14 Nov
2005
14 Nov
'05
8:06 a.m.
Allison wrote:
Everyone has done the PC emulating whatever. Why not
take a
single chip micro and use that to emulate another micro as
hardware? Same idea maybe slow but for a lot of things speed
is not the whole world.
I did it on paper for PDP8 and it was possible using an 11mhz
8049 to come within 1/10 the speed of a real 8. I guess a
36bit machine could be done as well (though really slow.).
Also a front pannel could be programmed into it as well
(maybe a speed hit). Any cpu that gets an instruction from
memory and then executes it can be emulated that way.
1/10 the speed of a real PDP8? Really? That is fast indeed!
My emulation of a PDP8, running on a 68B09 (at 2 MHz) gets to
approx 1/60 of the real thing. I measured that by running some
diagnostics. The diag says "bell rings after approx 5 seconds),
and timed how long it took on my 68B09...
I am sure that when I remove the front panel code things run
faster. The front panel requires time to update. But running
the emulation without the front panel is no fun.
- Henk, PA8PDP.
This message and attachment(s) are intended solely for the use of the addressee and may
contain information that is privileged, confidential or otherwise exempt from disclosure
under applicable law.
If you are not the intended recipient or agent thereof responsible for delivering this
message to the intended recipient, you are hereby notified that any dissemination,
distribution, or copying of this communication is strictly prohibited.
If you have received this communication in error, please notify the sender immediately by
telephone and with a "reply" message.
Thank you for your cooperation.
14 Nov
14 Nov
6:28 p.m.
Gooijen, Henk wrote:
1/10 the speed of a real PDP8? Really? That is fast
indeed!
My emulation of a PDP8, running on a 68B09 (at 2 MHz) gets to
approx 1/60 of the real thing. I measured that by running some
diagnostics. The diag says "bell rings after approx 5 seconds),
and timed how long it took on my 68B09...
I am sure that when I remove the front panel code things run
faster. The front panel requires time to update. But running
the emulation without the front panel is no fun.
Having got a new PC ... 2 GHZ??? who knows the real speed.
I have tried Spare time gizmos - pdp 8 emulator ( A minor bug
-- with windows how do you get the bell to sound ?).
I was running some sort of diagnostics and had the RTC displayed
and for about 10 minutes of real time, the clock advanced a hour.
That must be at least 6x faster than the real thing on this computer.
- Henk, PA8PDP.
7:33 p.m.
woodelf wrote:
...
Having got a new PC ... 2 GHZ??? who knows the real
speed.
I have tried Spare time gizmos - pdp 8 emulator ( A minor bug
-- with windows how do you get the bell to sound ?).
I was running some sort of diagnostics and had the RTC displayed
and for about 10 minutes of real time, the clock advanced a hour.
That must be at least 6x faster than the real thing on this computer.
If a 2 GHz (give or take) x86 CPU emulates a pdp 8 at 6x, it means either the
code is inefficient, or the code contains a speed regulator that doesn't work
properly.
The original PDP-8 took 10 clocks at 1 MHz to execute one instruction, a 2 GHz
CPU has 20 million cycles to interpret one instruction. So at 6x realtime, the
program is using 3.2 million cycles to interpret one instruction.
Most likely speed and efficiency weren't goals of the emulator, so I bring this
up not to discredit the program's author but rather to say: don't use that data
point as anything but a lower limit on what kind of horsepower it would take to
use a micro to emulate a PDP-8. I imagine an AVR device at 20-40 MHz should be
able to emulate a PDP-8 at real time.
8:01 p.m.
Doing some math, 2GHz/100KHz is 2000MHz/0.1MHz or 20,000. At 6X the
cycle count is then 20K/6 = 3333 (not 3.3M).
One would think it takes well less than 3000 X86 instructions to emulate
a PDP-8 instruction (assuming one instr per clock).
Jim Battle wrote:
woodelf wrote:
...
Having got a new PC ... 2 GHZ??? who knows the
real speed.
I have tried Spare time gizmos - pdp 8 emulator ( A minor bug
-- with windows how do you get the bell to sound ?).
I was running some sort of diagnostics and had the RTC displayed
and for about 10 minutes of real time, the clock advanced a hour.
That must be at least 6x faster than the real thing on this computer.
If a 2 GHz (give or take) x86 CPU emulates a pdp 8 at 6x, it means
either the code is inefficient, or the code contains a speed regulator
that doesn't work properly.
The original PDP-8 took 10 clocks at 1 MHz to execute one instruction,
a 2 GHz CPU has 20 million cycles to interpret one instruction. So at
6x realtime, the program is using 3.2 million cycles to interpret one
instruction.
Most likely speed and efficiency weren't goals of the emulator, so I
bring this up not to discredit the program's author but rather to say:
don't use that data point as anything but a lower limit on what kind
of horsepower it would take to use a micro to emulate a PDP-8. I
imagine an AVR device at 20-40 MHz should be able to emulate a PDP-8
at real time.
8:14 p.m.
Don North wrote:
Doing some math, 2GHz/100KHz is 2000MHz/0.1MHz or
20,000. At 6X the
cycle count is then 20K/6 = 3333 (not 3.3M).
One would think it takes well less than 3000 X86 instructions to
emulate a PDP-8 instruction (assuming one instr per clock).
I would say it takes a few instructions since you have to mask for 12
bit words, and remember the PC is running windows and all
sorts of other stuff. The point is the emulation is at least 6x faster
or more compared to
the original. Faster may not allways be better for emulation.
Jim Battle wrote:
9:11 p.m.
New subject: Grid 1520, 2 Data I/O 212's
Hey all, just rescued a grid 1520 laptop (looks like it has an addon
batterypack on the top) and two Data I/O 212's (one of them includes a
small data card plugged in to the side).
I have no need for them, so whoever wants them just needs to pay shipping.
I'm always looking for stuff in trade as well though if anyone wants to go
that route. ;)
Marty
11:53 p.m.
Don North wrote:
Doing some math, 2GHz/100KHz is 2000MHz/0.1MHz or
20,000. At 6X the
cycle count is then 20K/6 = 3333 (not 3.3M).
One would think it takes well less than 3000 X86 instructions to emulate
a PDP-8 instruction (assuming one instr per clock).
eh, what is three orders of magnitude between friends. :-(
nevertheless, the point was there are thousands and thousands of cycles at 2 GHz
to do a straight-forward interepreter. the ratio of host to emulated
instructions I used for my AVR estimate was 200-400 per emulated instruction,
which is probably overkill because if you are using an AVR, you are also
probably hand coding things, so the count doesn't have to be nearly so high.
10:31 p.m.
Jim Battle wrote:
The original PDP-8 took 10 clocks at 1 MHz to execute one instruction
really?
I'm probably off, but at a cocktail party I would would have said that
some instructions took 8 clocks (f0,f1,f2,f3,e0,e1,e2,e3) and some took
12 (f0-f3, d0-d3, e0-d3).
That's according to a recent look at the "blue book", but I may have
misunderstood. I did write up some verilog which uses those states and
it seems to be (mostly) correct, but it's done all debugged yet.
-brad
11:57 p.m.
Brad Parker wrote:
Jim Battle wrote:
I'm no PDP-8 expert. My source was this web page:
http://fixedreference.org/en/20040424/wikipedia/PDP-8
"The PDP-8 was a 12-bit computer with 4096 words of memory. It had only eight
instructions, one full register, the accumulator (AC), and a single-bit
register, the link (L) bit. The machine operated at a clock rate of 1 MHz, and
took 10 clocks for each instruction, so that it ran at 0.1MIPS."
The original PDP-8 took 10 clocks at 1 MHz to
execute one instruction
really?
I'm probably off, but at a cocktail party I would would have said that
some instructions took 8 clocks (f0,f1,f2,f3,e0,e1,e2,e3) and some took
12 (f0-f3, d0-d3, e0-d3).
That's according to a recent look at the "blue book", but I may have
misunderstood. I did write up some verilog which uses those states and
it seems to be (mostly) correct, but it's done all debugged yet.
-brad
15 Nov
15 Nov
12:16 a.m.
On 11/14/2005 at 5:57 PM Jim Battle wrote:
Time to add some commentary to the wiki, methinks.
IIRC, the cycle time of the original straight 8 was 1.5 usec. Instruction
times in cycles is given here:
http://www.cs.uiowa.edu/~jones/pdp8/refcard/65.html
"The PDP-8
was a 12-bit computer with 4096 words of memory. It had only
eight
instructions, one full register, the accumulator (AC), and a single-bit
register, the link (L) bit. The machine operated at a clock rate of 1 MHz,
and
took 10 clocks for each instruction, so that it ran at 0.1MIPS." 
12:27 a.m.
Chuck Guzis wrote:
On 11/14/2005 at 5:57 PM Jim Battle wrote:
Time to add some commentary to the wiki, methinks.
IIRC, the cycle time of the original straight 8 was 1.5 usec. Instruction
times in cycles is given here:
http://www.cs.uiowa.edu/~jones/pdp8/refcard/65.html
That seems to line up with the PDP8 manual I have.
Gordon.
"The
PDP-8 was a 12-bit computer with 4096 words of memory. It had only
eight
instructions, one full register, the accumulator (AC), and a single-bit
register, the link (L) bit. The machine operated at a clock rate of 1 MHz,
and
took 10 clocks for each instruction, so that it ran at 0.1MIPS."
10:58 a.m.
On 11/14/2005 at 5:57 PM Jim Battle wrote:
Time to add some commentary to the wiki, methinks.
IIRC, the cycle time of the original straight 8 was 1.5 usec. Instruction
times in cycles is given here:
http://www.cs.uiowa.edu/~jones/pdp8/refcard/65.html
I dug out the code for my original PDP8 emulator, which was written in
assembler on an 8MHz 8086 PC. In order to give the CPU a chance to
actually execute some instructions between interrupts from the emulated
teletype, I bumped a counter every time I went round the main loop. I
found that about 100 instructions gave a suitable (maybe a bit slow) 10cps
terminal speed.
That suggests to me that about 1000 instructions per second is possible on
a slowish machine.
Gordon.
"The
PDP-8 was a 12-bit computer with 4096 words of memory. It had only
eight
instructions, one full register, the accumulator (AC), and a single-bit
register, the link (L) bit. The machine operated at a clock rate of 1
MHz,
and
took 10 clocks for each instruction, so that it ran at 0.1MIPS."
12:59 a.m.
Brad Parker wrote:
Jim Battle wrote:
Well that number now that I think about it is way off, from memory
speeds in the
PDP -8 Faq. The handbook lists the pdp-8/e as 1.2 us for jump and
operate instructions.
2.6 us for memory instructions.+1.2 us for indirect.+1.4 us for
auto-indexed. A 5 mhz clock
looks to be used. I/O seem to have a variable time but 2.6us is the
general speed listed.
The original PDP-8 took 10 clocks at 1 MHz to
execute one instruction
really?
I'm probably off, but at a cocktail party I would would have said that
some instructions took 8 clocks (f0,f1,f2,f3,e0,e1,e2,e3) and some took
12 (f0-f3, d0-d3, e0-d3).
I downloaded a PDP-8 maintance manual (pdf) once and give the full
details of every time
state. The machines of that era of tightly coupled to the core-memory cycle.
That's according to a recent look at the "blue
book", but I may have
misunderstood. I did write up some verilog which uses those states and
it seems to be (mostly) correct, but it's done all debugged yet.
-brad
PS. Finds tinyfoot note. PDP-8,PDP-8/I 1.5 us memory cycle. PDP-8/L
1.6 us PDP-8/S 8us
16 Nov
16 Nov
12:20 a.m.
woodelf wrote:
Well that number now that I think about it is way off,
from memory
speeds in the
PDP -8 Faq. The handbook lists the pdp-8/e as 1.2 us for jump and
operate instructions.
2.6 us for memory instructions.+1.2 us for indirect.+1.4 us for
auto-indexed. A 5 mhz clock
looks to be used. I/O seem to have a variable time but 2.6us is the
general speed listed.
For the pdp-8/e, it is actually a 20MHz clock driving a 28 bit shift
register from which 4 timing states are generated. The memory cycle is
1.2us.
12:30 a.m.
C. H. Dickman wrote:
For the pdp-8/e, it is actually a 20MHz clock driving a 28 bit shift
register from which 4 timing states are generated. The memory cycle is
1.2us.
Now if DEC used that clock speed for marketing, I'd bet they would of
sold more 8's. That is a fast clock.
Hmm 200 Mhz would give you a .120 us and 2000Mhz .012 us :D
17 Nov
17 Nov
12:13 a.m.
C. H. Dickman wrote:
And they could have claimed the 11/45 was a 33MHz machine (that is the
frequency of the master clock crystal). It always amused me that my 11/45
had the same clock frequency as 386 PCs some 25 years later.
Of course as _we_ all know, clock speed is not a particularly useful way
of determining the speed of a machine.
-tony
For the pdp-8/e, it is actually a 20MHz clock driving a 28 bit shift
register from which 4 timing states are generated. The memory cycle is
1.2us.
Now if DEC used that clock speed for marketing, I'd bet they would of
sold more 8's. That is a fast clock. 
14 Nov
14 Nov
11:48 p.m.
On Mon, 14 Nov 2005 13:33:01 -0600
Jim Battle <frustum at pacbell.net> wrote:
woodelf wrote:
...
Well, probably on said modern PeeCee, you could run eight or ten instances of the emulator
without bogging the machine down. What degree of 'nice' (I know it doesn't
exist in windoze, just wondering about the equivalent thing) does the emulator run at?
The PC probably still has all the dancing, flashing foolishness that gatesware implies
running in addition to the emulator.
Having got a new PC ... 2 GHZ??? who knows the
real speed.
I have tried Spare time gizmos - pdp 8 emulator ( A minor bug
-- with windows how do you get the bell to sound ?).
I was running some sort of diagnostics and had the RTC displayed
and for about 10 minutes of real time, the clock advanced a hour.
That must be at least 6x faster than the real thing on this computer.
If a 2 GHz (give or take) x86 CPU emulates a pdp 8 at 6x, it means either the
code is inefficient, or the code contains a speed regulator that doesn't work
properly.
The original PDP-8 took 10 clocks at 1 MHz to execute one instruction, a 2 GHz
CPU has 20 million cycles to interpret one instruction. So at 6x realtime, the
program is using 3.2 million cycles to interpret one instruction.
6991
days inactive
6994
days old
16 comments
11 participants
participants (11)
-
ak6dn＠mindspring.com -
ard＠p850ug1.demon.co.uk -
bfranchuk＠jetnet.ab.ca -
brad＠heeltoe.com -
cclist＠sydex.com -
chd_1＠nktelco.net -
chenmel＠earthlink.net -
frustum＠pacbell.net -
gordonjcp＠gjcp.net -
henk.gooijen＠oce.com -
wgungfu＠csd.uwm.edu