Originally posted by ringu61
Futuremark = Can of Coke
NV = Guy with a Straw
ATi = Guy with Mouth
Normally people drink can of coke with mouth straight away, ok ATi drink like that too. Nvidia drink the can of coke with straw, ok why not, just another way of drinking.
WHy can't NV work around with their driver to make it it run better on 3DMark03 without quality reduce or watever. Can u even tell with ur naked eyes without looking at the scores and ****?
This is a really weird analogy... but I'll take a stab at a reply.
It's a chugalug contest... the prize... more consumers are apt to buy your product if you win.
The Rules... Human power only.
Obviously... guy with straw can't chugalug as fast. Guy with straw attaches a pump, powered by electricity, to win. That pump only works with coke, sprite, dr. pooper, and mountain dew (More will come as they tweak the pump, and get other soda's not to jam the pump). The pump also isn't human powered, hence cheating (As far as the coke chugalug contest rules... not cheating for sprite, dr. pooper, mountain dew, etc... different rules).
Mouth works with everything... and also works better than the pump in some cases.
Enough of the analogy... 3DMark delegates a simulated workload. A lot of that workload is unseen (like massive overdraw). It measures performance based on that simulated workload. If you "work around" that simulated workload (Like hand-shader replacements or shortened clipping planes)... you are not doing the FULL workload even though you aren't affecting the image, hence the measurement is Invalid, and does not give a TRUE performance comparison.