I downsampled all of my high res audio to 44/16 using MBIT+ in the iZotope RX software. It saves a massive amount space, with no sonic downside that I can hear using the DX50. I'm beginning to believe high res playback for practical uses is little more than a gimmick.
Even if you are a staunch supporter of high res playback, if you had to choose between native high res playback with deeply flawed codec (DX50 stock firmware) or downsampled playback with a flawless codec (Rockbox), which would you choose? My suggestion would be to try both before reaching any conclusions.
I used both, I have had RockBox on iPods before so it's not that I never tried it but I also tried 24/96 on the DX50/DX90/X5 and I can truly feel the difference, it's clearer, more detailed. By the way let's not get into that hi-res war, for me 24/96 is very good, less it's a downgrade, more is a waste that's just my personal preference, yours can vary and I respect that.