Re: External RAMdisk

rzh_nocal@xxxxxxxxx wrote:
On Jan 12, 9:34 pm, Herb Johnson <herbrjohn...@xxxxxxxxx>

Interesting thread!!

Maybe folks will be interested in this? I designed a Z80 SBC (which
ended up on 2 PCBs -- long story!). It runs at 10MHz, and has a 1 MB
battery backed RAM disk (SRAM based). The battery backup is via a
Dallas chip (DS-1202?? -- not sure if that's correct), and two D cell
flashlight batteries. I mounted the whole thing in the floppy bay
(floppies long gone) of a KayPro II that I got in a silicon valley
surplus place for $20. I goosed the KayPro II up to 5 MHz, and just
use the keyboard and screen as a "terminal" for my computer. The only
hitch is that I have to use a stall routine on my computer after it
sends out a linefeed, because the 5 MHz KayPro II can't scroll and
update its screen fast enough to keep up! The "R-Pro" (as I call it)
boots CP/M from EPROM (32K). Pictures on my Web site
(www.hanscom.us), but you'll have to drill down to find them! Oh, and
it also has an IDE interface (thanks to Phil Ruston -- www.retroleum.co.uk),
and runs on a 1 GB hard drive from Weird Stuff Warehouse, that cost me
all of $9.95. I use Xmodem to transfer files to/from the "R-Pro", and
am working on getting Zmodem up and running.

This is a *VERY* responsive system when running from the RAM disk, and
only a little slower running from the IDE drive.

Roger


FWIW< I used to have what I thought was a neat "first test" for a compiler: Compile and link the null program, then call the debugger and load the file. People used to tell me that this wasn't much of a test, because all I was measuring was the overhead time of the compiler and other tools. But to me, that's the point. I try never to write large programs that have to be compiled in one huge glop. So to me, the overhead is pretty much the entire task.

To put things in perspective, running this on a VAX 11/780, with BSD Unix, took 20 seconds, and generated a 20k object file. That doesn't include invoking the debugger. This was also at night, when I was the only user.

The same program, compiled on VAX/VMS, took a full 60 seconds and generated a 50k object file.

On my Kaypro (4 MHz) with an HD, I had a script file that did the whole thing. Compiling under BDS C, it took 0.6 sec, and generated a 2k object file.

Doing the same thing on the 2MHz Kaypro with RAMdisk took 0.2 sec. That's basically one tick of my stopwatch.

I told this story once, to New Zealander John Spray, who wrote a neat little Pascal-like compiler for the 68000, called Whimsical. I was bragging on the small size of the BDS C object file. He said, "Try it on Whimsical." I did. The file was under 256 bytes. This blew me away. But John said, "There's more to it than that. If you use the right compiler flags, you can get it down to two bytes, one 68000 instruction." The instruction? RETS.

I also like one of John's dictums a lot: "The best way to optimize the object code from a compiler is to not let it generate bad code in the first place" <g>.

Jack

On my 2MHz
.

Relevant Pages

  • Re: namespaces and main()
    ... would make the compiler read type information for OtherClass ... >>object file formats used to communicate between compiler and linker ... what does "full static type inf." ... of the AST for the current translation unit, ...
    (comp.lang.cpp)
  • Re: What a translation unit is.
    ... >>So therefore it only makes sense to me that the most usefull translation ... > But let's keep assuming the traditional compiler system. ... >>something inbetween states of tranlation then the object file is most ... >>The source file is a unit to be translated. ...
    (alt.comp.lang.learn.c-cpp)
  • Re: Help! simple compile question
    ... That would depend on what compiler you use, ... sub1 calls sub2 and sub3. ... and sub3 to end up in the same object file. ... the compiled result from multiple source files in a single file, ...
    (comp.lang.fortran)
  • Re: namespaces and main()
    ... >> an object file for a translation unit which contains enough ... >> information for the linker to check if any exported type information ... You completely ignore that the main use of a header ... if a compiler would produce object ...
    (comp.lang.cpp)
  • Re: Michigan / Ohio / Indiana Help Needed
    ... readable with standard PC running any software. ... I will try to do more testing of the floppies tomorrow orSunday...but ... in Forth and created a compiler compiler to create a given system. ... discuss getting these onto a KayPro... ...
    (comp.os.cpm)
Loading