3 ms·
5mb/s is on the low side. Isa ran on the same clock as the cpu. We started there, but things got better quickly. I think the hard limit was a bit below 40MHz,
by hyperman1 3y ago
5mb/s is on the low side. Isa ran on the same clock as the cpu. We started there, but things got better quickly.
I think the hard limit was a bit below 40MHz, as people still used 74HC chips for glue logic. Combined with EISA 16 bits = 2 bytes per cycle this might give you a 80MB/s theoretical EISA peak. I had an EISA based 386 running 40MHz at the time.
Interestingly, VESA's 133MHz is double this, so maybe the majority of the speed gain was doubling bus width to 32bit?
- rasz 3y agoWe started at 8 bit 4.77MHz with 4 clocks minimum so 1MB/s. We ended at around 12MHz in some crazy Turbo XT/AT clones, anything faster was back to ~8MHz with fancier BIOSes giving you option to overclock. You are thinking cpu fsb, ISA never ran at 40MHz. Even EISA ran at a fixed ~8MHz. Btw EISA is not those longer black ISA slots on normal boards, EISA is brown slots with too many pins between normal pins because its secretly a 32bit bus, found pretty much only on workstation/server boards from 1989-1994.
- hyperman1 3y agoYou made me curious, so I looked it up: https://en.wikipedia.org/wiki/List_of_interface_bit_rates https://en.wikipedia.org/wiki/List_of_interface_bit_rates It lists: The original case: ISA 8-Bit/4.77 MHz - 0 W/S: every 4 clocks 1 byte , so basically what you say. The best case; EISA 32-bit/8.33 MHz : 33.32 MB/s So a lot worse than VESA, but still better than 5Mb/s