- Back to homepage
- Related articles
  Related articles

FSP and Silverstone PSU Compared in Crossfire Stress Test

Cases & PSU/Power Supplies by geoffrey @ 2007-07-27

We take a look at two high wattage power supplies from companies known for their solid product lines; the Silverstone DA750 model features a single 12v rail and is rated at 750W. The FSP Epsilon has four 12v rails and combined offers up to 900W. We stress test these units in a real world environment with the most power hungry vga cards out there, two ATI HD 2900 XT in Crossfire. Read on to find out if they pass our tests

Test setup & results

Test setup

Geoffreys' Intel Test Setup

CPU	Intel E6600 @ 3,15GHz
Cooling	Zalman 9700 LED
Mainboard	DFI LanParty UT ICFX3200-T2R/G
Memory	2x1Gb TEAMGROUP Xtreem 800MHz 4-4-4-10
Other	Silverstone DA750 / EPS Epsilon 900 PSU Maxtor 80Gb PATA HDD Seagate 200GB SATA HDD 52x CD/RW 20" Dell UltraSharp 2007FP TFT monitor Antec 900 housing

The CPU was running at 3,15GHz by setting the front side bus to 350MHz and keeping the multiplier at default (9). The memory was running @ 400MHz (800MHz DDR) with 4-4-4-10 timings.

In order to add extra power consumption we ran our CPU at 1,55V in bios and also increased the DDR2 Vcc to 2,3V.

Results

Power conservative people won't be very satisfied with the consumption of the system we've put together for our tests, when heavy stressed the system used over 550 Watts! We measure power consumption at the wall outlet, this way we can see how efficient our PSU's are. For heavy load we ran SuperPi 32M together with 3D Mark 2003's Mother Nature test at 1600x1200 with 4AA enabled, this way we could put quite some stress on the VGA's while the load on the CPU is combined to calculating SuperPi 32M and keeping the VGA cards busy in the 3D Mark Test. One thing that must be said here: it is really important to stress the VGA's, don't load the CPU too much because it could be possible that the power consumption is going down instead of upwards. This, we think, is due to the CPU not being able to deliver enough data to the VGA's in order to keep them running at constant load.

IDLE means that our system has Windows loaded without doing anything else, Standby means that the pc is turned off, the AC power cable is still connected and the PSU's on/off switch is still active (on).

To recapitulate, the average numbers of this setup:

IDLE: ~ 260W

LOAD: ~ 580W

On to our results:

Madshrimps (c)

We let our system stress for multiple hours but we must conclude that even our CrossFire/Core2Duo setup wasn't able to bring one of those PSU's on their knees. That's a good thing, the volt rails remained very constant too although we did notice that with both PSU's the 12V line dropped its voltage a little bit. Now, 0,2V isn't anything to worry about and is well within specs.

Madshrimps (c)

Both PSU's are rated for over 80% efficiency, that's very good for such high-end components but nothing ground breaking as >80% efficiency is in fact common those days. In idle mode they both are very on par, though once our systems was being heavy stressed we could easily spot some difference between the efficiency of each PSU. We retested each PSU but the results remained unchanged, FSP's Epsilon 900 is definitely the better of Silverstone's DA750. We saw the same results in Standby mode, so you can save money even when you're not using your 'puter, though we do find that ~30W power consumption is still not looking very interesting for a pc that is not being used.

Madshrimps (c)

We tried to get a good idea of the heat that is being generated in each PSU, we installed a temp probe at the back of the PSU where hot air is blown out of the housing. Silverstone's PSU measured quite a bit warmer.

The Silverstone DA750 does have more components, the extra heat isn't caused by inefficiency of this PSU, otherwise we wouldn't have measured the idle power consumption being on par with both units.

Madshrimps (c)

Both candidates remained very silent through our tests, in fact you can barely hear them with our system booted although we did block every fan (besides those of the PSU's) in our housing. The 120mm fans are doing what they supposed to do and could without doubt be combined with a silent water-cooled high-end rig.

Comment from MTBF @ 2010/03/10

FORTRON / FSP Power Supply Units are unreliable hardware devices.
Failed just after 784 days of normal use (about 3500 hours only)!.

Comment from jmke @ 2010/03/10

MTBF doesn't guarantee you a minimum life-time expectancy FYI

Comment from MTBF @ 2010/03/10

Yes, 2 years but warranty was over since 1,5 month only.

Comment from jmke @ 2010/03/10

how many FSP PSU do you own?

Comment from MTBF @ 2010/03/10

Just one, Epsilon 700w definitively useless by now also it was properly protected with an UPS.
Such material is normally designed for 100,000 hours.

Comment from jmke @ 2010/03/10

Quote:

Such material is normally designed for 100,000 hours.

no, it has MTBF of 100.000 hours. And MTBF != LifeTime

Quote:

nother common misconception about the MTBF is that it specifies the time (on average) when the probability of failure equals the probability of not having a failure (i.e. a reliability of 50%). This is only true for certain symmetric distributions. In many cases, such as the (non-symmetric) exponential distribution, this is not the case. In particular, for an exponential failure distribution, the probability that an item will fail at or before the MTBF is approximately 0.63 (i.e. the reliability at the MTBF is 37%). For typical distributions with some variance, MTBF only represents a top-level aggregate statistic, and thus is not suitable for predicting specific time to failure, the uncertainty arising from the variability in the time-to-failure distribution.
http://en.wikipedia.org/wiki/Mean_ti...sconcepti ons

you can't decide on experience with ONE sample that a whole range of products is broken/bad.

Comment from MTBF @ 2010/03/10

I agree with you and the above MTBF but when you pay US$200 for a device you don't expect that you should do it again 2 years later... :-)
Especially when the breakdown is obviously the device itself.

Comment from jmke @ 2010/03/10

I've had laptops break down a week after their warranty has expired, sometimes you get the short end of the stick

the other 200 laptops kept working years after their warranty expired.... that's life

Comment from EsaT @ 2010/03/24

Quote:

Originally Posted by jmke

you can't decide on experience with ONE sample that a whole range of products is broken/bad.

Epsilon platform based PSUs commonly show "ripple happiness" from the start (even some advertised PSU specs breaking ATX specification) and FSP uses mostly cheap capacitors avoided by many makers.
So with such product propability of getting that short end of the stick is simply higher.

Comment from ToyTen @ 2010/04/01

I am a lucky owner of an 800w Epsilon PSU then. Since the Geforce 8800GTX and I used those in SLI with a Q600 for more than one year. My PSU is still there.
Not saying at all that problems never occur of course, but it seemed to me that a positive feedback about those PSU's was welcome here, supplying a 5870 and an I7 950 @ 3200Mhz with 4Ghz UCLK, 3 HDD and 6Gigs of memory. Also lots of fan for my water cooling.

All information and graphics contained in Madshrimps are sole property of the Madshrimps crew and may not be reproduced or copied in any manner without written permission from us.