Quantcast
 


SKYMTL's 500W+, Sub-$105 PSU Roundup

by Michael "SKYMTL" Hoenig     |     March 23, 2007

PERFORMANCE TESTS:

Instruments Used:
Belkin 1100VA UPS
Rexus PSU tester
Fluke 187 Digital Multimeter
UPM Power Meter
Tektronics TDS3032B Oscilloscope

Test Platform:
DFI Lanparty SLI-DR Expert
AMD X2 3800+ (at 2.6Ghz)
2GB Corsair PC4000 Ram (at 520Mhz)
EVGA 8800GTS (Stock, OC 650/1900, SLI)
1x Samsung Spinpoint 250GB SATA Hard drive
Antec Sonata Case
Pioneer DVD Writer

Important note:

Because of processor limitation, 8800GTS cards in SLI are seriously bottlenecked in Company of Heroes. Thus, while they still drew quite a high amount of power, when coupled with a higher end system or playing at higher resolutions they would probably draw much more.

One way or another, I would NOT recommend anything under a good 700W power supply for a pair of 8800GTS cards. These tests are done as benchmarks ONLY.


Efficiency Testing:

To test efficiency, plugged in my UPM power meter to the Belkin UPS and determined what the highest sustained power draw from the outlet was over a 1 hour test period. All tests were run twice and if there were anomalies, I tested a third time.

The first efficiency test’s “Load” value was done with an overclocked processor and the graphics card at stock speeds while running Company of Heroes. The values are the highest sustained power draw over the 2 hour test period.

The second efficiency test’s “Load” value was done with an overclocked processor and a heavily overclocked (both 2D and 3D overclocked to the same value) graphics card. Company of Heroes was played while Orthos was running on the processor in the background.

The third and final efficiency test was run with 2 8800GTS 320MB cards in SLI running at stock speeds with the processor overclocked to 2.6Ghz. Company of Heroes was then run for 1 hour to determine load values.



Efficiency Test #1





By looking at this chart we can understand why some computers fail to boot when coupled with a cheap or under-powered power supply. Upon startup, a computer consumes an amazing amount of power which may be too much for an inadequate power supply.

Results:

FSP Blue Storm: I was pleasantly surprised with the efficiency of this power supply. Even though it didn’t live up to the numbers that the Silverstone was pushing out, it was quite close across all of the tests. A very good showing by FSP.

Cooler Master 600W: This power supply really proved to be a disappointment because I was really hoping to see some better efficiency results. Instead, it gobbled up over 40W more power at load than the Silverstone. Just to be sure, I ran all the tests with the CM three times, all ended up leading to the same efficiency results.

Nmedia MP-500: Considering this is the least expensive power supply in the entire roundup, it put out some extremely respectable numbers. Even though we have once again proved that the “over 85% efficiency” listed on the box is false, the results are nonetheless impressive for a $60 power supply.

Enermax FMAII 535W: The Enermax netted the second highest load consumption which is not exactly something to be proud of but it did not come close to the power draw of the Cooler Master.

Silverstone ST50EF: Quite simply this is the most efficient power supply in this test. The FSP tried to give it a good run for its money but in the end, the Silverstone won by a fair margin. Considering this unit consumes nearly 20% less power than the Cooler Master, you can see that its 80+ certification is not just window dressing.

Mushkin HP-550: While this power supply is not as efficient as some of the others, it is a solid mid-pack performer. What should be noted is that even though the “Load” value is only 1W below that of the Enermax, the average consumption of the Mushkin was quite a bit below that of the Enermax. Since the values fluctuate so rapidly, I have no way to accurately log these peaks and valleys in consumption. Thus, this result only reflects the PEAK power consumption.


Efficiency Test #2





Here I am putting quite a bit more stress on the power supplies by running the overclocked processor with Orthos and bumping the speeds of the 8800GTS to 650/1900.

Results:

Once again the Silverstone is the clear winner but not by that much of a margin; the FSP is still keeping pace and even gaining a bit of ground in this test. The nMedia MP-500 is also keeping pace quite well in this test and once again I am stunned by its performance against power supplies that are much more expensive. Finally, the Mushkin’s peak numbers reflect its actual performance a bit better even though it stays firmly in the middle of the pack while the Enermax and Cooler Master continue to push the results to new highs (or lows if you look at it that way).



Efficiency Test #3




This is the one test that made these power supplies crap their pants and call for mommy; 8800GTS cards in SLI.

A note about the Enermax performance numbers:

You can see that the even though the Enermax was able to run this test I experienced serious performance issues in Company of Heroes. While using all of the other power supplies (except the nMedia) I received a minimum of about 52fps in the CoH SLI performance test. When using the Enermax, I received a minimum of 24fps. This test was run 4 times, each with the same result. Thus, the Enermax fails this test.

This illustrates how important a proper power supply is to the performance of your computer.


Results:

Right off the bat you can see that the nMedia called it a day and decided to shut off when the 8800 cards were put under load. While this is a bit of a disappointment, you can’t expect miracles from a $60 power supply. The performance of the Enermax has already been discussed but I will say it again: even though it consumed power, it still failed the test.

Other than those two hiccups, all of the other power supplies performed as they did for the previous ones with the Silverstone on top, the FSP still gaining ground (it is SO close) and the Cooler Master guzzling power like no one’s business. Observing the HP-550 is somewhat interesting as it seems to be getting more efficient as the power demands increase.


Voltage Regulation Testing:

To test voltage regulation I used the same tests as the efficiency. They were all done over two tests of 1 hour where the voltage drops were logged with the Fluke 187 multimeter installed on a typical Molex connector. The tests were as follows:

The “Idle” value was done with an overclocked processor and the graphics card at stock speeds while running the Windows Desktop.

The “Load” value was done with an overclocked processor and the graphics card at stock speeds while running Company of Heroes.

The “Load (OC)” value was done with an overclocked processor and a heavily overclocked (both 2D and 3D overclocked to the same value) graphics card. Company of Heroes was played while Orthos was running on the processor in the background.

The “Load (SLI)” value was run with 2 8800GTS 320MB cards in SLI running at stock speeds with the processor overclocked to 2.6Ghz. Company of Heroes was then run for 1 hour to determine load values.


+5V Voltage Regulation:

Once again, I am going to keep this short and sweet; because I do not have (and the typical user does not have either) enough components that draw power from the +5V rail to put much stress on it. Thus, I did conduct the tests with the system I had and all the power supplies passed the tests (except SLI for the Enermax and Nmedia) within +/- 5% of +5V. I did not add a chart as it would look VERY boring.


+12V Voltage Regulation





Results:

FSP Blue Storm: For some reason the Idle test showed the +12V rail at 12.3V; while this is still within the 5% +/- of 12V, it is the highest in the test. Overall, the voltage regulation of the Blue Storm is rock solid across all of the tests.

Cooler Master: Once again the Cooler Master proved a bit of a disappointment. Even though it is still within the specifications, the voltage drop in SLI was quite significant. Other than that, it performed well within the norms.

nMedia MP-500: The nMedia once again provided good performance in this test with great voltage regulation until the SLI test….where it crashed under the weight of two 8800GTS cards under load. Up until this point the nMedia is proving to be a good power supply but it does not hold up to the others performance-wise.

Enermax FMAII: While the Enermax seems to have passed all of the tests, the dreadful performance provided by it in SLI mode carries on to this test as well. Once again, I saw frame rates in Company of Heroes fall through the floor. On the other hand, it did show decent +12V voltage regulation.

Silverstone ST50EF: The proof of this power supply’s industrial-grade components affecting performance becomes apparent in this test. The first time I ran the OC test, I thought my multimeter had up and died; the voltage didn’t budge from 11.94V. All of the other power supplies had the voltages jump up and down constantly. I believe the other tests speak for themselves.

Mushkin HP-550: While not quite as steady as some of the other units tested here, the Mushkin nonetheless displayed great voltage regulation across the board. There was a bit of a fall-off when the SLI tests were running but as you can see it was well within the ATX-spec’d norms. This is a very good showing for the only modular power supply in the roundup.


+12V AC Ripple Testing

This is a very significant test in the fact that AC Ripple can be the cause of many common computer problems. Short term effects of excess ripple can be anything from an unstable overclock to memory errors while long term effects can include premature component failure and decreased component performance. The ATX v2.2 ripple tolerance is anything below 120mV on the +12V rail.

To test for ripple the following tests were run twice for 30 minutes while the ripple was being measured by the Tektronics o-scope. The values were the highest peak ripple measurement.

The “Idle” value was done with an overclocked processor and the graphics card at stock speeds while running the Windows Desktop.

The “Load” value was done with an overclocked processor and the graphics card at stock speeds while running Company of Heroes.

The “Load (OC)” value was done with an overclocked processor and a heavily overclocked (both 2D and 3D overclocked to the same value) graphics card. Company of Heroes was played while Orthos was running on the processor in the background.

The “Load (SLI)” value was run with 2 8800GTS cards in SLI running at stock speeds with the processor overclocked to 2.6Ghz. Company of Heroes was then run to determine load values.

Results:



Observations:


FSP Blue Storm: I was a bit worried with this unit considering the 400W FSP / Sparkle unit I tested in the last roundup had quite a few problems with ripple suppression. Luckily, this power supply passed with flying colors. Even when running the system full-tilt in SLI mode, this unit displayed amazing results for a $100 power supply.

Cooler Master 600W: This is one test I was hoping the Cooler Master would do well in considering its OEM’s background. It did not disappoint…until the SLI test that is. Things were going beautifully until the SLI’d 8800GTS cards kicked into high gear and the corresponding ripple shot up into the 90s. While this is still below ATX specifications allow for, it is more ripple than I want to see on a power supply in this class.

nMedia MP-500: In these tests it looks like the nMedia is starting to behave a little more like a $60, 500W budget power supply. Under load, the ripple was much higher than I would have liked to see during each of the tests. Sure, it was still below ATX specs but it was significantly higher than any other unit in this roundup. I was really hoping for some great results (like in all the other tests) but looking at past offerings from Shenzhen, I am not all that surprised.

Enermax FMAII: The results for the Enermax are firmly in the middle of the pack throughout all of the tests. It should be mentioned that even though I included the results from the SLI test, the Enermax was not providing enough power to the SLI’d graphics cards. Thus, the 80mV score should be largely ignored.

Silverstone ST50EF: Once again, Silverstone’s choice to use Enhance’s server-grade gaming power supply chassis for their unit is paying off in spades. With barely a twitch from the O-Scope through ALL of the tests, we actually ran these tests four times just to recheck the results. The results are very good from the start to the finish.

Mushkin HP-550: I was actually very happy with the results I got with the Mushkin in these the ripple tests. It was better than many of the other power supplies and was very close to the FSP across the board. I was actually quite surprised at how quiet the +12V rail was in all of the tests.



Noise and Heat

The power supplies were fully installed in my Antec Sonata case and I disabled all of the case fans except the rear 120mm which I left at the “low” setting. The only test that was conducted was the “Load” test of 1 hour of gaming with Company of Heroes. Any power supplies that could run the SLI test were tested using the pair of 8800GTS cards as well. The case door was closed and the case was put in my desk approximately 36” away from my ear.


FSP Blue Storm: Since the Blue Storm uses a 120mm fan it was dead silent until the 20 minute mark of the standard test and the 15 minute mark of the SLI test. Luckily, even though it spun up to fast speeds it was still nearly silent and could barely be heard above the whooshing noise of the 8800GTS cards. Let me tell you…those cards get HOT. The temperature coming out of the back of the FSP did heat up around the time the fan spun up but quickly cooled off again.

Cooler Master 600W: Everything with the Cooler Master went completely smoothly in this test. The fan was barely a whisper throughout the tests even though it really kicked itself into gear around the 40minute mark. The heat (or lack of) coming out the back was fairly consistent throughout the tests.

nMedia MP-500: Looking back to my hair-raising experience with the MP-400’s jet-turbine of a fan, I approached this power supply like an armed stick of C4. Luckily, all my worries were soon put to rest as the fan on this unit is dead silent. The titanic fan spins its lazy rotations with the barest whisper until…..the 20 minute mark of the tests. It then spins up and the dust bunnies behind my desk go scampering across the room faster than if Elmer Fudd was chasing them with a tank and a bucketful of attitude. Let me tell you; at full rpms this 140mm fan moves a crazy amount of air. The PSU stays totally cool throughout the tests….duh.

Enermax FMAII: Even though this power supply has the smallest fans of the roundup, they stayed blissfully silent. The only thing that concerns me is that by the 40 minute mark of the normal test (not the SLI test) the heat coming out of the power supply was quite intense even with the fan knob maxed out.

Silverstone ST50EF: Silence is the name of the game here as well. The 120mm fan on the Silverstone keeps quiet even powering the SLI system. Something I did notice at first was a faint clicking noise when the fan sun up but that quickly disappeared never to return again. The air being expelled through the back stays quite cool throughout all of the tests.

Mushkin HP-550: This power supply was really something to behold. For the life of me, I couldn’t hear the fan spinning at all. At one point I had to check if it was even running; it’s that quiet. This would be a huge problem if the power supply was on the verge of overheating but the air coming out was room temperature. Maybe I wasn’t putting enough load on the thing….
 
 
 

Latest Reviews in Featured Reviews
August 15, 2013
Corsair's Carbide AIR 540 is a unique case.  It uses a distinctive dual chamber design which not only saves vertical space, improves airflow and houses large E-ATX motherboards....
August 11, 2013
Galaxy's GTX 770 GC 4GB doubles up on the reference card's memory while also utilizing higher clock speeds.  It also happens to be one of the most expensive GTX 770 cards on the market but does the ad...
August 5, 2013
Gigabyte's new Z87X-UD5H may be targeted towards the upper end of the mid-range market but it incorporates features normally found on flagship models.  But with such cutthroad competition among Haswel...