On 2025-02-15 11:27 a.m., Frank Leonhardt via cctalk wrote:
Running anything like Algol on a machine with drum memory seems a bit
optimistic!
Remove "Like Algol" and the statement is even more valid.
I guess that was why the PDP-1 was successful, it had early core memory.
I keep forgetting about drum memory, on most early machines.
Around what time did core memory drop in price that one had ample main
memory to compile with? I am guessing the late 60's.
The PDP 8/e, being simple could have a fast cycle time 1.2 uS.
I have a rather full featured, 18 bit design being run in real hardware
using 10 ns CPLD's (emulating ROM) and the best speed I can do is 1.6 uS
core memory cycle time with a 2 phase clock.5 Mhz Oscillator.
Most of the delay is with the slow 74LS219 register file and
ROM accesses time. Working backwards, ballpark timing for a 70's design
using SSI is a 2.17 uS. 3.6864 Mhz clock.
I keep finding out how the early computers got away with so little
hardware and memory and did so much with that.
I wonder how much progress computers would have made had ACSII
and Algol not kept changing standards every few years?
Some how 6 bit characters seems more standard, text wise,
compared with the mess with accented characters and money characters
of today.
Is there any thing a ALGOL compiler needs for good code generation
other than ample index and GP resisters?
Ben.
