That's a good question, basically related to the question, what is the optimal size of the academic science community? Keep in mind that part of this growth is because other countries have significantly increased their science funding, so more of these papers are being contributed by countries that previously did not contribute on the scale of the US, UK, Japan, and few continental European nations.
I'm of two minds on this: on the one hand, there is a LOT of good research left to be done out there, and you could have a significantly larger scientific community still doing productive research - which means that we just have to resign ourselves to the fact that nature is huge, and it's a lot of work just to keep up with a tiny corner.

On the other hand, many journals aren't selective enough, and a lot of useless stuff gets published. The NIH is more selective than most journals (although I don't know what the optimal acceptance rate for a journal should be) - in any given grant application cycle, 7%-20% of the applications typically get funded, and a large number just get triaged because (in most cases) they've got major deficiencies. So the government is being fairly selective.


It would be helpful to have more numbers - are individual labs publishing more, on average, or are there simply more labs publishing. For the bests labs, costs are probably going up, because the latest technology is expensive (although again, it would be great to see some numbers), and today's genomics research is much more reliant on 3rd-party technology than cutting-edge molecular biology was 20-30 years ago.

So I don't know.

I didn't count, but my educated, off-the-cuff guess would be... 90%. 
OK, maybe not, but they're in love with GWAS.