Seafog the 3rd wrote:
And how old are you Acolyte? Early 40's? I wasn't born yet when man last walked on the moon. I'm 29, and waiting rather impatiently for a permanently manned moonbase.
I've been waiting longer. Yes, I'm 41. A bit old to still be an acolyte, but some of us are slow learners, I suppose.
Mirober wrote:Alpha's (if I understand correctly) were some of the first computers to use a 64-bit architecture (something I'm apparently going to become intimately familiar with, as I help in porting the 32-bit code from the VAXen).
They were more like one of the first 64-bit micro
processors. There were earlier ones in the mainframe world. I used to work on a CDC Cyber... um... 960 I think, which was a few years older. Funny thing about that machine is that if you were using the older NOS instead of the latest and greatest NOS/VE operating system, the upper 4 bits of each word were locked off which left you with the same 60 bit word as on older Cybers.
You will note 60 is not divisible by 8. We used octal, and 6-bit bytes. Yes, that was a bit limiting, but it was close to I was used to at the time. The mainframe my college had for us to work on was a DECSystem-10, which had a 36-bit word. This was the early 80's. The college had a VAX, but it was for research only. If you weren't a grad student with a thesis to complete there was no way you were going to get an account on it. Hell, the only reason I had a timeshareing account on the DEC-10 was because I was a CS major. Eveyone else had to drop their cards in the bin and pray.
Alphas were actually being made until just about a year ago or so. Following a lawsuit a lot of that tech got licensed to Intel, which used it in the Opteron. Screwed it up along the way of course, but that has a lot to do with why HP (which bought out Compaq, which bought out DEC) is porting VMS to the Opteron, and why Compaq/HP chose to terminate the line at this time.
Our own VAX to Alpha conversion effort was about 10 years ago now, and I don't remember what the major issues were. Some system calls behaved differently. There were alignment issues. You had to twiddle the linker differently to get Fortran commons and C globals to map onto each other properly. You have to be careful when reading in binary floating-point zeroes from files written by the VAX. (Alpha has this amazing FPP, but it's real cranky about floating point numbers not in canonical form that the VAX would just deal with.) Other than that, I think it went pretty smoothly. C ints still default to 32 bits, as do Fortran INTEGERs, so we didn't have much of that kind of thing to deal with. There's even a compiler that lets you run VAX MACRO on the Alpha with minimal tweaking. It was still an effort: when you're moving a couple of million lines of code from one platform to another even issues that are relatively rare can crop up with some frequency.
Geez, listen to me. I ought to have a beard and suspenders. Well, I've got the beard anyway...