What we're talking about with Niagara is six time to 15 times improvements in performance for certain kinds of workloads.

What we're talking about with Niagara is six times to 15 times improvements in performance for certain kinds of workloads.

At the top level, there's not any appreciable difference [in performance]. It really doesn't mean anything. If you were talking about going from 2000 to 3000, then yes, but not at this level. We thought this was an appropriate place for us to set a limit to protect the game from any future unknown developments.

On Friday, we had a ratio of 12 to 1 of losers to winners. That's the worst performance since 1997. We also had over 3,000 declining issues. This shows real weakness.

Out of his last three starts, this was his best performance. We've got to have him now through the playoffs.

Current accelerators costs about $15,000 each and deliver little performance improvements beyond what you could achieve by buying more blade servers for that same price. We have taken the approach that we must deliver three times the price-performance of a standard blade.