www.computerbase.de suggests in a review that ATI maybe cheating in connection with doubtful filter algorithm optimisations.
Unfortunately it seems ? based on these knowledge ? that most of the benchmarks published on the Web are not comparable and virtually useless.
Read the entire review:
My personal opinion:
When comparing the performance of products, these products should perform identical/comparable operations. If an ?undocumented? optimization is used the results are no longer comparable. In such a case a chip vendor may suggest having a faster algorithm, but claiming to have the fastest hardware may be misleading.
Consider a 200m dash and someone is running a shortcut. Would he win the Gold Medal or would people rather lough at him? Remember Matrox Mystique turning vertical sync off to look faster without even being able to filter bilinear. However, the reaction pattern of the chip vendors will never change. They will most likely issue a statement calling the ?cheat? ?the latest patent-pending blessing offering great value for the end user?. Amazing, isn?t it? (wt)
Del med dine venner