OK, long post warning.
SACD of course uses DSD, and that is not PCM, but is roughly equivalent to 8 x 176.4 PCM (I wrote it that way on purpose).
Yep it is typically encrypted, but is available over HDMI, and there are AV receivers with HDMI input and true DSD-decoding DACs. They are typically multi-channel. I don't know of any 2-channel standalone HDMI DACS that handle DSD. Perhaps they exist, but I don't care, as I now explain.
Ask yourself -- how was the music recorded and mastered? The answer is, at 192 or 176.4 typically. OK, not always ... but if you care about 2-channel stereo, careful downsampling (decimation) of DSD to PCM at 176.4 is all you need. Now you have PCM on SPDIF to feed to your DAC, to your soundcard, whatever. You have freed the music from the SACD jail. And unless your SACD is magic, the sound will be as good this way as it will be listening to the SACD player analog output.
This is even true of DSD downsampled to 88.2, IMHO on most commercial SACDs I have tried. In theory 88.2 is not enough, but I don't hear a difference.
I have a Wadia 781i, considered one of the finest SACD players in the world. It also outputs 88.2 PCM from the 2-channel SACD layer. And as I have explained, I can use the PS3 to get a 176.4 PCM rip from the same layer, using a Sony-engineered algorithm.
Now I can listen 5 ways:
1. The redbook layer of the SACD, at 44.1, direct from the Wadia.
2. The redbook layer of the SACD, ripped at 44.1 by the Wadia, and played back from a computer, fed to the Wadia DAC, or to one of my external DACs.
3. The 2-channel SACD DSD layer, ripped at 88.2 by the Wadia, and played back from a computer, fed to the Wadia DAC, or to one of my external DACs
4. The 2-channel SACD DSD layer, ripped at 176.4 by the PS3, and played back from a computer, fed to one of my external DACs (the Wadia can't handle this).
5. The 2-channel SACD DSD layer, direct from the analog outputs of the Wadia.
So the theory says 5 > 4 > 3 > 2 > 1 (2 > 1 because nothing is spinning, so there are fewer bitstream errors reading the PCM in real-time, and if care has been taken with the soundcard, DAC and cabling then there are no noise or timing problems introduced by the PC).
To me, 3,4, and 5 always sound the same. And 1 and 2 always sound the same. I do think that (3,4,5) sounds better than (1,2) on a number of SACDs. [Techincal note: 4 is new to me, and there are some computer problems with the files, so I have only limited flying time with 176.4, but I still stand by my conclusion, even at 88.2].
My conclusion -- you lose nothing ripping SACDs. My hearing might not be good enough (I'm 61). My system is ultra-resolving, so the only issue is my hearing ... we'll be repeating these tests sometime in 2010 with younger members and golden ears.