>>>>>And too bad the DEC Alpha wasn't popular enough. I saw it run some version of NT in 1995, and it was way ahead of what x86 could offer at the time. Too bad it didn't survive.
>>>>
>>>>Never had my fingers on those - multitasking I can imagine being more sophisticated than 386 single core. Which areas impressed you?
>>>
>>>The screen :). I said I only saw it. The girls at the show were doing their regular stuff - editing documents, inserting graphics etc, and that was waaaay faster and smoother than we saw on the Pentium boxes that year - and it was running the same Windows NT.
>>
>>For a while I ran NT Server 3.51 as my main dev box. It was a P133 with 32MB RAM. For NT having enough (i.e. lots) of RAM was key. I think the required minimum was 8MB but 32MB was a realistic minimum, otherwise the system spent all day paging to disk. Trouble was in those days RAM wholesale was $50/MB so just the RAM for my system was $1600. A lot of people weren't willing to spend the extra money for the extra RAM so NT ran slowly for them.
>>
>>In 1995 NT4 hadn't been released yet, I imagine what you saw was NT 3.51 with the Windows 3.x GUI. NT4 had the Windows 95 GUI.
>>
>>The DEC workstation you saw probably had tons of RAM and fast SCSI hard drive(s). IIRC DEC priced them from a mainframe/mini POV rather than a microcomputer POV, which explains why they weren't "popular enough". ;-/
>
>I finally switched over to x86 on a 386 with 8 MB of RAM in the late 80s (after some 8088 Turbo Pascal I went with 68000 for at least 2MB of RAM) - RAM alone would have bought a nice used car. Ran a non-GUI version of OS/2 on it, which was a speed demon, as extended DOS specs were not established fully back then. Was able to port all the Motorola Pascal and Fortran direct array access stuff using linear memory access. Extended DOS later allowed nearly same compiled stuff to work on DOS, but dev work was much smoother with OS/2 command line multitasking. NT4 on a 32MB years later was good enough to make me move (then under Modula2 for data access replaced Pascal and C replaced FORTRAN for bottled statistics routines and .dbf access), but dislike for GUI never left me totally.
>
>Give me a problem, data and define type of access (disc or mem) and leave me alone ;-))

I first ran Turbo Pascal on a CP/M Jim Ferguson Big Board I, souped up to 5MHz with a Z80B and 64K of 200ns static RAM (cost $200 somewhere around '82 or '83). Ported some programs from that over to the company's IBM PC, performance was surprisingly similar (or maybe not surprising, original PC was 4.77MHz). The PC had bitmap graphics so I could do some graphical analysis/output (the BB was text-only and used a VT100 terminal or emulator), I switched over to that pretty quickly then bought one for myself.