All, Somewhere at home, I still have a couple of the FAMOUS TI MAGNETIC BUBBLE MEMORIES. Hows that for nostalgia? Regards, Jim On Wed, 19 April 2000, "Quitt, Walter" wrote: > > WOW! > > All this of old computing. It brings back memories of > core cache. Talk about electro-magnetic tiny little > things. > > Wire wrapped interface boards to PDP/11-45s. > > Ah, the late 70s, early 80s.... > > Sniff... > > Walt... > > -----Original Message----- > From: Russell McMahon [mailto:apptech@CLEAR.NET.NZ] > Sent: Wednesday, April 19, 2000 5:19 AM > To: PICLIST@MITVMA.MIT.EDU > Subject: [OT] Ah yes those were the days (but this is now)... > > > I copied my comments on paper tape punches and switch programmed SC/MPs to a > friend and this was his response to me. No PICs therein but it contains a > good old reminisce (and he's some 10 years younger than me AFAIR). I thought > others may enjoy a similar opportunity or wish to comment on any of this. > Must be about time for another walk-to-school-in-the-snow-uphill-bothways > story or two. Goflo can tell us about his arcane collection of older iron > etc. I have some core memories here, and an IBM ??? with a "program" read by > the multiple fingers on a motor driven rotating PCB selector and wire spring > memory and .... > Then there's mercury bath memory and CRT phosphor memory (yes memory!) and > ... > An vacuum tube character generators that scans an anode through a stencil > which contains images of the characters to be displayed and ... > And the magic 4004 / 8008 / 1702 / 2708 / 2716 triple supply (wow!!! 2kB > !!!!) > 2758 - wow!!! - only ONE supply ... > 2114? > .. > .. > . > > > Notice the Litton drum computer he had (has?). An amazing beast to us and > seems so archaic but what are rotating magnetic disks going to look like to > our grandchildren? The Litton had even the accumulator on the drum, written > multiple times to allow fast access, and many many heads to allow wide > access. > > > RM > > +++++++++++++++++++++++++++++++++++++++ > > Russell, > > Yes I fondly remember assembling my first computer (A Cosmac in my case) and > shortly thereafter a D2 kit (officially an MEK6800D2). The Cosmac system > was very basic compared even with the D2 but did have one fascinating > uirk - the static memory device (silicon-on-sapphire) was capable of > retaining its contents for weeks without power. I too could write M6800 > machine code from memory (I still remember that BRA = 20H, BNE = 26H, RTS = > 39H, etc.). > > I also helped a friend build a Scamp evaluation board (which I never warmed > to) and later bought the first 6809 device (for my thesis - it arrived too > late and was never run - I still have it) and the first 68000 processor > ever sold in NZ. The data sheet for the 6809 (which I also still have) was > so preliminary it was partly written in longhand (official Motorola > document!). > > I also recall the hand-made paper tape reader (made from two pieces of black > Perspex with a shallow groove for the tape cut into the surface of one using > a Black & Decker router and holes at the right pitch drilled right through > to take Motorola photo-transistors. It used ambient light (an Equipoise > lamp was best) and the tape was pulled through by hand. Hooked up to a PIA > on the D2 it worked a treat. > > The back of my thesis (dated July 1980) includes the complete listing for a > radix-2 FFT for 8-bit complex data written (using the D2 kit) in association > with Ross McMillan towards the end of 1978. It could operate on data sets > up to 256 points in length and included ancillary routines for bit-reversal > of data and data smothing (via convolution with the impulse response for a > raised cosine). It is possibly the earliest FFT ever implemented on a > microprocessor (certainly I have never seen reference to anything earlier). > Regrettably my thesis does not document the execution speed but I recall it > was quite respectable compared with the hardware (hard-wired and bit-slice > microprogrammed) FFT processors we were building at the time. It > exclusively used self-relative addressing (so was fully relocatable) and > coded entirely by hand (no access to an assembler back then). It occupied > 687 bytes of code plus 256 bytes for the sine and cosine twiddle-factor > tables. A driver program (not included in my thesis) occupied most of the > remaining 81 bytes of code space - it read sample data from a home-brew ADC > and outputted the transform via a home-brew DAC for viewing on a scope (the > ADC and DAC were built on Veroboard using CD4000-series CMOS with a > binary-weighted resistive ladder assembled from hand-picked 5% resistors). > I remain very proud of the FFT code it to this day. > > The there was my Litton - now there's a machine for you. A serial CPU > inplemented in TTL with ALL storage (CPU registers, program code, and data > memory) on separate tracks of a rotating magnetic drum. It had originally > been used for accounting purposes by the Egg Marketing Board. Peripherals > included a drum type "teleprinter" (a bit like an IBM golfball typewriter), > a paper-tape puch and reader, and twin floppy drives (8" hard-sectored with > 80kb per drive). It could be programmed using switches and lights on the > fron panel of the processor unit and programs could be saved to or restored > from paper tape using a short boot loader entered that way. The floppies > could only be used to store data and were driven directly by the users's > application program (no BIOS here mate). I wrote a 0's and X's program for > it but didn't do much else with it. By rights I should still have the drum > memory but my wife accidentally (or so she claimed) threw it out. The same > fate shall not befall my prized ferrite-core memory module ! > > But that was then (sigh). > > Back to the present - have you had time to peruse the 87LPC768 data sheet, > and have you had time to play with http://surpluslistings.com ? I don't > recall seeing any commentary from you on either topic. > > Regards, > > Ken Mardle jim@jpes.com