Welcome Guest LOGIN | REGISTER
James Gorbold

Input/Output - New Custom PC benchmarks are go!

Posted at: 5th September 2007

We've updated our benchmarks in readiness for the multicore revolution - is your PC prepared?


As the deputy editor of Custom PC, one of my main tasks is making sure that we have the best tools available to measure the performance of new hardware. After all, every review in the magazine, and many of the news stories and features, are based on the all-important question of whether or not the product is worth buying.

This fundamental question can only be honestly answered if you possess two pieces of information - namely, knowing if the product performs well and how much it costs. While the latter is fairly easy to find out, usually involving a telephone call or email, measuring how well a product performs is far more complex.

This is why Custom PC has expended a lot of time, effort and money over the last four years on ensuring that we have the best tools available to measure performance. However, while most of these tools are standard off-the-shelf items, from the £10 power meter we use to measure the power consumption of PCs, to the £20,000 Chroma 6314 PSU load tester, there's one area in which we've gone it alone - our Media Benchmarks.

The first version of the CPC Media Benchmarks made its debut back in Issue 1, although it also made a sneak appearance in the sample issue that we put together to show the industry what the mag was going to be about. This version set the pattern that later versions were to follow, comprising three individual tests that combined to produce an overall score. However, just as the way that we use PCs has changed, so have the benchmarks. For example, the first version ripped several tracks from an audio CD into MP3s, encoded a video and edited photos. The next version of the Media Benchmarks debuted in Issue 12, and introduced a video encoding test; for the first time, this was multithreaded, so it was able to take advantage of multicore systems. The third incarnation of the Media Benchmarks ditched the CD ripping test entirely, since by that time (mid-2005) even entry-level PCs could rip CDs very quickly. We replaced it with a multitasking test, which performed the photo editing test while creating a file archive in WinRAR.

However, all good things must come to an end, and this month, we're introducing the 2007 version of the Media Benchmarks, which can be downloaded at www.custompc.co.uk. While the latest version still comprises three individual tests, it marks the biggest change in the history of the benchmarks. The 2007 version is easier to install and use, as it's entirely based on open-source applications, so you don't have to worry about any of the applications timing out as in previous versions. In addition, the applications don't have to be installed; they simply run out of the directory to which you decompressed them. This means that if you have more than one hard disk, or perhaps an SSD, you can simply copy and paste the whole directory, and run the benchmarks from there. Another benefit of this approach is that the benchmarks don't install data in the Registry, so they don't leave an unwanted footprint in your system, even if you already have other versions of the applications used in the benchmark installed. Unlike previous versions, the 2007 version of the Media Benchmarks is compatible with Windows XP and Vista, allowing you to determine if there's any performance difference between the two platforms.

The first test uses GIMP to open, edit and compress lots of very high-resolution digital photos, before assembling a massive panorama. This test not only tells you how well-equipped your PC is to handle digital photos, but it also puts a lot of strain on the memory and I/O sub-systems, in addition to the CPU, so it's a great way of measuring hard disk and RAM performance. The second test encodes a high-resolution MPEG-2 video taken from an HD camcorder into H.264 using Handbrake. This test is very CPU-intensive, so it runs much faster on PCs with more than one processor core. The final test attempts to play back the output file from the video encoding test while creating a massive file archive in 7-Zip.

We've also changed the way the benchmarks display the results. In the new version, the individual and overall test results are displayed as a relative score to our reference PC, which is based around a 2.66GHz Intel Core 2 Duo E6750, which scored 1,000 points. Therefore, if your PC were to score 1,200, this would mean that it's 20 per cent faster than the reference PC. As the benchmarks are freely downloadable from the Custom PC website, this means that you can easily compare the performance of your PC with other readers' machines on the online leaderboard, as well as the products reviewed in the magazine.

Not content with developing an entire new benchmark suite, we also introduced the RTS game Supreme Commander to our benchmarks. This replaces Need for Speed: Carbon as the standard game test for all motherboards, PCs and laptops. Supreme Commander is a great test to use; as it simulates huge complex battles, it responds to faster CPUs and GPUs. Of course, we'll continue to test using other games and applications, but this issue marks the single biggest change in our test software for two years. As usual, we'd love to know what you think of the new benchmarks, so why not download them now and see how your PC compares to the Jones' machine?

Submit to:  
Print-friendly version
Comments

Make a Comment

Broadband Dongles

Compare prices

Fastest, cheapest 3G mobile broadband dongles from 3, Vodafone, T-Mobile and Orange
from just £10/month

Button link to Mobile Broadbandgenie.co.uk
Powered by
Broadband Genie