On Sun, Aug 07, 2016 at 02:38:09PM +0200, Gerrit Heitsch wrote: > On 07/31/2016 04:48 PM, Segher Boessenkool wrote: > >Same with VIC-II: the 6566 is a very nice and clean design (what I can > >reconstruct from it anyway, I haven't seen an actual die photo)... The > >DRAM versions, not so very clean. > > The interesting part would be to compare the 6566 and 6567 dies to see > what was changed and how much is the same. Where did they squeeze in the > refresh counter, the RAS/CAS logic, were things moved around for it... http://segher.ircgeeks.net/vic-ii/8565-blocks.jpg (that is 8565, not *entirely* the same as 6567/6569, but I labeled the blocks here). This is 6569 (R3): http://retronn.de/imports/cbm_chips/vic2_overview_gray4.jpg and this is R1 (warning, huge): http://mail.lipsia.de/~enigma/vic2r1/fullvic_lowquality.jpg (not much has changed in R3). Refresh counter sits next to the VM counters (it also writes the low address bits, and the VM counters were handily on the end of things already). Here's a piece of the pinout, including all changed signals: 6566 6567 17 phi0 phi0 18 phiin #RAS 19 phicol #CAS 20 Vss Vss 21 A0 phicol 22 A1 phiin 23 A2 A11 24 A3 A0/A7/A8 25 A4 A1/A8/A9 26 A5 A2/A9/A10 27 A6 A3/A10/A11 28 A7 A4/A11/A12 29 A8 A5/A12/A13 30 A9 A6/A13/1 31 A10 A7 32 A11 A8 33 A12 A9 34 A13 A10 So, phicol and phiin were moved (colour clock and dot clock inputs). Note the weird routing of the current colour clock stuff, and the wacko block at the start of the signal. Those would make a much neater layout in the original location. Note all the address bus routing on the new layout; a lot of it is very ad-hoc and strange. Things would fit great with the original pinout. The clock gen / redriving and the colour buffer ("C bits") seems very untidy as well, but I cannot figure out how it should have more logically been before. Btw, the RAS/CAS things are self-timed, using analogue delays, long inverter chains etc. I want die pictures of the 6566, too :-) Segher Message was sent through the cbm-hackers mailing listReceived on 2016-08-07 20:00:13
Archive generated by hypermail 2.2.0.