One good idea ids to ask the tech support at the various vendors a complicated question. Most will not have a front line person who can answer, then see how long it takes before they come back to you.
It does not help much to have the best tools in the world if you have to sit idle for two weeks before you can continue without that error you can not figure out why it is posted.

I have found that the variance in the quality of tech support is much more important than the variance in tool performance.

Of course, you will not suffer a 100% increase in code size to gain a better tech support, but the difference between all I have seen with one exception is well within the acceptable. Actually in one test a came out 8% better than b, while in another b came out 10% better than a.

Do not ask me to publish my findings, the analysis of 5 toolsets was done 5 years ago and the figures may have changed dramatically.

Again, if the support is bad, what good does it do that the tool is good.

Erik