Re: "Taking over two minutes to load a 64 kilobyte into memory was maddening."
It shouldn't have taken that long. According to "Commodore - A Company on the Edge", Robert Russel considered the disk access on the VIC-20's serial bus to be too slow. So for the C64, they replaced the 6522 VIA (with its broken shift register) with a 6526 CIA and added some high speed lines to the serial port. Same deal with the 1541. It would have been 20 to 30 times faster than the VIC-20.
Problem was, during the period when prototype boards were being reworked into their final production layout, somebody removed the high speed lines. The decision to reuse VIC-20 cases for the C64 imposed some serious space challenges. Some engineer mistakenly thought they weren't being used, so they were omitted to save space on the board. Since the Kernal's serial routines hadn't yet been updated to utilize the new burst mode, nobody realized the mistake until several hundred thousand motherboards had been manufactured. By then, it was too late.
When marketing said that the 1541 also had to be backwards compatible with the VIC-20, that limited how the serial routines could be written even further. There was also a looming deadline. So Bob Russell wrote the routines over a weekend to stay on schedule. They worked, but they sucked. Programs such as Epyx Fastload and CMD JiffyDOS worked by replacing the serial routines with optimized versions.
Anyone who has used a 1571 or 1581 on a C128 knows how much faster they are. That's what the 1541 on the C64 was supposed to be like. But due to a few bad decisions, the 1541 ended up dirt slow.
Biting the hand that feeds IT
