More Chart Sneakiness

Okay, I cannot claim credit for this. It was all George Ou's work. I did say that this editorial is not about bashing AMD for making those incredibly-positive tables. However, he has discovered something else about the charts provided by AMD that I think you should not miss - another way to mess with the charts. This time, it involves selectively picking data to skew the chart in favour of their own product.

He posted about this in his recent ZDNet blog post. Apparently, they "omitted the better Intel results because of an optimization technique called auto parallelization". As a result, the Quad-Core Opteron was only appeared 5% slower than the Intel Xeon even though it was really 13.4% slower.

Of course, AMD had their reasons why they omitted those better results. However, the point really isn't about why they omitted the better results, but the fact that they did so. This only goes to show that you just cannot trust charts that companies provide. Here's a quote from his blog post :

AMD used these slower Intel E5345 results which had auto-parallelization turned off instead of using the best Intel E5345 results which would have meant AMD’s 2350 should have trailed by 13.4% and not just by 5% as shown in the slide. Nothing at the live event disclosed this omission.

Again, we would like to stress that this is not a practice exclusively restricted to AMD. Many companies do it. It is up to us to look at these charts with a properly-critical eye. Look past the bars and see the actual numbers. It won't be easy, unless you redo the charts yourself. On our part, we will do our best to correct all of such "optimized" charts when we can.

Questions & Comments

Please feel free to post your questions or comments here!