Hands on: Save power and save money

Unfortunately, such a configuration would not necessarily be very efficient, consuming more power than necessary while also running hot, requiring potentially noisy cooling. Indeed, unless it was a high-quality unit, it may not even deliver sufficient juice on the required rails to reliably power the system in the first place.

Most decent PC power supplies perform at their most efficient when delivering roughly half their rated output, so if your components typically draw around 200W, then you’d ideally be coupling them with a power supply rated at 400W, or maybe higher still if you planned to upgrade.

The important thing to remember is that a power supply won’t constantly consume electricity at its maximum rating. It will only deliver the amount of power the PC requires at any given time, so even something as hefty as a 1,000W supply may only draw a fifth of that in normal operation.

So a well-designed power supply with a decent rating will only consume what it needs, but will do so more efficiently than a lesser model, thereby reducing power bills along with heat and noise. Indeed, the only downside in buying a higher-rated PSU is paying a little more for the unit to start with.

The test
To put this theory into practice I upgraded the power supply in my own work PC. This is based on an Asus P5W DH Deluxe motherboard with a Core 2 Duo E6700 processor, 6GB of Crucial DDR-2 memory, a Gigabyte NX66256DP (Nvidia 6600) graphics card and four 300GB Seagate hard disks, three of which are in a Promise hardware Raid 5 configuration.

The existing power supply was a 365W model from Enermax. This is a very respectable company, but the model itself was bought and fitted over two and a half years ago when I ran a single-core Pentium 4, a lesser graphics card and just one hard disk. The power supply had been more than adequate running this original configuration, but over the years I’d upgraded the processor, graphics card and added no fewer than three extra hard disks, so it was clearly under greater pressure.

With Vista booted and disk activity essentially idle, I measured 170W being drawn from the mains socket using a Maplin power meter; this rose to 209W under 100 per cent load when transcoding a video file. While the latter was still far from the maximum rated output of my power supply, I still wondered how a newer and more powerful model would cope.