As promised, as soon as the results and split times from Bolt's unbelievable 9.58s performance were available, we'd be analysing them
Turns out they became available rather quickly (not 9.58s kind of quick, but quick nevertheless). So here is the analysis of Bolt's race.
All the splits
First of all, here is the table published by the IAAF this morning, containing all the splits and times for 20m intervals, for both the semi-final and final:

It reveals how Bolt achieved the fastest split at all intervals, which is amazing considering his supposed slow start. He led the race at 20m, and continued to grow that lead. I have shown his times on the chart:

The relatively slow first 20m (because of the start) makes interpreting the last 80m of the race quite difficult, so I've taken the liberty of looking only at the last 4 intervals, for a more obvious comparison between Bolt, Gay and Powell. It makes the point that Bolt was dominant from start to finish, as shown below.

Translating those times into speeds is a simple but interesting exercise (shown in the graph below), because you then perceive just how fast the top end speed is. For Bolt, the fastest interval (60m to 80m) was run at an average speed of 44.72km/hour. If you're wondering how that compares to last year in Beijing, read on...

The next graph shows the gap between Bolt and Gay and Powell for each 20m interval. I felt this would be interesting to look at because it would show whether Bolt did all the damage early and then held the lead, or whether he grew the margin progressively through the race. It turns out to be the latter option - he builds a lead at each interval. It's 0.03s/20m for the first 40m (over Gay, that is), and then Gay makes something of a comeback (but never actually closes the gap, he just limits its growth), before losing another 0.03 seconds in the final 20m.
For Powell, the margin grew steadily throughout the race, and opened up at an increasing rate after 60m. The comparison is interesting because it serves to highlight:
a) How incredible Bolt was, from start to finish of the race - who'd have thought he'd lead at 20m?
b) How well Gay did to hold onto Bolt as much as he did

A comparison between Berlin and Beijing
Now, for the comparison everyone has been making already - how does Berlin 2009 compare to Beijing 2008?
I must confess that I'm a little skeptical of the Beijing figures - at the time, there was some contention about the accuracy of these splits, because they were worked out from TV footage, whereas the latest Berlin result is a very specific research study, and thus (you'd think), more accurate. However, it's an interesting comparison, but just keep in mind that there may be some error in the Beijing numbers (you'll see what I mean below)
First the split times. I've looked at Bolt in Berlin, Bolt in Beijing, and then Gay in Berlin as well.
What it shows is that Bolt's first 20m in Berlin was actually slower than his first 20m in Beijing (now you see why I'm skeptical). Anyone who saw last night will testify that Bolt started brilliantly in Berlin, and so I can't believe he was slower to 20m. However, from then on, he's faster, and by 60m, is 1/100th of a second ahead of "himself in Beijing", then four hundreths by 80m, and adds another 0.09s to that gap in the final 20m.

As for the Berlin Top 3 comparison, here are the gaps at each 20m marker, but this time Bolt's performance in Beijing has been added (in green). Positive numbers mean Bolt is ahead (as was the case the whole way in Berlin!) and negative means behind, as the splits suggest for 20m compared to his Beijing run. So Bolt was behind at 20m (by 0.02s), slightly ahead at 40m, then grew the lead, to 0.04s at 80m, and ended up 0.11 seconds ahead of his Beijing time at the finish line.

Once again, here are the speeds based on the available times - they show that Bolt in Berlin hit a higher speed at 60 to 80m than in Beijing. I've indicated Bolt's speeds on the graph, and the smaller number at each point is the speed of Tyson Gay, just for comparison purposes.

There is a lot more to be said, I unfortunately have a pressing deadline, so I'll have to say it later!
Until then, thanks for the comments!
Ross
P.S. One thing I didn't mention in the original post (and should have) is the difference in wind speed between Beijing (0.0m/s) and Berlin (0.9m/s tailwind). So here is a comment from Rob, which pretty much summarizes the impact:
One thing that I think needs mentioning is the effect of the tailwind, which was 0.9 m/s in Berlin and 0 m/s in Beijing. This might account for why Bolt/Gay in Berlin had higher top speeds than Bolt in Beijing -- at least based on your graph results (the other factor is the Bolt might have already started celebrating before the 80m mark in Beijing!).
Secondly, if we adjust for the affect of the 0.9m/s tail wind [Ross: this depends a little on which 'formula' you use to adjust for win, but it illustrates the point well enough, and the numbers are about right], Bolt would have run 9.62 instead of 9.58, which isn't much different than a possible 9.64 in Beijing had he run through the finish line. Tyson Gay's adjusted time in Berlin would have been 9.75.