Can old PCs run Vista?

Posted at: Monday 14th May 2007 by David Fearon

How well can an aging PC handle Microsoft's shiny new operating system, and is it worth upgrading your old machine as far as it can go? David Fearon tinkers with an Athlon XP PC and a Socket 478 Pentium 4 PC to find out.

Although Prey didn't respond to a lowering of the settings, with a bit of tweaking, the system had enough basic processing grunt to provide decent frame rates in the other games. Need for Speed: Carbon responded well to switching to trilinear filtering, with the shader and world detail at Medium. After that, we saw a minimum frame rate of 25fps for longish periods, but nothing we did could avoid the low-memory-related swap-file blips at certain points, which reduced the speed to way below 10fps for 10-20 frames.

The Upgraded Intel machine

We then ripped out the old Pentium 4, graphics card and RAM modules, and replaced them with a 3.4GHz Northwood Pentium 4, four 512MB PC3200 DIMMs and an AGP HIS Radeon X1950 Pro.

The difference this made to the Media Benchmarks 2005 was impressive, with the overall score shooting up from 0.47 to 0.72. The biggest gain was in the image editing test, in which the score went from 0.62 to 1.05. The main reason for the big gain is, once again, the Achilles heel of Vista - memory. The extra 1.5GB mostly avoids the need for the OS to hit the swap file, providing a huge speed increase.

The difference to gaming was also massive. From being a disk-bound, low-detail-only gaming rig with little to boast about in the way of frame rates or gaming experience in general, spending a few hundred quid on our system transformed its gaming potential. Need for Speed: Carbon became a feast for the eyes. Basic polygon models were treated to motion blur, glow and particle effects that hugely enhanced the look of the game, and simply weren't feasible with the old card. The raw figures for the minimum and average frame rates don't tell the whole story compared to the difference in raw playability, and the entertainment factor that the upgraded machine provided. It's also worth bearing in mind that the minimum frame rates in the results were due to occasional stutters, but the high average frame rates still meant that the games were playable for the most part. Basically, if you're prepared for some occasional stuttering, particularly when the game first starts, then it could be worth upgrading your old machine; if you're a stickler for silky-smooth frame rates then you'll need a new PC. Prey continued to be a problem, though, averaging just 12fps.

The AMD machine

Our AMD test system was roughly the same vintage as the Intel machine. Those with long memories will remember the megahertz wars, the legacy of which continues today. Intel had the upper hand as far as the ability to clock the bejesus out of its silicon was concerned, courtesy of the otherwise inefficient NetBurst architecture. However, AMD had superior performance-per-clock, so it came up with the notional '+' rating; unlike Intel's parts, this didn't relate directly to the clock speed, so our Athlon XP 2400+ wasn't a 2.4GHz CPU. Instead, it was a 2GHz Barton-core chip with 512KB of Level 2 cache. AMD and Intel were close on performance at the time, with AMD reckoned to have the advantage in general. However, Intel gained the upper hand when it came to serious floating-point work, such as media encoding.