Could you do one thing just to test another approach? Please export the RAW file to TIFF and push the shadows in the resulted file. I saw very good results someone shared here using this method.
Mathematically this should not give any difference as noise is already in the image.
You might see improvements in SNR a little bit but only if during this conversion from RAW to TIFF some small per pixel details are lost e.g due to down sampling
I didn't find the message with sample images that lacked visible noise, but here is another message about the same thing:http://www.techradar.com/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/canon-eos-5d-mark-iii-1074186/review/page:5#articleContent
the techradar test:QuoteTIFF images (after conversion from raw) have a consistently good signal to noise ratio across the sensitivity range comparing closely to the Canon EOS 1Ds Mk III and just having the edge over the Nikon D800. It also shows a slight improvement at all sensitivities over the Canon EOS 5D Mk II.QuoteTIFF files (after conversion from raw) have a high dynamic range with results comparing closely to the Nikon D4 and D800. Compared with the Canon EOS 5D Mk II, there is a marked improvement showing over 2EV greater range at the lower end of the sensitivity scale.
i thought techradar is a DxO certified lab?
I do not know if TechRadar is certified DXO lab, I think they just use some DXO equipment but how they use that and how they interpret results I do not know. Have seen that article before but did not trust much what they were publishing .
I more rely on pure Math - all the time in my professional life that was first stage in any systems development. And then implementation results were compared against initial math results to see if design was OK or something was missed or done wrong - initially or at later stage.
So as you suggested I did noise comparison for original RAW and TIFF (lossless converted from RAW, to16 bit TIFF, color space is the same Adobe RGB, no re-sampling - the same pixel count).
Noise measurements were done using Noise Ninja plug-in PS which gives NUMERIC noise profile of the image.
For initial RAW noise with shadows raised by +3EV index is 56 (Luminance=15, Chroma=41)
For RAW converted to TIFF and then shadows raised by +3EV noise index is 54 (Luminance=15 and Chroma=39).
As expected these are same results for lossless conversion with normal statistical distribution as Noise Ninja was selecting areas for noise profiling automatically and they a bit different.
See below results two 200% crops for RAW and TIFF and that snapshots of NJ noise profiler screen
Anyone who has NJ plug-in in PS can do the same and see numeric results