Now Trident's fingered over 3Dmark01SE results

We now think that Trident can be added to the list of potential witches and we're inspecting its body closely for the tell-tale "Devil's Bench Mark".

A few weeks ago, we received a letter from a reader and he'd bought a notebook powered by Trident's XP4 - a promising chip that never made it to the wider world of the desktop marketing dorks, although it looked pretty good on paper.

Our friend tested the Toshiba Tecra M1, using the Trident XP4 and a Pentium-M 1.6 GHz with 512MB RAM versus his dad's Dell Latitude D600 with a 32MB, 64-bit (most are 128-bit) Radeon 9000, PentiumM 1.6GHz and 512MB RAM and WinXP.

His score dropped by half if he just changed name of 3dmark01SE to, let's say, inquirer.exe.

After we received this email, we contacted Trident and got its official explanation on this. The Trident executive told us that Trident hardware has 50 per cent of the transistors of its competitors but still delivered about 70 per cent of its competitors' performance.

To achieve this performance grade it invented a "proprietary 3D algorithm optimized for multiple pixel rendering" powered by "implement proprietary software + hardware support" that is meant to accelerate the 3d feature that was considered as a bottleneck by Trident.

So what Trident did is that it designed its driver to use the names of popular apps of interest to the market.

But if you you change the name of the application there will be no performance optimisation for that program.

Trident clearly stated: "This is not cheating in the driver, it is selective performance optimization of popular applications of interest by using an unique combination of software + hardware algorithm proprietary to Trident."

Our advice to Trident would be to try to recognise frame streaming in an application. That is not a hard task to do and it can ask Nvidia how to do this since it has some experience in this field.

So you now you know, don't you? µ