Financial Inversion, cont'd

But the PDP did not need a PC nanny to rin a SDK.

But the PDP did not need a PC nanny to run a SDK.

And the C story:

You are constructig a contradiction that never was there in the first place.

My C comment was in the line mentioning the PiPiCo and for a good reason was not a statement about which compiler was in use on the PDP11.

But the PDP did not need a PC nanny to run a SDK.

And the C story:

You are constructig a contradiction that never was there in the first place.

My C comment was in the line mentioning the PiPiCo and for a good reason was not a statement about which compiler was in use on the PDP11.

(End of thread for me.)

The PDP11 raw processor needed an external box of electronics from which to load programs and in that respect was no different to the RPi Pico. Such external boxes would be associated with storage media such as paper tape or spinning disks.

Eh? Wot? Was English not your mother tongue?

Well I don't follow that but there's no reason why the RPi Pico could not run a C compiler and then store the result in off-chip storage.

Having used a PDP11 back in the day, with I think 128k of RAM, the C complier (and the C CROSS compiler we had on it, to generate 6809 code) were both far more primitive than today's offerings.

Its almost impossible to compare like for like.

ISTR that the PDP11/20 to which I was introduced 52 years ago as an undergrad intern had 16K (32KB) of core store.

But no high speed reader punch; it took over 1/2 hour to load the assembler itself via the 10CPS ASR teletype paper tape reader!

No wonder that as a prototype nerd (hacker as it meant in those days) I found machine code entry via the blinkenlights to be more rewarding :-)

The basic PDP-11 architecture limited addressing to 64 kilobytes, with 8 last kilobytes reserved for I/O, leaving 56 kilobytes for program use.

Well the one I used had ran Unix, had 64k of program memory *and* 64k of data memory an IIRC a 20Mbyte hard drive, and we accessed it via serial consoles.

Late 1980s that was. We used PCS running probably DOS 2 or thereabouts as serial terminals and as text editors as 'vi' was clumsier than WordStar.

... and the Unibus was rated up to 56 ft long, the transit time putting ot out of the running for today's processing speeds!

Talking of speeds, ISTR calculating the slowest instruction (apart from RESET which took 50 mSecs) to be ...

BICB @2(R1), @4(R2)+

... which with all the memory accesses including the write-back cycle took something like 17 uSecs.

This is certainly true - you could however almost certainly port an OS of the era (say Tripos or an early Unix) to run on the Pi Pico and result in an experience fairly similar to a PDP-11 era mini.

UNIX in a total of 128K? The kids of today with their VDUs needing several MB of video storage (3 bytes for each colour of each pixel) just won't believe you!

Did the pdp11 have an MMU? or was it an option? I believe there is a non-mmu version of Linux. So with loads stripped out it might be a goer.

I programmed a CTL Modular 1 in 1970 in assembler. It had an MMU and multiuser interactive OS but can't remember how many Kb it had. I think

It was a pretty "modern" computer. Unlike the various PDP it had very few flashing lights and no control panel. It had a built in boot program read in from a fairly fast papertape reader to get it going.

Prior to X windows, that was not far off what SCO Unix ran on. I remember 'Venix' on a 286.

Cant remember how much RAM, but less than a MB I am sure

Likewise SUN SPARC stations.

Unix was created on a PDP-11.

Yes. It was

"Some models, beginning with the PDP-11/45, can be set to use 32K words (64 KB) as the "instruction space" for program code and a separate 32K words of "data space". Some operating systems—notably Unix since edition V7, and RSX11-M+—rely on this feature."

Wiki

There seem to have been three possible memory models, the original 64k map, the twin 64k maps as described above, and full MMU allowing up to I think 256K bytes.

Which suggests that *early* linux could probably run well on a Z80 :-)

According to Dennis Ritchie[1], Unix was, in 1969, cross-assembled on a GE 465 running GECOS, targetted for execution on a spare DEC PDP 7. It wasn't until

Yes true - I should have said /portable/ unix was created on the PDP-11 but then I tend to think that unix didn't really become unix until then.

It is always difficult to fit gradual analogue evolution into the frame of Boolean thinking so beloved of hoi polloi.

The stupid man asks 'is it dangerous?' The intelligent man asks 'how dangerous is it?'

When I started this train of thought, musing upon my experiences amongst the undergraduate apprenti (colloquial plural used in Westinghouse, Chippenham) I was talking about 52 years ago in 1971.

I find it to be salutary to think about all that had been achieved in the previous 52 years taking us back to 1919 when even valves (tubes to the Yanks) were largely unheard of.

Looking at the technical advances in the 52 years on from 1919 it is no surprise that great strides were made in the next 52 years from 1971 taking us from the single processor integer-only 16 bit PDP11/20 to the multi-processor RPi 4B with shedloads of on-chip memory and not only floating point capability but also vectored forms of floating point with 64 32 and 16bit processing all contained in a single silicon chip.

When I've finally got around to refubishing all the old clocks that are waiting my attention I might at last tackle the RPi400 that I bought 2 years ago for some 64 bit assembly language :-)

And on the (OT) subject of horology how the prices of long case (grandfather) clocks have fallen when you can get a good one for £200 when 30 years ago you had to spend £2K!