> > + {3, 1536, 16000, 24576000, 0x20, 0x02, 0x10, 0x35, 0x8A, 0x1B, 0x1F, 0x3F}, > > + {3, 1625, 8000, 13000000, 0x0C, 0x18, 0x1F, 0x2D, 0x8A, 0x0A, 0x27, 0x27}, > > + {3, 1625, 16000, 26000000, 0x0C, 0x18, 0x1F, 0x2D, 0x8A, 0x0A, 0x27, 0x27}, > > + {3, 2048, 8000, 16384000, 0x60, 0x00, 0x00, 0x35, 0x8A, 0x1B, 0x1F, 0x7F}, > > + {3, 2304, 8000, 18432000, 0x40, 0x02, 0x10, 0x35, 0x8A, 0x1B, 0x1F, 0x5F}, > > + {3, 3072, 8000, 24576000, 0x60, 0x02, 0x10, 0x35, 0x8A, 0x1B, 0x1F, 0x7F}, > > + {3, 3250, 8000, 26000000, 0x0C, 0x18, 0x0F, 0x2D, 0x8A, 0x0A, 0x27, 0x27}, > > }; > That seems bit confusing, can't you just have something like > coeff_div_v0 and coeff_div_v3 and select one depending on target? Using one clock makes it easier for users to compare the register differences between the two versions of the chip. It is also easier for developers to check whether the register changes are correct or not.