"Bling," like LED fans, Techflex sleeving, and modularity serve to improve the attractiveness of the product. To keep power supplies flying off the shelves, it takes more than just another SMPS slapped with ATX holes. Today, we take a look at two Power Supply Unit (PSU) from Silverstone and FSP. Both are characteristically different and aimed at rather different market spectrums.

In the next page, we'd take a brief look at the two and do a little pictorial walk-through. Prior to that, keep in mind that this article is not a shootout, but a curious investigation of what is on the shelves right now.

Let's start off by looking at the power figures that determines the stature (and price) of the two items. Silverstone's Olympia is rated for an amazing 1k2W, while the FSP Everest comes in at a comparatively shy 600W. These power figures can get misleading for a very simple reason - the value is calculated from the sum of all the supply rails' capacity. This means that one can grab a power supply that promises insane amounts of power and still not get a PC to function simply because there exists a weak rail incapable of feeding the PC in question. Of course, there is the problem of manufacturers overrating their PSUs. However, that by itself, is another story altogether.

In that case, is the 1k2W Olympia twice as powerful as the 600W Everest? In absolute terms, assuming neither PSU makers are liars, total power capable of the Olympia is indeed double that of the Everest without accounting for manufacturing variablity. Now, look carefully at the two PSU labels on the boxes and the differences begin to show. First of all, the Olympia has a single 12V rail against the Everest's four. Secondly, the combine 12V rail is capable of 1080W on the Olympia, which makes it considerably more than the 864W afforded by the Everest.

Here comes the twist: FSP states the maximal combined power of its 12V rails to add up to a mere 575W. This is a far cry from the sum of individual rail capacity added up. It would become apparent in our later tests how power ratings do not actually reflect real world performance. Apparent in the sheer high current capability of the Silverstone is a penchant for power-guzzling graphics accelerators and multicore CPUs whose Voltage Regulation Modules (VRMs) buck off a 12V rail.

What use is all the power if they do not exit the PSU? FSP chose a modular approach, using captive leads for regular, often-used connectors. Extra cords are supplied should the user intend to power more components. With over a hundred amperes on hand to handle, the Olympia took the no-frills approach with captive leads on all its outputs.

Size-wise, the less muscular Everest measures 165mm short, and the Olympia some 220mm long. With all that mass, potential users of the Olympia should ensure their cases have a support strut to deal with rotational forces ready to tear off case rear panels.

Somewhat oddball is the use of a 16A IEC socket rather than the 10A one found on the majority of conventional PSUs. Known as the IEC320 16-20A plug/socket arrangement, its larger contact area allows greater current delivery from the mains. Whether the impressive looking power inlet is a mere marketing tactic or a serious necessity remains a question. Assuming 80% efficiency with 1k2W power draw, 1k5W will pass through the mains inlet. Maximal current (on 110V sockets) amounts to 13.6A, theoratically justifying its use. Never lose the cord supplied, for the last IEC320 20A plug I bought set me back a cool SGD20 (around 14USD with the current tumultuous exchange rate). On a separate note, the supplied cord does seem to lack grip on those nickel plated pins.

A 14cm fan draws air into the Silverstone while a 12cm unit does the job for the FSP. Bling-haters rejoice! Neither fans are lighted even though the transparent Yate Loon of the FSP looks suspiciously so.

In the following pages, we will dissect the two units and peek at what makes 'em tick. First victim was the smaller FSP which was subdued in no time and put to sleep.

Like most of today's PSUs, a single fan draws cool air from the bottom-face, impinging on the heatsink fins before passing heated air through the perforated rear. FSP has supplied some mugen-style black chrome thumbscrews for vanity's sake. The fan is thermally activated, making the Everest a passive power supply when idling on low loads. Like the Silverstone, the Everest is autoranging, quelling worries of wrong voltage settings.

Velcro cable ties are supplied alongside a molded powercord for the FSP. Bridging cables are all sleeved, with the secondary PCIe power connector sporting a red wrap. Goes well with ATi cards eh?

Popping the lid reveals a clean layout, with the modular sockets sitting on a PCB at the outlet end of the Everest. Thermals are taken care of with simple heatsinks that are nontheless effective. Where insulation is secondary, heatsink compound is sparingly applied. Mechanical support also includes adhesive on the toroidal chokes.

Capacitors used appear to be low ESR offerings from Taiwan manufacturers. Understandably, these capacitors were engineered with value in mind. Time will tell if they are a compromise over brandname Japanese offerings. The SMPS transformer is neither obese nor anorexic, appearing to be well constructed and capable of longterm operation. For the imaginative, there exists a lone trimpot on the PCB.

The Silverstone put up an Olympian struggle before we finally got the screwdriver on it. A peek in a bonnet revealed some thoughtful engineering deserving of the flagship identity.

Construction was on par with the FSP with notably higher rated devices. Thermal contact on the Olympia was often done via silicone thermal transfer sheets. Had insulation been secondary, heatsink compound would have been superior. Outputs are derived from a bundle of wiring that has been soldered directly onto the PCB.

To accomodate the hefty 12V single rail, SMPS transformers used are on the fat side. Fortunately, big is beautiful when it comes to conducting current. Chokes used on the output filters are huge too, with thick motor wire to ensure minimal resistive loss.

Attention is paid to mains noise, with common mode chokes spotted right at the input. With 1k2W to pump, it had better not trip the house every now and then. A pair Panasonic capacitors handle high voltage filtering. Sad to say, neither the Olympia or Everest adopted high temperature 105 Celsius rated parts. Again, Taiwanese capacitors lined the output.

Power supplies used in computers have to power multiple components that draw huge currents. The only method to efficiently regulate voltages, is to switch them. Switchmode Power Supplies (SMPS) make up PSUs we know of. Even then, efficiency varies. The industry is literally "turning green" today. One step for power efficiency, and one step on the competitors' face. First up, a simple setup to simulate an average enthusiast "rig" consisting a lowend Core 2 Quad or an overclocked Core 2 Duo with a single GPU. This review will feature Asus' latest X48 board with its energy processing unit (EPU), which is a programmable digital interface with VRM regulation and CPU performance calibration features.

Consisting Asus P5E3 Premium and a low-clocked Kentsfield at 1.2V on Vcore, Sappire HD3850 and 2GB or DDR3 memory, this system isn't the fastest around. It is made to represent generic setups that gamers afford and use. Here are the standby and idle readings of the two.

We've included standby readings of the two not to compare efficiency, but rather to show that power is still drawn when a system is shut down but not taken offline. This minor cost can add up in the long run, so do your part for the environment by shutting down and flicking the switch. 5VSB is still supplied to USB ports, logic and the like, hence the power draw. On idle, it is visible that the FSP commands better power efficiency, saving near 20W of power on the desktop.

Prime was run with four threads, and 3DMark06 looping to simulate a multithreaded game in botmatches Two Seagate SATA 7200RPM drives and an inactive Pioneer DVD writer formed part of the load.

Again, we see the FSP gaining higher grounds on efficiency over the Olympia. Pure muscle? We shall see. What it does to a higher end setup. Next page!

With 1.375V on the Vcore and CPU at 3.33GHz, we ran Asus's 790i Ultra motherboard with a triplet of Inno3D 8800GTX graphics accelerators. G80 are infamous power guzzlers, so we load 'em all 'PUs. The Everest was not designed for such a setup, being lower-rated than NVIDIA recommendations. We pulled out the adapters anyway just to see how each one fares in different scenarios.

By now it is apparent that the FSP holds its front against the Silverstone where efficiency is concerened. A 30W lead is saved with the setup in idle. What about brunt then? You ask. We loaded the system on both PSUs and the Everest halted operation as a safety precaution after a few runs. The Silverstone ramped up fan speed, making turbulence audible.

The last reading we captured on the Everest prior to cutoff was 760W. Assuming 80% efficiency, this would have meant that the PSU was putting out over 600W of power, meeting manufacturer specifications on worst case condition. The Silverstone ran the setup stably for hours. Wanting more, we ramped up Vcore just to test the capabilities of the Olympia.

Much to our surprise, the Olympia shut down after 3 loops of Canyon Flight. The power supply was unhappy about the extra 12V load on it's purported 90A rail. A quick scan of the power meter revealed the peak power draw to be 836W. With Silverstone's efficiency figures, the power capability of the Olympia seems to fall short of its claims.

Rail stability has always been an issue of concern, as it affects the stability of the system. HDDs especially, suffer error and permanent damage of chipsets when subjected to abnormal or noisy rails. Here we see the deviations between the two. The Olympia has a higher-than-normal 3.3V and 5V rail.

Under load, the Olympia droops by a fair magnitude on the 3.3V. RAMs (Graphics and System) derive power from regulators bucking off the 3.3V rails. Next we look at the 12V rails.

Both units sport healthy 12V rails. Putting them under load was no issue, even for the Everest, which sustained a reasonable voltage under SLI attack. The below was taken before cutoff.

It is clear that power supply construction on the whole has improved across the board, as much as we understand that the two samples come from manufacturers with track records of producing quality products. FSP especially, has been heavily involved in the OEM business. Under intense testing, the FSP proved its worth against the pricier Silverstone, which let itself down under intense 12V load.

Since efficiency of SMPS in general improves when loading occurs near the maximal continuous rating, it is possible to save significant power by using a PSU that offers headroom without excess flab. Big numbers do not indicate the practical power reserve, with the Silverstone being the case in point. Your pick will depend on your components and daily application. Gaming on 2560x1600 and putting your PSU up to the challenge with the rig above, where a power saving and cost efficient setup would've been the least of your concerns, with the main objective being to power up your monstrous system.

What about that HTPC rig running in the hall? The Everest with a little capacitor-swapping sounds will definitely good. High efficiency, small form factor and a good price, there simply is't much other factors to consider.