Nice!

Thank you.
I got in the new 5V power supply and it is 50% bigger than the old one, and POWER to spare.
Testing now a .750 uS clock cycle.

Ordered parts for a .650 uS clock cycle, the design limit for this computer.
This includes bigger bypass caps.
Now In the ballpark of PDP-8 memory speed here.
Most of my time will be now on software and mico-code revision.
I want to see if I can do simple 36 bit design and that means limited character operations. The current design
was designed more for Small C.

Things have got too messy,so I need to start fresh. A 1.00 uS clock or .868 uS clock is planned.
I am short a few 16 pin sockets so I plan to chop down some 20 pin sockets. This will keep me busy over
the week end, as I build new pcb's,

I found some lose 16 pin sockets, so I got the cpu board made.
Timing is .918 uS, for a 1.8 uS emulation. Faster may not work, so
need more testing. .833 uS is the next trial. Revised the order code to have ,Read Switch Register,
and made the PC (p) a valid register for load. Added base registers E..L,
load store only, giving more a 1960's feel.
It may not compete with the PDP 11, but it does compete well the IBM 1130.

A plcc socket cracked, so that PCB is toast.Testing using the other hacked PCB.
Gone to 4.77 Mhz, for PC speeds This build seems stable so far.

Found a 4.9Mhz oscillator, in my junk box and now testing with that. Removed the pc as valid destination. The irq service call, now jams 007000,2 into the IR, and state counter,permitting a push of the pc on the stack,

Back 4.77 Mhz. Gone to 504 sectors of 768 bytes, for 378Kb 8" floppy with a simple FAT file system. ~8 Kb used for file buffers and 8Kb for stack. EEProm is Kb and 8Kb expected for the Shell. 32 Kb used just for a floppy OS. Another 8K is needed if I use a 1 Mb hard disc. 192 x 2 heads x 4 sectors.
.

768 byte sectors is an interesting size. Is that including space for a preamble? Formatted sectors are often a power of two.

The SD card routines read/write standard 512 8 bit bytes per block. That gets gets converted internally to read 256 x 9 bits.
1 Kb sectors work just a bit better for a simple 1 block directory over 768 bytes per block. Now I can have 8.3 file names rather than
7.2 file names.
Looking for Rat C compiler source (A C compiler 1st ed).
Also looking for a simple file system algorithm.
Not having a compiler is big problem.

Juggling numbers so I can use a 5 meg disk in the space I have. Block size is 1.5 K
32Kb for stack, fat and buffers. 8Kb for the shell.8 Kb rom for i/o and basic functions.

Had to re-think the memory mapping since I would be using 8K word memory cards.
Rom is a 1-4K word rom card. Upper memory is 2-8K cards.Lower memory 4-8K cards.
1K blocks. Now I can move forward with the software.

A major milestone, I have the file routines finished, but not tested.
Console IO routines like printk,and gets, read a octal number have been debugged.
Now I need write routines to format the drive,read the fat table,read a directory and other utilities.

I can never seem to get the file routines finished myself.

What language are the routines written in? There are file routines in the C standard library. It is necessary to write about five or six basic routines to get things going.


Have you seen the FatFs library? It might have some routines that could be duplicated.

I have 9 bit wide bytes. Everybody expects 8 bits in software.

The end goal is to be self hosting,so I am limited to just assembler at the moment.The Small C is port is mostly done.
Small C generates rather large code, so I have no rom space for a file system programed in C.
I really can't find any good file system algorithms around the 1970's. After that you only have the crappy (need a more polite word here) Unix and MS-DOS file systems. CP/M and others are floppy sized media.

I can read, write and delete files now.The C file interface still needs testing.
I may make some hardware changes as well.