8 Jun
2010
8 Jun
'10
12:21 p.m.
On Jun 7, 2010, at 7:59 PM, Jerome H. Fine wrote:
The current design has an mmu. I has all the guts for a full I&D 11/44 style mmu.
I didn't go to the limit and add all of the 11/70 features, but may some day.
It's currently configured as a 11/34 style mmu with no split I &D, using defines.
Well, the basic clock is 50Mhz, which is 20ns. Each instruction takes a minimum
of 3 clocks. The most complex take about 20 clocks. So on average I think
it's about 5 clocks/instruction. My last stab said about 10MIPS.
So, I think it will beat a 11/44, but won't compare with simh on a modern pc.
The disk controller makes an IDE disk look like a single RK05 drive. You
can just copy a simh rk05 image onto any ide disk and go.
Brad Parker
wrote:
I am quite confident that TSX-Plus requires an MMU to run. If you can confirm,
when I get to testing the TSX-Plus code, I can produce an RK05 image for you
as well. Well, as I
said, the no-mmu version runs at 50MHz. I could improve that. The mmu version probably
won't run faster, mostly due to the 20ns rams on my fpga board.
Can you translate that in terms of the speed of a standard PDP-11? When you used RSTS/E, did you run only basic programs
or
did you also use the MACRO.SAV program?
I used basic+, as well as Whitesmith's C under RT-11. Most
of the TSX-11 work I did was in Pascal, with a little bit of macro
driver work. I also did a lot of fortran on a a standalone 11/23
in a geology lab using RT-11. I wrote a great spacewar clone in
fortran which used a scope with z-control :-) brought all the grad
student's progress to a complete halt, which didn't make me popular
with the dept. head.
I'm afraid I can't help with .Chain semantics.
-brad
Brad Parker
Heeltoe Consulting
781-483-3101
http://www.heeltoe.com