Truth is, these letters are related to symbol rates in MHz - they are Roman numerals!.
D = 500. the first D-PHY spec in 2009 said HS speed was intended to be 80 to 1000Mb/s. The upper limit has since been raised.
M = 1000. The first HS Gear was 1.248Gb/s (HS-G1A) or 1.4576Gb/s (HS-G1B). Later specs added HS-G2 and HS-G3 (and stay tuned for HS-G4).
C = 100. C-PHY is derived from D-PHY (except it doesn't forward a clock, it is embedded) but since it transfers 2.28 bits per symbol the symbol rate for the same data transmission rate can be less than half as much as D-PHY would need. According to the latest draft C-PHY spec it is "intended to define a solution for a symbol rate range of 80 to 2500 Msps per lane, which is the equivalent of about 182.8 to 5714 Mbps per lane."
It is interesting to note that C-PHY balances the current in 3 wires, does not need an extra pair of wires for clock, and achieves speeds up to M-PHY's HS-G3 but at a symbol rate 1/2.28 of M-PHY's bit rate. It reuses D-PHY's low speed mode ... I think in some cases C-PHY could be viewed as a substantial improvement over real D-PHY.
Thank you Joe. For clarifying the "C" in CPHY. I was guessing on the naming convention given the heavy focus on CSI/DSI support in this revision to DPHY. I agree with you that CPHY offers great promise to increasing DPHY throughput without hanging more symbol-based overhead on.
C-PHY took a departure from the D-PHY and M-PHY naming convention. The C doesn't stand for "100 mbps", it stands for "channel limited". As in, it can run over lower quality interconnects (at least compared to what D-PHY or M-PHY would require). This presumably makes it cheaper to implement.