Shameless Words from Shameless Geeks

Was the infamous Duke Nukem vaporware actually released today? Can the granddaddy of all gaming jokes be at an end?

Naaaah. We can relax. It’s only a re-hash of the old Duke Nukem 3D title for XBox Live Arcade. They almost had me believing that Duke Nukem Forever was actually released. I suppose that I should have known better. The universe still exists, and hell is… well… still hell. Please go back to fragging.

OK… so maybe it’s just another study that relies on “expert” opinion to derive it’s numbers. In order to avoid being a hypocrite, I have to say that I don’t believe the results any more than I believe the studies that tell me how bad video games are for development.

That being said, it’s nice that psychologists “opinions” are beginning to balance out. Someday, I’d like to see the end of opinion based studies. Until that magical day comes, I’ll settle for a world where the “opinions” balance out to a no-op.

I just saw a /. post entitled PCMark Memory Benchmark Favors GenuineIntel. I’m a little concerned about how quickly the crowd went straight into “Intel paid them to cheat” before thinking about other possibilities. I’m not trying to say that Chipzilla is virtuous in this matter, but lets be fair about it.

First of all, how many of the libraries that they use are written by Intel for promotional purposes? I can remember (many moons ago) writing some DSP and graphics code for DirectX. At the time, the best code for the job was the Performance Library Suite that was available for free. Naturally, it was used in a lot of places. The problem was that it was written by Intel for the purposes of pushing their own chips. Each level of the Pentium (yeah, it was a while ago) ran it’s own pre-compiled library. Anything not identified as a Pentium defaulted to software emulation of some of the SSE features. This, of course, led to a skewed result when using the libraries for comparison.

Secondly, it is ultimately the responsibility of a vendor to make sure that their code doesn’t skew results. If their code doesn’t work properly on some CPU’s, they should clearly indicate so. I haven’t used PCMark2005, so the software may already have some kind of indication built in. A feature like this would have warned the user not to trust the results of the test.

Lastly, VIA should take an active role in making sure that they don’t get shortchanged on tests. How about a library that takes advantage of the chip’s strengths? Maybe even partnering with the spec vendors? It seems to me that they would be foolish to trust that a secondary chip vendor would get equal treatment without having someone help things along.

Honestly… the whole big-business-is-cheating conspiracy theory thing is getting old. It really doesn’t sound good when I hear it coming from fellow geeks.