efficiency graphs. So, Smallville8vdm, I would suggest that you use a good PSU calculator (the one boinker suggested is great), double the wattage it suggests and see if your PSU can handle that. That's the wattage you should go for. You don't want to have your PSU loaded to 80%, as it it less efficient then, aim for around 50% load [as paulktreg suggested] (and like some people above have said..some power supplies cant even get to 80% load because they cant supply the power and some even blow up before they reach it ) If you want to learn more about PSUs I would suggest you read up on this article - can be a big help: Everything You Need to Know About Power Supplies Edited December 30, 2010 by Alexandre Share this post Link to post Share on other sites More sharing options...
paulktreg Posted December 30, 2010 Posted December 30, 2010 The Power Factor (not the 80+ rating - that's all pretty much garbage when you see how they do the rating) is probably your best bet in determining the efficiency... In my work system, I dropped the power supply down from a 750 to 550 just because it was determined to be oversized (QX6800 + HD3850 + 5 Hard drives + burner + bunch of accessories, fans, USB devices, lighting). Compared to newer components, these actually draw more power off the wall, and the 550 still works without a problem Power factor has nothing whatsoever to do with efficiency, totally unrelated! I wouldn't call 80 Plus garbage. I agree it as its faults but at the end of the day you won't find a cheap generic that qualifies. Your justification for spending extra money on 550W units to replace already installed 750W units makes no sense unless quality was an issue? I am looking at 3x servers in front of me - all running 2x E5630 + 18GB RAM @ average load - all off a single 1000W power supply (there's actually more that I'm not including - hard drives, a 24 port gig enterprise switch that's powered by the same power supply). Your power supply might work out just fine - I've noticed that about 90% of the people tend to get power supplies much larger than they actually need. If you already have the components, the best thing to do - test it... Three servers running of one 1000W power supply? No redundancy? Are you serious or just trying to wind me up? Share this post Link to post Share on other sites More sharing options...
Alexandre Posted December 30, 2010 Posted December 30, 2010 (edited) Power factor has nothing whatsoever to do with efficiency, totally unrelated! was about to say the same thing Btw Smallville8vdm I would probably trust paulktreg more than anyone else here because he's done some good PSU reviews here at OCC Edited December 30, 2010 by Alexandre Share this post Link to post Share on other sites More sharing options...
_TheAlexO Posted December 30, 2010 Posted December 30, 2010 just do whatever Paul says Share this post Link to post Share on other sites More sharing options...
sticknstone Posted December 30, 2010 Posted December 30, 2010 GPU benchmarking software is used to simulate a typical gaming load and not fully load the power supply. When you say 80% I take it you mean your computers full load is 400W so fit a 500W power supply? If a benchmark is set to maximum settings wouldn't that be considered fully loading? What else could a user do? A computer reads binary. It was from memory, I was stating 500w so buy 400w psu. The source was some kind of guide written in pdf. I do not even know where to find it. It stated that a typical computer never uses all of its features all the time so choosing a psu at 80% is acceptable. ie, odd, or usb ports are not being used. Share this post Link to post Share on other sites More sharing options...
sticknstone Posted December 30, 2010 Posted December 30, 2010 If you can check some books at the library. Two books with different methods than I stated are Bigelow's "Troubleshooting, Maintaining and Repairing pcs" or "Acing the A+ certification exam". They give each component an estimated wattage based on ampsXvolts=watts and add up the total. Share this post Link to post Share on other sites More sharing options...
Alexandre Posted December 30, 2010 Posted December 30, 2010 I was stating 500w so buy 400w psu. You saying that if you require 500w supply you should get a 400w because the computer doesn't use all the power it needs at one time? This is a recipe for random computer crashes Like it was said before, you want a PSU which is loaded approximately 50% most of the time for best efficiency, low noise level and room for hardware upgrades. Share this post Link to post Share on other sites More sharing options...
mcw Posted December 30, 2010 Posted December 30, 2010 Power factor has nothing whatsoever to do with efficiency, totally unrelated! I wouldn't call 80 Plus garbage. I agree it as its faults but at the end of the day you won't find a cheap generic that qualifies. Your justification for spending extra money on 550W units to replace already installed 750W units makes no sense unless quality was an issue? Three servers running of one 1000W power supply? No redundancy? Are you serious or just trying to wind me up? I never said 80+ was garbage - I said that it doesn't mean much at the end... Keep in mind that not all power supplies even need to get that rating done (and it costs money to do it - it's more marketing than anything else). I also never said I spent extra money on a 550W (where I work, we have access to many parts/test parts/engineering samples/etc) - it was determined, after testing, that the 550W was more than adequate (and still has room to grow) with the loaded configuration. The 750W was overkill. And yes, 3 servers off a single 1000W... I can't tell you more than that (NDA), but it's there, it works... The power supplies and components have been commercially available since the end of 2007, but the configuration isn't up for discussion at this time. Of course, that's worst case scenario... In commercial use, there is a LOT of redundancy (it's not a single power supply). And no, you can't just buy the power supply and expect to use it like that... Power supply calculators are nothing compared to hands on testing.... And we do that quite a bit here. (Oh, and for the 1000W, the power factor played a big part - if I was dealing with something less than .98, I'm sure that 1000W power supplies wouldn't be adequate). Share this post Link to post Share on other sites More sharing options...
paulktreg Posted December 30, 2010 Posted December 30, 2010 I never said 80+ was garbage - I said that it doesn't mean much at the end... Keep in mind that not all power supplies even need to get that rating done (and it costs money to do it - it's more marketing than anything else). "The Power Factor (not the 80+ rating - that's all pretty much garbage when you see how they do the rating) is probably your best bet in determining the efficiency" Power supply calculators are nothing compared to hands on testing.... And we do that quite a bit here. Power supply calculators are a useful start but I'm of the opinion that most of them include a safety factor and consequently quote high to cater for Joe Bloggs and his $20.00 generic power supply? (Oh, and for the 1000W, the power factor played a big part - if I was dealing with something less than .98, I'm sure that 1000W power supplies wouldn't be adequate). You'll have to explain this because I don't understand what you getting at? A power supply rated at 1000W will deliver this power whether it's PF is 0.75 or 1.00, it doesn't matter. You don't work in a large IT based commercial organization with hundreds of servers/PC's do you? If you do then the utility company will charge extra for poor power factor levels and I could understand the need to use power supplies with active power factor correction. Share this post Link to post Share on other sites More sharing options...
mcw Posted December 30, 2010 Posted December 30, 2010 "The Power Factor (not the 80+ rating - that's all pretty much garbage when you see how they do the rating) is probably your best bet in determining the efficiency" Power supply calculators are a useful start but I'm of the opinion that most of them include a safety factor and consequently quote high to cater for Joe Bloggs and his $20.00 generic power supply? You'll have to explain this because I don't understand what you getting at? A power supply rated at 1000W will deliver this power whether it's PF is 0.75 or 1.00, it doesn't matter. You don't work in a large IT based commercial organization with hundreds of servers/PC's do you? If you do then the utility company will charge extra for poor power factor levels and I could understand the need to use power supplies with active power factor correction. I stand corrected (on garbage line) - didn't mean it as ignore it - meant it as "it's pretty much marketing" (and not all "Great" power supplies may even apply for or have these ratings... I really hate getting a power supply much larger than necessary - it's an added expense that's wasted... I'd rather calculate myself than use an online one, just because they ARE going to overestimate... Some are worse than others. I'm not saying Power Factor is the basis of the whole decision - I'm simply stating that power supplies with better power factors tend to have better manufacturing processes behind them. Which goes back into the "getting a larger power supply than necessary" - it's probably because the components aren't the best that they could be... And I can't say where I work... Share this post Link to post Share on other sites More sharing options...
paulktreg Posted December 30, 2010 Posted December 30, 2010 And I can't say where I work... FBI, CIA, NCIS, Walmart? Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted December 30, 2010 Posted December 30, 2010 FBI, CIA, NCIS, Walmart? Damn it, you're weren't supposed to say Walmart! Now he's boned. Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now