DEC Alpha
ฝัง
- เผยแพร่เมื่อ 28 พ.ค. 2024
- In the 90's the Alpha processor was one of the most powerful CPUs you could lay your hands on. With a number of operating systems available for it, from both DEC (VMS, Tru64) and 3rd parties. It even have the most successful none x86 port of Windows NT. For many the Alpha was then next machine up if their x86 PC just did not have the CPU power they needed.
BN: In the intro I say the PDP-11 is 8bit, its not its 16bit. That was a slip of the tongue, I even wrote 16bit in my script, is that what I said no, did I spot it after watching through the video many many times, no.
Information about Sixel github.com/saitoha/libsixel
This video is sponsored by PCBWay (www.pcbway.com).
0:00 - Introduction
0:14 - A word from our sponsor
0:23 - The background
1:30 - VAX
3:05 - VAX Cluster
4:05 - Terminals
5:32 - ReGIS & Sixel (the best thing most people don't know about terminals)
6:51 - Cisc/Risc
7:40 - Prism
9:51 - Riscy VAX
10:38 - Alpha
11:18 - Alpha Personal Workstation 500 Tour
16:03 - VMS on Alpha
16:34 - Tru64 on Alpha
17:21 - Trouble at mill
18:39 - Infamy infamy they've all got it in for me
20:19 - The state of the market
21:04 - Compaq buys DEC (but Alpha grows market share)
26:14 - The demise of Compaq
26:53 - End of the road for Alpha
28:13 - What did DEC ever do for us ?
29:56 - Thanks for watching - วิทยาศาสตร์และเทคโนโลยี
I should mention Linux could also emulate x86 code on Alpha, but most code was complied for Alpha, things like Netscape and Acrobat used x86 emulation.
Not many know... For some ephemeral time, actually Compaq tried to really push Alpha to the middle market at least.
They actually sold kits to "assemble your own Alpha PC", called... Well... AlphaPC kits. These came with an ATX form factor motherboard, and an Alpha CPU for it. These boards were compatible with SIMM EDO and later DIMM SDRAM standard PC memory. It had PCI and ISA slots... Well, actually was indistinguishable from a late 90s PC board except for the Alpha CPU and some chips in the board. These boards could also run SRM firmware to run VMS/Unix/Linux... Or AlphaBIOS ARC firmware to run Windows NT 4, the upcoming at time Windows 2000 (NT5), and Linux (not common config for linux to run). Even when Compaq finally killed Alpha altogether, it seems they kept producing AlphaPC boards for some time, which ended basically as the computers of some big Scanning and Printing systems like Canon, Fiery and Konica ones.
I still feel future for PC wasn't MIPS, PPC or even ARM as it is today, but Alpha. How different would be the world if Alpha could become the real successor of x86? Is was really a shame it got killed by Intel's Itanic lemon. But at least there is a relief on the fact some tears of the Alpha CPU soul outlived the fate of both DEC and Compaq in the hands of AMD... In the form of the Athlon CPU and cpu to chipset communication bus protocol, the EV bus.
The “future” we have today is multiarchitecture. We have ARM and RISC-V, even MIPS is still around in some form (LoongArch), as is POWER.
What ties them all together is a common software development and deployment stack, built on Linux.
@@lawrencedoliveiro9104 In terms of chips shipped in volume we only really have 2 architectures left x86_64 and Arm, Risc V seams to be the one waiting in the wings to start doing real volume but its not there yet. Power is a lovely architecture but its not a volume scale architecture, commercially still viable for IBM but they have very few high value customers. MIPS sadly is just in a very drawn out decline into being embedded IP into larger chips, in the same way 6502 and Z80 are still being manufactured just embedded into other chips (e.g. Flash controllers). Technically UltraSparc is still alive for one more generation from Fujitsu but its basically dead now.
@@RetroBytesUK Doesn’t really matter. Linux runs on them all. I know the guy who has been single-handedly maintaining recent Debian builds for Alpha on a bunch of decommissioned departmental machines (some of which I used to admin).
@@RetroBytesUK
POWER is particularly sad, because it's always had potential for mass market relevancy, but IBMs enterprise customer tunnelvision led to a product that wasn't even suitable for their enterprise customers.
It's crazy that IBM still to this day operates with the same faulty mindset and model that lost them the desktop market in the 80s... Especially when you consider that Apple, who shares part of that mindset and model with them, still managed to evolve enough to stay in the game.
I worked for DEC for 34 years and saw the whole parade. I can vouch for what a great place to work it was. Here’s a story that for me symbolizes where it went: after Ken Olsen came Bob Palmer, a gunslinger hired to decommission the place. One day he came to our office in Greenbelt, MD to speak. There was a parking garage in the basement, the conference room on the first floor. They defined a walkway for him with theater ropes and a red carpet for him to walk on all the way from the garage to the conference room. Heaven knows you wouldn’t want to actually encounter some riff raff along the way. I’m sure Ken Olsen would have blown his stack at such a display of imperiousness. BTW, I heard an engineer talk at the end of Alpha who said there wasn’t much headroom for growth in the architecture. Another tidbit is that DEC’s chip manufacturing folks had feature size and clock speed that was WAY ahead of competitors at the time. The integration of those people and compiler folks also made Alpha what it was. All THAT was sold to Intel. RIP Ken and DEC. I’m forever indebted to you.
Good to hear from you Glenn.
@@robertthomas5906 wow! You too! Hope all is well and your wings don’t have any ice on them…;-)
My standard joke was that I could ruin DEC for half the money that Bob Palmer was making :-). DEC/Compaq/HP 1991- 2005
I worked at DEC and I loved working there. The only reason I left was Bob Palmer, I felt he was an idiot.
Didn't they all refer to the company as the name digital not DEC
I worked for DEC between 1985 and 1995. The only company I ever worked for that DEMANDED you to go on trainings for at least 8 weeks a year, all expenses paid. The amount of groundbreaking technology that came out of this company is mind boggling, and I still have to see a modern cluster system that runs as effortlessly and scalably as VAXclusters back at the time. Also: the search engine (Altavista), the firewall (Marcus J. Ranum and the DECwrl team), etc. Compaq bought a wounded swan and butchered it.
Thats just the fantastic thing I now expect to hear from people who worked for DEC. I sure beets the "we have a ping pong table" you hear from modern startups
I did my VAX/VMS Systems Manager course at DEC's offices in Manchester in the mid-80s - the two VAX machines in the cluster were named Hazel and Nut - the hazel nut cluster - it still makes me chuckle😂
@@andrewcollie as DEC field service engineer I saw a lot of "mars" and "venus" , the node names used in the system manager's manuals :-)
Former DEC employee (electronics tech -> software engineer). I spent a lot of time on PDP-11s as a tech. PDP-11 was a 16-bit computer, not 8-bit. It was capable of operating on 8-bit data, however, the instructions were 16 bits as were all the registers.
Indeed!
The mistake, which I also spotted instantly, is noted in the video description.
Fellow DECy, I worked for SEG at Hudson actually started at Westboro, chip testing.
Chuck I think I know you not sure which site but I spent 20 years at DEC, Augusta, Maine, Billerica and Littleton Commons in Massachusetts. I was a Principal Engineer at Dec. Sure do miss working at DEC except after Ken got booted and Palmer took over and sold it off piece by piece. Nick Gildred
I have lost many friends from the earlier computer days and I wish I could sit down with them and have conversations FOR DAYS. Please do the world a favor and find somebody near you that is enthusiastic about your earlier days. Let them chat with you, record it, and make it available to the world. I cannot stress this enough. The knowledge, experiences, and memories in your head are more valuable than you could ever imagine!
There was also Altavista, the go-to search engine in the early internet years. Created by DEC and running mostly on Alpha machines, as far as I know.
Altavista and Webcrawler were my jam back in the day. Kids these days don't even know...
Before altavista , I remember using Archie and Veronica. Those were the big ones. There was another in that set but I’ve forgotten it
Altavista let you craft your search so that it would return exactly what you were searching, no trying to guess what fits best. It was a valuable tool that the internet now completely lacks.
Oh, really?! I remember Altavista, was THE search engine I used in the beginning of my internet use, had no idea.
Hotmail was on top of Digital hardware and software still for several years after Microsoft bought it.
I had a short stint as a VMS admin at the time of the Alpha transition. VMS is, to me, one of the, if not THE greatest operating system ever. Seeing DEC fall was gut wrenching. Watching Compaq/HP absolutely throw VMS into the garbage the way they did was also infuriating. Personally I think the world in general and computing in particular is worse off with the demise of DEC.
I wish I had been around for VMS hayday or that VMS had held on sufficiently long for me to use it as part of my day job. Its still being actively developed, but I dont have cause to use it for work stuff, and I doubt many new places are designing solutions based on it.
can't agree more, every word you said.
Which login did you use, SYSTEM MANAGER or FIELD SERVICE? LOL
@@RetroBytesUK There was a book published called "The hitchhiker's guide to VMS" by Bruce Ellis and was in a similar story form as the Douglas Adams, but it taught you the inner workings of VMS. Got a copy right on my shelf, from 1991 back when I last played with VMS.
@@pauldunecat I may have to try and find my self a copy.
Alpha was the first CPU architecture that got Linux after x86. Linus Thorvalds wrote his Master's thesis about the port.
Also the first 64-bit architecture that Linux ran on. So it went 64-bit at the same time as it went portable.
Technically the first port (not by Linus) of Linux to non x86 was to the Motorola 68K. The Alpha was the first port done in tree and could be considered the first _proper_ port.
I guess it made a lot of sense because the Alpha had so much compatible periphery at the time.
Wasn't also true that the SMP featured of Linux developed as Alpha as well ?
@@autohmae SMP on Alpha came a little after Intel and hyperSPARC.
@@andrewclayton5859 ahh thanks, then I guess it was the 64-bit I most have remembered as someone already mentioned.
Another Alpha legacy: DEC licensed the EV6 bus architecture to AMD for their Athlon platform, which saved them from having to spend piles of cash they didn't have developing their own, or licensing the GTL+ bus from their biggest rival, Intel. The Athlon CPU was arguably the turning point for AMD and legitimized them as a competitor to Intel, and not just a knock-off company. You could easily argue that AMD wouldn't be what it is today if it weren't for DEC/Alpha.
And I remember back in the late 90s I snagged a DEC AlphaStation slimline desktop from an auction site, and managed to get NT 4.0 running on it. It would probably be more interesting today since getting it connected to broadband would be trivial. As it was back then, home networking was pretty much non existent and I never got motivated to get an external modem for it. But it was a great little machine and I was thrilled to finally make use of something on the NT install disk that didn't come from the i386 folder. I never did get around to finding a similar MIPS workstation. :D
"DEC licensed the EV6 bus architecture to AMD for their Athlon platform". Not only that, but they actually got some bits of the Alpha silicon architecture into the Athlon CPU. Actually is part of the interesting history behind the development of the original Athlon CPU.
@@hyoenmadan Yes, the fundamental architecture design of K7 was done by former DEC Alpha engineers, by repurposing a beefed up EV6 pipeline. The Alpha relation was in the fact, that the first Athlons already had SMP multi-core logic inside and the 64-bit extensions for x86 were already in R&D by the time of the platform release in 1999. The 64-bit K8 (Opteron) was a trivial "upgrade" of the original architecture, together with hardware virtualization support. Back then, AMD really made a life-saving bet on the farm, by snatching key talent from DEC. They milked the original Athlon design for a decade, may be too much, but it paid off well.
Also a lot of Alpha engineers went from DEC, to Compaq, to HP (to work on Itantium), some to AMD, but eventually most of them ended in Intel (due to HP-Intel cooperation), and were helping design Intel Core after Pentium 4 fiasco.
@@movax20h Think you'll find Core was a fundamentally different design after Prescott P4 by a small Israeli company, as is all the Core family today. That company saved Intel's butt.
A helluva lot of the improvements in new AMD/Intel CPUs were essentially following DECs roadmap for Alpha from about a decade prior for a few years there. Even today, if you watch what the markets the Alpha mainly competed in are doing when the budget calls for specialised hardware that usually is where our consumer CPUs wind up going later down the track.
I spent several years developing on VMS and even working as a sysadmin for an 11/750and 11/780. When the OS disk suffered a head crash I simply dropped a backup copy into the removable drive, swapped the ‘disk 0’ and ‘disk 1’ plastic activity lamp covers and booted. So many aspects were so thoroughly thought through. And the quality of the documentation… second to none.
I'm sad I missed out on that period, and getting to use VMS commercially.
Yes - Documentation was one of DEC's strong points. I worked on PDP & VAX systems, and had a girlfriend who worked at DEC support, so every now and then I'd have her sneak me into their library so I could rummage though the tech manuals they used internally, and then order my own copy from the reference#. I ended up with a mini DEC-library of my own, which all sadly got lost when the company I worked for went out of business. Still remember the deafening roar of the VAX system when walking into the server room equipped with Halon fire systems.
You're right I forget how good the documentation was for those systems, full shelves of documemtatom..
Yes lol, we had over 10 linear feet of DEC manuals for the 11/50 and all its peripherals and software.
They Grey Wall (used to be orange!)
That was my library when I was the VMS specialist at my employer in the 90s.
Long time DEC coder. RSTS/E on the pdp-11, and VMS. A lot of assembler programming. Back in the day I could churn out macro 32 code as fast as some did basic. I really miss how phenominally good all DEC's documentation was. No one else comes close. Those machines were a joy to program.
I also started on a PDP-11 running RSTS/E in 1977, and when I graduated from college, went to work for DEC, eventually ending up working in a group designing terminals. The PDP-11 was a 16 bit architecture, with a 64KiB virtual address space. Depending on the model, the physical address space was either 20, 22, or 24 bits (perhaps wider on the /70 and beyond; I don't remember for sure). Moving to VAX was a bit of an adjustment because the virtual address space was larger than the physical, and I had gotten really good at overlays and playing with the APRs to allow programs to occupy more than 32KiW. I learned engineering (rather than CS) at DEC during my time there, and my documentation style is heavily influenced by what I learned there. I knew several of the people mentioned in the video and one of them was a mentor to me for about 2 years.
The PDP-11 remains the best instruction set I have ever programmed in, with the exception of a few of the CIS instructions and MARK, which were between them abominations.
One of the core devs of RSTS/E was Mark Bramhall (MHB on the microfiche). He died recently :-(
DEC was serious about spreading their documentation too. As a teenager, I somehow expressed an interest in Alpha to them and they sent me the instruction set document!
@@filker0 I liked the IBM 360 and PDP 11 instruction sets..
I talked with an IBM publication writer who had worked for DEC before. She claim that where was a lot of IBM manuals pirated by DEC.
18:21 Fun fact: the VAXstation, and DEC itself, played a key part in the development of the X11 windowing standard. The precursor X10 system was actually tied specifically to VAXstation display hardware; the X11 rework made it portable.
Ooh!
@@kaitlyn__L The verison of VMS I have installed (when the HD works) default to the CDE desktop env.
More fun than fact. X was designed to be multi-platform from the very beginning, and X10 was shipped with workstations from DEC, HP and IBM, and also ran on Sun, Apollo and PCs. DEC did play a big role in funding project Athena (from which X originated) and also by dedicating a whole project team to X11 development.
Here’s a quote from Wikipedia: “Although X10 offered interesting and powerful functionality, it had become obvious that the X protocol could use a more hardware-neutral redesign before it became too widely deployed, but MIT alone would not have the resources available for such a complete redesign.”
@@lawrencedoliveiro9104 Yes, X11 was a major rewrite, but it's not true that X10 was only available on VAXstations. I developed X10 applications on an IBM RT (and had a very hard time later porting them to X11R2 on Apollos and DECstations).
I had a 21164 at 500MHz back in 1997/1998 and I adored it (I still have it in my storage). I ran Tru64 Unix on it and it owned everything. We had two GS140s with Tru64 and TruCluster for our back end. We made the first online actual animal production and health database. Which allowed us to predict locations of food and mouth disease outbreaks. We used them for fluid Sims to simulate the spread of Corona viruses in pig stables. Because we already know for 30 years that Coronas primarily spread by air as our systems also proved. And we could fend of disease by tweaking the air intake and are expulsion speeds. All that was floating point cpu breaking stuff that the GS140s did.
I was working at Air Force Research Lab Ins Rome, NY as a Jr computer engineer during college in 1991-1995. Our primary machine in my group at that time was a Sun machine, which to me was pretty hot stuff!
One day there was a DEC sales team that came to the lab to show off Alpha. I had heard a little about it through magazines but has never seen one until then. During the demo they ran the isohedron graphics demo (everytbing was software rendering back then) and it utterly blew away the Sun machine but like 10x. I was floored! Wanted one of those SOOOOO badly at that point.
It’s really too bad compaq killed it off. It really could have been a strong contender and might still be around today…
I was at Xerox at the same time, running Sparc too…saw the same demo from DEC at their European launch, with the same reaction 😎
This comment brought back a memory. We had an Oracle demo scheduled to determine if we were going to run our database servers on Alpha or Sun. Called up DEC first, described the workload and they were like "OK, cool...we'll bring out a 2100, install Oracle on it, and then we'll run your benchmarks." Then we called up Sun, described the workload and they were like "OK, we'll bring an E250 configured with for your benchmarking."
Whoever set all this up with Sun mentioned that we were looking at Alphas, too. Immediately the Sun rep asked what DEC was bringing out and Sun decided to essentially double the size of the system they were bringing for the demo. I _think_ they ended up bringing a maxed out E450.
I have no idea what the relative cost of these systems were but the Alpha absolutely destroyed the Sun in our testing.
Had one in the DICE lab
Another DEC legacy, is Windows-NT itself. Dave Cutler - having been a lead on VMS and previous lead on a couple of the PDP O/S's - not surprisingly was heavily influenced by his DEC background in designing Windows-NT (in fact the old joke is that WNT is VMS+1) and along with the ex-DEC team he brought to Microsoft, implemented a lot of core VMS concepts, and some code such as the task scheduler is remarkably similar - so much so that DEC took Microsoft to court over it, and MS settled at the last minute. I worked on PDP systems using multiple O/S options, and VAX systems using both VMS and Ultrix-32, and was always impressed with the quality of the equipment and attention to detail put into the designs. The other *BIG* thing about DEC was Documentation - it existed for almost everything and was extensive. These were machines and systems designed by engineers and then marketed by sales teams, rather than the other way around.
My last job I worked at we still had a cluster of Vaxstation 3100's (think they were model 76, but can't quite remember) running VMS still running in a production environment, this was at the end of 2020. Those machines were well built.
Yes, Windows NT "inherited" a lot from VMS - some of those concepts remain in Windows to this day. Yes, the case was settled out of court and thereafter, for a couple of years anyway, Microsoft was VERY nice to DEC giving it first dibs on new things instead of Intel - even though the relative market shares didn't really warrant it. I wonder if there were some dots to join up there? 🙂
I was working at DEC when Cutler responded on VaxNotes to the WNT = VMS+1 "What took you so long"
VMS was a true simultaneous multi-user OS. NT is a single user system, though it can have several accounts logged in at one time. So in that regard, NT was better described as VMS-1.
Somewhere around 1990, we got some new microvaxes with huge graphic monitors not too long after we had acquired our first PCs. But I was *horrified* at how long it took the microvaxes to paint the screen compared to the cheap PC's very fast update. More power, less graphics speed.
Ex DEC employee and VMS system manager here. Well done.
I was there when they ousted Ken. Super bummer time to work there starting at the heyday and then the downward slide. What a great company.
I wish you included the part where DEC gave Intel the secret architecture Alpha design plans to create a second Alpha source where Intel declined to become an Alpha FAB source but stole the design to make their chips run faster.
Yep, Intel did that. DEC sued Intel and then ... I recall DEC agree to license some processor designs to Intel ( I think that was a mistake ).
I worked for DEC from 1976 to 1993. You can put me in the same group that basically worshiped Ken Olsen. Not in a religious sense but in a human sense. I knew no one that had a bad word to say about him. He was so down to earth. Used to visit plants and just stop and ask workers what they were working on etc. He was interested in what you were doing. Wore patches over the elbows on his suits. Drove a Ford Maverick for quite some time even though he was worth well over 200 million at the time.
What many do not realize is that DEC had the best field service group in the world, some 40k strong. When Compaq bought DEC, they only did so to acquire the field service unit. Most of the hardware divisions had been sold. At one time, I believe DEC had 25 manufacturing facilities in the U.S. They had a fleet of helicopters that used to ferry employees between plants throughout New England. So, Compaq only really wanted the field service people and nothing more and that is a big reason why Alpha died.
If I remember correctly, the chip facility was in Hudson, Mass. I had heard a rumor that it was built on huge springs that kept the building from excessive vibrations that is vibrations from the rotation of the Earth. It took a lot to cut those wafers back in the day. Then you have the DCU. Had a branch in our facility. It was great. DEC was such a great place to work and they also really held to the open door policy where no office was allowed to have doors. Only conference rooms were allowed to have doors. Even the plant managers office did not have doors. Ken wanted an open atmosphere. I will concur, they day the board threw Ken under the bus was the day as they say, the music died. I really miss that place.
I worked in the Hudson fabs in the early 80s. From there, I got into operations in Marlboro with the AITG (artificial intelligence technology group). Stayed with them for six years. This facility had 7 Vax clusters in the main computer room.
Pretty sure it was a Ford Pinto
DEC service was insanely expensive as I recall. I had a core memory board with 3 cracked cores. A third party repairer was less than a quarter of the DEC engineer quote. It was great fun watching the engineer re-threading the wires through the cores.
@@alanhaywood01 Yes, Ken Olsen did drive a light blue Ford Pinto. I remember seeing his Pinto parked at the Mill in Maynard, MA. Yes, Ken did indeed walk around alone (no entourage in tow) having one on one conversations with whomever. That behavior is sometimes labeled 'management by wandering around'; MBWA. It works, because people respond positively when they know management CARES. Also, it is a great way to identify internal company problems, if mid level or first line managers 'filter' ( mischaracterize ) situations. I was there from 1977 to 1991 serving in customer service, network sales, and engineering. Great company.
I think the decline started about 1985, when DEC hired thousands of sales people ...that did not fit the culture. Lots to be said about that, but not today.
I worked at DEC from the early 80's until the mid 2000s. Our feelings about Ken getting ousted were exactly as described in the video. Things started getting really bad when Bob Palmer (AKA GQ Bob) was hired to dismantle the company and make it attractive to be acquired.
Palmer was just the DEC version of Chainsaw Al.
Remember his mink coat?
The Alpha project had an interesting and ambitious goal: 1000x performance improvement over its lifetime, broken down as 10x from process improvement (i.e. clockspeed), 10x from IPC improvements and 10x from multiprocessor environment. That doesn't sound as impossible when it's broken down like that, and it is scarily close to how the industry has moved, though perhaps not 100% as they envisaged.
Excellent video! I worked at DEC for 16 years and was one of the folks who left when Ken Olson was let go, because as you've said, it wasn't DEC anymore. I look back and miss the good times I had working with PDP11s and VAX. My last position was with the development team that ported DEC Unix to the 64 bit Alpha. I still have a VT100 and LA120 in the basement. Those were the days....
This brings back some very very happy memories of that time. I moved jobs and went from looking after HP-UX to a bunch of AlphaServers/AlphaStations. A couple of 1000s and 4000s and the horizontal alpha stations. Mostly of them ran Digital UNIX/Tru64 but we had another set with VMS as well for other things. Going from the HP-UX servers to the Alphas was one of the best times of my career, these things were such a joy to manage. The OS was fresh and modern, the hardware was so well built and easy to maintain. Plus,. for the size of machine they really really performed.
The highlight though was the way Digital dealt with their customers. It really did feel like a different company even when it was in the early days of being owned by Compaq. Hardware problem? If you could tell them what it was (and they were good to diagnose) they'd happily ship out the part for you to fit yourself, even if it was server mainboard. If you weren't comfortable or sure of something? Sure we'll send an engineer to do it with you. I remember hitting an obscure OS/TruCluster problem at one point and their customer support went from 0-60 getting it through 1st and 2nd line support to 'you are now talking to someone from the kernel development group'. They were really a passionate bunch about Alpha and VMS/Digital UNIX/Tru64 and you'd find those engineers helping people out on the various newsgroups.
I had one of the Digital UNIX key chains, 'CALIFORNIA Y W8 4 HP' (you can see one of them on the Tru64 UNIX web page). It was really sad to see it's decline and eventual death of their mentality within Compaq/HP.
Oh and the DEC Alpha story isn't complete without mentioning AltaVista, which for me at the time was the only web search worth using. That was an incredible example at the time of what the could be done on the Alpha processors.
altaVista is going to get its own video at some point.
Yep there was a number of company's back in the 80's-90's that truly had passion. Today pretty much the whole tech industry "passion" is just something that is said as empty words. No one strives for perfection... and instead is striving for not sucking.
As a young boy, I read the computer magazines and lusted after hardware I could never conceivably afford. The Alpha machines were one I drooled over but never actually laid hands on. Now they're almost as expensive as new or more expensive in some cases.... Never to be I suppose.
Back around 2002 I came across an Alphastation on ebay for 273CAD, or at least that was my bid, and I wasn't watching closely enough and it got sniped. I regret immensely not putting in a higher max bid, I have never seen a working complete one that cheap since. I just did the inflation calculation and that would be 396 dollars today which would still be a great price.
I bought a DEC Alpha Multia after it prime days were over and severed me well running Linux for years.
You have basically documented by computing work history. Each one of the points mentioned was a milestone in my working life. Thank you for the memories.
Having fond memories of working on VAX/VMS, 780s, MicroVAX IIs, disk and tape stations, terminals, DCL, and VAXC. Those were the days. We got so much done for so many people on those. Amazing bang for the bucks. VAX and VMS were professional tools at a time PCs were still barely toys. Thankful for having had that experience. Cheers, Ken!
Mr Bytes you are quickly becoming one of my favourite channels. Thank you for these videos they are endlessly fascinating.
Thank you, thats very nice of you to say.
I worked for Digital and Compaq. DEC had incredible products but it got a bit too complacent and didnt read where the market was going. They made the best products but it could have turned out differently with some better marketing and strategic direction (hindsight is a wonderful thing). There were amazing people working there and I owe a lot to them.
When I was at university in the late 90s and early 2000s, my university hosted our email on a VMS cluster. Very interesting OS. With my limited exposure, the biggest thing I remember is the built in versioning in the file system.
The versioning was an essential part of how the OS worked. During an upgrade new versions of OS files could be written while the prior versions were still in use. After a reboot a purge would clean up the previous versions. It's much better than the Windows approach where new files are written to temporary names and renamed after a reboot.
All our universities did. There was imap on solaris too.
Yeah versioning was "different". Once a guy came in with a 5-line program he couldnt get working. The file name was MEDICAL.PAS;173 He had gone through 173 versions and revisions of the file, a 5-line program that was still not working at all. A poor and totally clueless guy, at least with computers.
I spent more than a decade working on the VAX with VMS developing applications for a mail order company. loved it and, looking back, I wished my time with the VAX could have been longer. This was one of the golden eras of computing. I really like the fact that Dec machines are still available and that there is a thriving community online that supports these machines and operating systems. In fact, I have a whole collection of VAX equipment in my workshop thanks to eBay and an early donation from a former DEC service tech.
I do want to get my self a microvax box at some point.
You should join us online by connecting some of them up to HECnet. DECnet and the machines that run it are still alive and running :-)
Thank you! Great summary; thanks for sharing! I had a unique view of DEC; I worked at DEC from 1977 to 1995. My mom and sister also worked there too. I grew up near the Maynard Mill, later on, lived on Thompson Street, 2 doors from the main entrance. Was part of the DECtalk and PC/Networking Group. I went from order entry file clerk to senior engineering manager and worked at many DEC sites. Until I left DEC, I did not realize how progressive and supportive they were of women in engineering.
The first computer I got to work on was a PDP8 at my high school, it was replaced by a PDP11. The Ritchie C book was one of our text books.This was in a day when there was only one computer in the school and barely anyone went near it. The fact that I got to work on it was because my father had a Telex machine and it had the same terminal and paper tape machine which one day broke down at school. I told the teacher I knew how to get it running and proceeded to clean it out and get it working like I did many times before. A week later the optical card reader stopped working and he asked if I knew how to fix it. I lied and said I did because I was sure it was only dirt keeping it from working and I was right. From then on when DEC came to work on the machine my teacher would call me down from whatever class I was in. DEC was only a mile from the school and they came often to try out things including giving us a tapes with the game StarTrek and UNIX on it that we would run on Fridays. Normally it ran a DEC business operating system. In college I picked a school that had and used VAX machines in the labs, at the time many computer/science courses did not have direct access to the computer and were still using punch cards, I did not want that. After school I joined a scientific company that was a DEC Var, I would travel out to customers and get the machines running, first thing I would do was get the ISDN up and log into DECNet. Sometimes I'd stay all night and other DEC people from around the world would help me solve any problems, the Aussie DEC reps were awesome and often helped late at night. This was still the 80s but by the 90s we were installing multiprocessor VMS machines with SGI front ends. Soon other Unix machines would join in and we had IBM and Apple AIX file servers and Sun. We used the first Macs as low cost/tiny footprint terminals, an early Mac was still less expensive than a DEC terminal and sometimes you could put one where nothing else fit. It wasn't until the Intel Pentium Pro dual socket board came out that we ever had use for a PC. Even the DEC Rainbows ran circles around any PC of its time. At one place I worked I was asked to add some new fonts to VMS and I looked up the running time and the machine hadn't been rebooted in a few years so I dug around and found the whereabouts of the guy who last booted the machine, he helped to review my changes and was quite helpful after hearing I was still running on his boot. Let me repeat that VMS had YEARS of UPTIME without rebooting! My first experience with Alpha NT and a mail server was watching a guy not able to apply a change because the monitor was too small to show the bottom of the window, when he resized the screen it wanted to reboot.
I nearly got to play with a DEC rainbow this afternoon, unfortunately it had developed a fault this morning before I got there.
Alpha was an incredible platform. It reminds me of SGI in that it was way ahead of PC but finally failed commercially and was superseded by generally inferior PC.
Exactly
Everything now has turned into a cluster of high end PCs glued together by middleware.
I don't remember how much Alphas cost back when I first saw them but I remember thinking they were forever out of my price range. I couldn't blame the company where I worked for opting to stick with cheap PCs.
The PDP-11 was a 16-bit system. VAX's also ran a Unix derivative called Ultrix-32
Yeah, he kinda messed that up when he glossed over the PDP-11. The 11/20 had an 18 bit console, but you pretty much had to have an 11/45, 11/70 or a Q-Bus LSI (non Unibus) PDP-11 to address more than 64k. The memory management in the 11/45 would let you install up to 256kb of memory (18-bit addressing). The 11/70 would run quite well with 4-meg of memory (22-bit addressing). The VAX was 32-bit but you were addressing virtual memory on the disk as nobody would be able to afford 4-gig of memory (boy have times changed).
When he talked about PC's, he also missed the PDT-11, VT-180 (robin). DECmate, I and II, LINC, VAXmate
@@MrTherende Also missed the DEC Pro line (the 325, 350 and J-11 380). Sadly, except for a private copy of RSTS that Paul Koning had running on a Pro, all Pro's ran RSX. I had a lot of hope for the Pro line, but it never panned out. Loved the PDT-11s, but the VT-180 was more stable.
This was a great watch, thanks for putting it together. One interesting side quest that you didn't cover was Alpha being massive in the HPC market because it blew pretty much anything else at the time right out of the water for raw Floating Point performance. I was a Physics postdoc in Forschungszentrum Juelich in Germany during the turn of the century. Many of us had DEC and Compaq (more of these) alpha workstations running TRU64 (as I recall we even had an early version of KDE running on them). More interestingly we had two Cray T3E supercomputers that were in the top 5 of the top500 list at the time. These had 512 Alphas inside them and we wrote massively parallel code to run on them. I think these actually ran UNICOS but my memory is getting hazy and we also had T90s and J90s so I might be mixing them up.
Loved watching this, taking me back down memory lane. I’m a Longtime DEC user/owner from PDP 11/03 through VAX through VMS, Tru Unix, through to Alpha running NT 3.5 and NT 4.0. I’ve still got them all. Most importantly I still have my DEC coffee cup. If I’m going to be meeting with some of the old timers I make sure I pull out my DEC cup at the restaurant and get a fill of coffee. Although you covered the major steps along the way to the Alpha you stopped before getting to the Beowulf Cluster fad with a stack of Alphas.
Thanks again.
Started my career as Alpha was coming in. FX/32! was really cool - it made AutoCAD run quite quick. I had an Alpha Station 225 that came with NT 4.0. I ditched that an installed Red Hat to run our main organisational DNS server. I even had a sticker on it: "Red Hat Screams on Alpha".
That’s a waste of the alphas capacity. You could’ve used any other system for that. Alpha was an FP monster and this is all SpecInt and some IO.
And RH didn’t perform all that great, Tru64 was considerably faster. And it was at one point a 150 bucks.
But you at least didn’t “rape” it by running Windows NT, like so many others did 😆
Loved the video, as always. Did some DEC VAX back in the late 80's - upon a project to port ICL2903 PLAN assembler to DEC VAX C. Some salesperson thought it a migration could automate and well, fun times.
The C development environment was neat and well-featured and the performance tools were pretty darn cutting edge for a long time.
I had no idea there was a port of PLAN for Alpha, and I was working for part of ICL round the time Alpha was really starting to take off.
@@RetroBytesUK Nope, no there wasn't a port of PLAN for Alpha, though that would have been easier. No the project I worked on was porting an application written in PLAN assembler into DEC VAX C - yes the things salespeople would sign.
Sorry if that came across any other way initially.
I did work at ICL for a period as well, later on - Feltham in the late 90's.
So many memories brought back. I spent years working on DEC Alpha systems running OSF/1 then Tru64 (though almost never Windows NT) they were super systems. It was a sad day when Digital disappeared.
Wow, you brought back some memories!
I lived within the DEC ecosystem for quite a bit of my tech career. I started off by working for Visual Technology who made emulating terminals that competed directly with DEC's products. Visual's products emulated the DEC VT100 directly but was also able to emulate other brands such as Lear-Siegler. Visual also had graphics products that emulated the DEC graphics terminals as well as the Tektronix models as well. These were quite popular and were quite fun to repair.
A few years later, I worked for Infinet who I mentioned in a comment on your comms video. At Infinet, I ultimately ended up in MIS where I worked as a computer operator. Infinet had two VAX 11/780 clusters as well as an older 11/750 running ULTRIX. As time went on, they introduced a VAX 8350 to replace one of the VAX 11/780s. During this time, I got to know VMS pretty well and ended up learning Btrieve32 to build queries and reports during my non-operator time. For MIS tasks and operations, I used a DEC VT100 terminal, or a paper teletype console. They also had DEC Rainbows and I used one a few times for some programs including one to print out backup tape reel labels.
Many years later, I used a VAX 4000 and a 6000. These were the last of the DEC VAXs I used and my use of VMS. This product while still DEC didn't feel the same. They were newer but felt tired and weren't quite as reliable as the old VAX 11/780s I ran at Infinet. Honestly and truthfully, the 11/780s were the height of DEC's quality.
After that it was DEC's workstations at Polaroid. Polaroid Graphics Imaging (PGI) used both the DEC Alpha workstations and sold the DEC PC workstation products as RIP servers for their digital proofing products. I supported both products as part of the IT support group for that division. They both ran Windows NT at the time, and I remember the Alpha being a whole lot faster than the Intel PIII processor versions. There was a big fault, which I'm not sure you encountered with the workstations. When there were two hard drives placed in the drive-cage, the drives were so close that the magnetic field was felt when a drive was removed. The close spacing also caused drive failures due to the heat buildup in the case because there was little space for airflow in there with the drives so close.
My relationship with DEC also came about because I live in the state where they were located and family and many friends worked there, Wang, Polaroid or Western Electric (Lucent/AT&T). The plant where the DEC Alpha chip was manufactured is still in use today. Located in Hudson Massachusetts, Intel now owns the facility and manufacturers various processors and controllers there. I knew a few people that work there as well. From what I remember, Compaq sold the Alpha and the facility to Intel back when they ditched the processor.
I'm a relatively new sub, and gotta say, your stuff is great! Usually I happen to jump on board channels the moment before they blow up in popularity, so I'm hoping that tend continues with you
Very well done. Never worked for DEC, but certainly competed against them while at Data General 1982-1992 and Sun 1992-2001.I recall trying to meet Ken Olsen while manning the DG booth at Autofact in 1985. I went over to the DEC booth with my DG badge on, saw Ken, walked up to him and stuck my hand out and introduced myself. He started to reach for my hand, but when he saw my DG badge, he retracted this hand and turned around. A lot of bad blood between him and Ed DeCastro who worked for him at DEC and went on to found DG.
I cut my teeth in the working world on a VAX 11/780 at Caterpillar. GREAT machines with very advanced Operating systems...VAX mail which allowed you to split the screen on a VT100 and communicated with another user on the system , for example.
Larry, wasn't that VAX/VMS-Phone actually? I used to chit-chat with this girl from one of the Paris offices using Phone, from Melbourne in the late '80s...
ciao, lm
I ran a VAX 11/785 back in the day and also got the 3rd Alpha off the line. I used to go to DECUS, the yearly meet up for DEC admins. I'd leave DECUS with several tapes or tools and games for the VAX and Alpha! I had two hard drives on the VAX that were 30 Megs each. They had five platters each and were the size of a clothes dryer. In order to access the data on a different platter you had to dismount the drive, spin down the drive, pressurize the drive, open the top then unlock the platter to pull it out and place it in this case that looked like a cake carrier. Then lock a new platter in, close the door, evacuate the platter chamber, spin up the drive then mount it. The whole process took about 10 minutes!
I had a number of Alphas years after they were new. One was a dual 21264 CPU that I unfortunately never got working. Insane CPU modules the size of small cigar boxes. I remember hearing about the Alpha while in high school, with its CPU running at hundreds of megahertz while it was still pretty normal to have a 386 running Windows 3.1 so Alpha always had a mystical legendary feel to it. Even around the year 2000 Alpha was destroying seti-at-home an order of magnitude faster than top end PCs.
The EV68 was a monster, we had many many servers and workstations including a GS1280 , then I left that company never knowing that was the last time I touched Tru64 or VMS in a production mission critical environment. Makes me sad what happened to that architecture.
Yes, I was always seeing all those Alpha’s at the top of the seti@home rankings, it was magical
This was a really awesome video! Thanks so much for putting it together. I learned so much about DEC (especially that I probably shouldn't have tossed that old Alpha in the trash years ago). I also had NO idea about DCU and what it stood for!
I worked for a NASA contractor in the early '90s at JSC on a project to replace the MSC IBM mainframes with networked UNIX systems that would present Shuttle data in color, 3D, with 'counts' from the hardware replaced by engineering/scientific units, and AI augmentation.
As a SBIR project we were severely underfunded, and would scavenge old Dec and VAX stations, MASSCOMPS, and other systems from the monthly surplus dumps, as well as the occasional demo Sun and other minicomputer systems from the vendors trying to sell into NASA. (Thanks to this oddball collection our C code was truly portable between systems ;)
I watched the tech magazines as the Alpha came to fruition, and got the DEC rep to 'give' us an early Alpha, 100 MHz, 32 Meg RAM, compared to all our other machine lightning fast!
I drew up a new network design (literally on a napkin at lunch) based on the Alphas and showed it around my RTDS team and other NASA coworkers, it drew a lot of interest but we didn't have the money for it.
Finally we had a failure of the mainframes during a sim and our demo machines in the front room were the only devices tracking the simulated ascent of the shuttle. Very shortly there after the largest IBM mainframe buy for NASA/JSC was canceled and my napkin drawing with upgraded Alpha models and our distributed network was implemented to replace the monochrome IBM monitors and mainframes in the front room.
The descendant of that system is still in use, although I don't know what computers it uses now.
I had a DEC Personal Workstation running Window Maker on top of a Red Hat variant (iirc) as my daily driver from ~2000 - 2004. I really enjoyed it.
Don't know if it was intentional, or common. BUT everytime I see Window Maker it translates to Widow Maker in my brain. Miss the good ol days of watching the WidowMaker motorcycle on TV..... (old CRT tv on Saturdays/Sundays).
This is quickly becoming one of my favorite YT channels. Well done!
The main thing I remember about Alpha was its 30 Watt power consumption and how high that seemed for a CPU then
As an undergrad I had a soft spot for DEC’s brief dalliance with R3000 MIPs DECstation range running Ultrix. Later as a postdoc we got one of the first line of Alpha machines - I was gobsmacked, it totally blew away pretty much all of the SPARCstation machines I’d been using before.
just come across this one, best narration! thanx for the memory lane trip.. at DEC from 1972 till 1992, unforgettable times.
Awesome video
I don't think I've come across another video that covered the history so well. Thank you!
I worked with OpenVMS for close to 30 years, love the OS
It's being ported to x86 currently, first prod release due this year in fact (it's already out in field test)
Some of the original developers came out of retirement to work on it
They have already got it running on VMware, virtualbox and I think Xen already
Lots of things planned for it too
Google VMS vsi (vsi is the company who now owns the rights to keep developing OpenVMS)
I have a very small museum of things from my 50 year association with Digital Equipment. Although I never actually worked for the company, "Uncle Ken's" culture informed much of my approach to computing and life in general. These include some early promotional material from the Alpha based systems. Perhaps some of the strangest things include different rubber overlays for the numeric keypad on the VT100 and VT200 keyboard. These were custom overlays when the soft keys were used by different applications such as an editor, although we always used to joke that they were to prevent the user from getting pregnant.
The customs overlays! I had totally forgotten about those rubber things.
I used those overlays all the time. The WPS-layout flavor of the EDT editor was much nicer than the standard EDT layout, but hard to memorize the layout, so the rubber overlays were the solution. I'll take EDT + WPS overlay any day versus any Unix/Linux editor you can name.
I remember DEC Alpha being in one of my first IT jobs at BT Labs in 1997. Yes, Exchange Server on Windows NT!!
Later in my career (I worked for HP 2000-2006), saw quite a bit of Tru64. Excellent clustering capabilities, shared filesystem and context aware soft links. Perfect platform for running Oracle RAC.
Rumour: Each letter in WNT is one letter up from VMS
Thanks for this wonderful video! In 1995 I worked as a mechanical designer building 3D solid CAD models using Unigraphics on a 233MHz DEC Alpha running Unix CDE. It was so much faster at processing than any other workstation we had in our office, it was like driving a Ferrari amongst economy cars. In 1998 I ended up finding a still in the box new-old-stock 166MHz DEC Alpha on an online bidding site for $200 and used it as a hobbyist for several years.
My favorite story about how VAX was the CISC architecture par excellence is when I learned that it had an instruction (POLY) for evaluating a polynomial. I wondered how big of a polynomial they let you calculate, thinking "it's gotta be something reasonable, right? Just a quadratic, or maybe a cubic for a nice power of two. Four coefficients ought to be enough for anybody."
Because this is a computing story that contains the phrase "N (things) ought to be enough for anybody," you'll have spotted already that I was wrong: the arguments to the POLY instruction are the argument to the polynomial, the degree of the polynomial, and a pointer to up to *thirty-two* coefficients. 😳
Apparently it wasn't used much, and not every VAX supported it fully, but still, absolute legend.
And completely pointless. VAX was an object lesson in how doing things with simpler, more general-purpose instructions, while they might take more instruction bytes, were usually faster than complex purpose-built instructions.
@@lawrencedoliveiro9104 wow man, talk about living in your own world. The entire Digital Signal Processing industry, you know , every media processor on earth that is responsible for the modern world, might want to have a word with you. Tiny example, TI's TMS320 DSPs, 1st one the C10 released in 1982 had "MAC" instruction (polynomial eval) which is the core of everything media processing (think digital audio to video)... that line went for 20 yrs.. up to ST's current STM32 series that has the same thing. Those are consumer examples, with DSPs and their "polynomial" eval (massively CISC inst) that can execute in ONE cycle, everything from cell phone base stations to aeronautical guidance disappears... oh and the core of all the "AI" hype today is that "MAC" instruction too... so yeah..no .. you may want to broaden your perspective. Oh and this info is not from a google search..
Nice post man, yeah a lot of ppl "used" the POLY inst indirectly since it accelerated MATH libs like BLAS etc and fortran compilers.. why the VAX series absolutely ruled the engineering space.
@@gregd6022 Multiply-accumulate is a common instruction, not just in DSPs. It takes a fixed number of operands -- just 3 -- to produce its result. Not the same thing as the VAX POLYx instruction. Not nearly as complex, for a start!
I guess/hope DEC had compilers that knew to use that instruction? Unless you were going at it in assembler (I knew a guy who did; he was challenged to write code to change other processes' process names) I can't imagine how you'd invoke it unless the compiler was crazy smart.
I had a boss like Ken, and when he left it went from a environment to a routine ... I am still at that company and im not very happy
Hey friend. I really appreciate your videos and find them invaluable. I work in the tech industry and have worked with a few people who worked at DEC in the 80's-90's. It's insightful to see and learn all of this, and the way you present it is perfect. Thank you, and keep up the good work!
Brought back so many memories. One of my first jobs that I didn't ask for (I already had a job in design...but they thought I should do more) was to be the "computer operator" (basically IT guy) for the corporate VAX Cluster. We had a network of DG machines as well that they threw in, so I had to keep doing my design job and this besides. It wasn't initially my idea of fun. Nobody ever compliments the IT guy...:) But your video brought back memories of raised floor rooms with massive A/C and three phase 480V lines dropping like tree trunks from the ceiling, and trying to figure out the right parameter in the "lookaside lists" in the VMS setup and banging my head against that for quite a while.
I had one of the Dec Workstations in the 90s, I remember running SETI@home on it when it came out and it was great. Also playing pinball on it with nt4. That was a fun box.
Would have been fun had someone slipped some Bitcoin code into the SETI@home code. Wonder if their timelines overlapped any?
Not only was UNIX created on DEC machines, but the C programming language was created to make UNIX portable to the PDP-11.
My heart belonged to Unix when I joined DEC in 88, and I remained "the unix guy" in the team, always running Ultrix, and there was plenty of friendly banter about which was best Unix or VMS. Happy times and great colleagues who politely put up with my dinosaur quips. For me , Digitals greatest legacy was being the birth platform for Unix and hence OSX, Linux and Android
@@jives11 Hi Jonathan. We did have fun baiting the VMS guys didn't we!
@@alanmusicman3385 Yes indeed, but they took it pretty well. I was based in Basingstoke, UK. They were quite happy that all my VAXmail and ALL-IN-ONE mail was forwarded to my Ultrix box via the Ultrix-VMS connection sendmail gateway. Worked on DECwrite, DECpresent plus any 3rd party products we sold on those RISC machines like Uniplex and Alis. I think the latter was the Great Great grandfather of Libre-Office ?
@@jives11 and with VMS leading to MSDOS and thus Windows, the Unix/VMS war continues to this day 😋
@@SimonBuchanNz Windows NT was the OS with roots in VMS, but I take your point.
Great video. I worked as a system test and repair technician on PDP, Vax and alpha systems in the Digital plant in Ayr, Scotland. Great company and products. Watching stuff like this certainly brings very good memories . Many thanks
I always wanted to get my hands on one of these. Thanks for the history and breakdown -- I appreciate the hard work that went into this video!
I had a 533MHz 21164 version of this exact workstation back in the 90's, and you actually needed ALPHA compiled binaries for Windows NT programs to get any speed benefit, as when running x86 Windows NT applications on it required x86 jit translation, and contrary to what is said in this video (2x as fast or so), In my experience it was actually slower than using a contemporary Pentium based system/workstation/server instead.
And unfortunately there were only a handful of alpha compiled Windows NT applications available...
I had a special version of LightWave 3D for Windows NT/Alpha and used the workstation for rendering my 3D animations at speeds at least 2x to 3x as fast as a contemporary Pentium 180 MHz machine...
After it became old and Intel machines were faster with Pentium III etc CPUs, I relegated it to my garage and turned it into an OpenBSD web/mail server over DSL for a small web/mail hosting company I was running from home in my spare time, and the thing stood in my dirty, dusty garage with a small UPS and it had an uptime of more than 6 years before I migrated all the web/pop/imap accounts to a new x86/freebsd rack unit at a hosting provider... Very well built machine indeed! :)
IBM had the same problem with their RS/6000 series and AIX. There was a emulator that allowed you to run x86 code, but it was slow, real slow. I remember it as performing somewhere between a 286 and 386 depending on the program. So it was more a case of the system being able to limp through x86 code. Certainly not something you wanted to do if performance was in anyway important. I never even tried it myself as all programs I worked with either was supplied as source code to be compiled for the system or was available in native POWER and later PowerPC versions.
Nice treatment of DEC/Alpha. Very well presented.
I'd love to see a similar video on Data General.
Would love to hear about the storage systems. Was a Sys admin for 20 years, Oracle on DS10, DS20, ES40 45, GS 80, 160 and 320 systems and cabinets of DEC storage from vintage 90s through EVA systems. Great hardware, Tru64 and DEC field service.
Great video. One of my first ever machines I admin'd was a DEC Alphastation and I loved those things. They were unbelievably fast for the time. Tru64 was a bit funky but usable and I really enjoyed working with that whole platform.
Another great video thanks. I was an application developer on VMS for a few years. After earlier roles developing in the likes of COBOL & APL on IBM S/370 TSO/SPF & VM/CMS and TRANSACT/IMAGE on HP3000 MPE, VMS with Fortran was a real breath of fresh air. It felt like the OS developers really understood the app developers’ point of view and how to make a real leap in effectiveness. Little things like default file versioning and big things like the ease and power of VMS DCL scripting. I then worked for a spell on Intergraph VAXes programming 3D graphics apps. Intergraph basically added graphics cards to standard VAXes and repackaged the systems in sleek black cabinets with more coloured flashing lights than could ever be explained. I remember being sent away on fun training courses at DEC offices like “Programming VAX/VMS internals using Fortran IV & 77” where I met DEC staff and felt the wholly different vibe compared to staff when I worked for a short while at IBM. Things got messier as the DEC PC “competitors” like DECstation and VAXstation came and went, clustering and networking exploded, and I was pulled into IBM SAS (stats) and SGI Irix 3D stuff. But to this day my fondest programming memories are on those VAX systems.
The VT100 wasn't a dumb terminal, it was a "Smart" terminal. They communicated with the computer to determine cursor position and formatting etc
That's the definition of a dumb terminal. It displays the data that arrives (including control sequences to move the cursor and change the formatting) and only sends back the keystrokes you type, it doesn't run its own programs.
@@vink6163 No, dumb terminal just received/displayed characters like a teletype. in fact, "dumb terminals" were often called "glass teletypes" for this reason.
A terminal that runs it's own programs is called a "computer"
@@vink6163 From Wikipedia:
"Dumb terminals are those that can interpret a limited number of control codes (CR, LF, etc.) but do not have the ability to process special escape sequences that perform functions such as clearing a line, clearing the screen, or controlling cursor position. In this context dumb terminals are sometimes dubbed glass Teletypes, for they essentially have the same limited functionality as does a mechanical Teletype. This type of dumb terminal is still supported on modern Unix-like systems by setting the environment variable TERM to dumb. Smart or intelligent terminals are those that also have the ability to process escape sequences, in particular the VT52, VT100 or ANSI escape sequences."
@@the123king Interesting, I stand corrected, I had always assumed dumb terminals were the ones that didn't go beyond processing escape sequences, unlike say an X-Terminal which required a much more intricate bidirectional communication protocol.
A fun fact is that the OG AMD Athlon was designed by many of the DEC Alpha's designers iirc.
Dirk Meyer was the chief engineer of the Athlon. He gave AMD the solution to its biggest weakness compared to the Pentium - the monster floating point unit that was the Alpha's biggest strength. He eventually went on to be the CEO of AMD for a while after the success of the Athlon.
@@bdwilcox Jim Keller joined DEC in 1982, and worked there until 1998.
Jim Keller was the lead architect of the AMD K8 microarchitecture (including the original Athlon 64) and was involved in designing the Athlon (K7) and Apple A4/A5 processors. He was also the co-author of the specifications for the x86-64 instruction set and HyperTransport interconnect. From 2012 to 2015, he returned to AMD to work on the AMD K12 and Zen microarchitectures.
I was going to comment that the PBP-11 was 16 bit, but you already did so in your intro remarks. Great video, and great memories for a former DEC employee and PDP-11 programmer. Thank you.
I worked for DEC for 28 years and was involved on a porting team to get ISV's onto the Dec Tru64 Alpha platform. We also had Windows NT running on Alpha for a while. Those were the days.
0:53 The PDP-11 was considered “16-bit”, not “8-bit”. All the registers were 16 bits, remember.
Its a 16 bit machine indeed.
Slip of the tongue , I even wrote 16bit in the script. I've added an info card to point out I meant to say 16bit.
@@RetroBytesUK You are forgiven 😉
@@RetroBytesUK I also hate it when I edit and unrealized I said something wrong. But you are convinced you said it right. But yeah, we don’t have script supervisors 😅 At least I don’t, I you do… you may want to fire that person 😜
But seriously! Your content is brilliant and well researched.
I was thrown into the deep end with DEC Alpha's and VMS clusters in 1999 doing a great deal of large scale SCADA systems (city water systems), they are excellent for mission critical systems and fun to work with and learn. Yes, we had to do a lot of Y2K stuff, was one of the reasons I got the job.
You've been doing this for 2 years?! How am I only finding this now!?! Awesome stuff! Dropping a sub and i'll soon be enjoying the backlog! I've been computing since the single digit mhz days, this is absolutely my jam!
Very informative! I thought I knew a lot about computers but all I knew about Digital/DEC was that I saw it on some terminals as a kid, but you both revealed and plugged that gap in knowledge. Good videos
In 1999 worked in the UK on contract during the merger. In the UK had 20,000+ Exchange users. The Exchange environments of DEC and Compaq were being merged using a Microsoft utility I spent weekends watching thousands of users disappearing from DEC to Compaq environments. The DEC users were on Alphas consolidation meant that most of these had Exchange stores (IS) with circa 2000 users, the Alphas were powerful.
Absolutely loved my DEC Alpha Machine (433mhz IIRC). Ran WinNT on it with Lightwave 3D and Video Action NT with a DPS Perception video board. WAY faster than any PC at that time. It's still sitting in my garage. Have no idea if I could get it to boot now or not.
Back in the mid-90s, I worked on software that was pretty CPU-intensive and the state-of-the-art x86 processor clock speeds were around 250MHz. We bought a couple of Dec Alpha workstations that ran at a screaming 600MHz! They were amazing for the time.
I worked for DEC at South Queensferry 92 to 97 During that time we had a contract to make 486s for AMD on the Alpha platform. We had to redesign some of it to make our 486s run slow enough for the AMD market
Very interesting history lesson here, especially as I experienced a good portion of this history. My first job in the computer industry was managing and running a PDP 11 based system in 1979. It was a marvel at the time, having a series of removable 2MB hard disks that used software-based mirroring to provide data redundancy. Unfortunately Hurricane Frederick soon reduced a system of high redundancy to a single disk system. I later worked with a company that sold customized VAX system clusters and knew people who worked in DEC’s HQ. DEC was a well respected company at that time. I do have to say though that the PDP 11 was a workhorse and was very reliable and the VT100 was so easy to use and easy to write code for (interactive COBOL). Thanks for this trip down memory lane!
Fascinating stuff. Never came across one in the wild, and now I understand why. Thanks for this. Subscribed.
I worked in the VMS operating systems group from 1987 to 1991, and worked on the port of VMS to the Alpha. I enjoyed working on the Alpha port, but it was becoming obvious by 1992 that in spite of it's technical merits the industry wasn't going to embrace it. The layoffs started in earnest in the following years and I left the company in 1993. When Compaq bought the company at the end of the 90s it was kinda sad. But I still use DCU as my bank decades later!
I remember my "Sandpiper" workstation in the early 90s. Everyone else was ooh-aah about Windows 3.1 and I had a 64-bit computer with VMS. I had been a VMS developer since the Vaxen came out and RSX on PDP before that. I have a couple of the PWSs you showed in the video, but mine aren't in as nice of shape. I managed to get all my old files off the hard drives recently using a program called ods2reader cuz I couldn't remember how to get the TCP/IP on VMS to work!
Alpha had such an insane memory model, that proper multithreaded programming for it was a nightmare.
Very entertaining and educational. I worked at 3M Camarillo, where they produced the DEC DLT cartridges and had PDP-11s on that line to test the cartridges.
Alpha systems are very much still out there too. My company still runs them, albeit they are very close to the door and will be decom'd in the next year or so but they are still there (GS1280's if I remember)
If at some point they'll be decomm, make sure they don't end up in the e-waste pile at some recycler, a GS1280 if I', not mistaken can be disassembled into smaller machines.
love these stories. brings back a lot of memories, like working on RSX-11 and RT-11 and getting down and dirty with hardware. I planned to get an alpha to port my OS to, but sadly my company life conspired against me... Now I tinker away on ARM
My RSX-11 system used to crash regularly at 3PM and 3PM. No one worried about the afternoon crash, but the morning crash caused an Outcry. On the back of this I went on their RSX-11 course, and tried to blog the source code for the OS. All they gave me was a memory map. Anyway after a few weeks found that the shared memory count was only held in a byte, and it incremented more than it decremented, thus reaching zero and declaring the shared memory as free. Boom. I never did find the cause, but a little scheduled program to poke 20 into the shared memory counter kept the beast running.
I was never at DEC. most of the equipment that I was responsible for used DEC CPUs, which was how I know what I do about DEC. I worked on Genrad 227X and 228X in-Circuit testers. The 227X (I worked on the 2270 and 2276(XP)) were PDP 11 family based, the 2270 was a PDP 11/03, and the PDP 11/43 was in the 2276 (later updated to a 2276XP) whrn the CPU was changed to a PDP 11/73). We also had a programming station ( with 2 VT-100s) which ran on a PDP 11/23. Operating system was RSX-11M+. When we outgrew that, we moved to the Genrad 228X family, which were uVAX 11 based systems that used VAX VMS. It was a night and day difference between the two systems. The programming station was a VaxStation Midel 30 with maxed out RAM.
Another thing that should be mentioned is that DEC seemed to have its networking capabilities from the earliest days. The 2276XP was actually networked to the 2293 programming system using DEQNA cards on the QBUS on the two machines ( I know we used a rigid Ethernet trunk cable with vampire taps as physical wiring, but I may not have the correct name of the interface card). The VaxStation the the 3 228X testers were also networked the same way.
Awwsome watch. Loving your work.
IBM bought PWC Consulting, not Arthur Andersen Consulting, which had bought itself out, and eventually rebranded as Accenture…but love the vids, keep then coming 😁👏🏻
BTW, the first Alphas had a big design flaw: they were designed for scientific use but most of DEC's cistomers were business users.
They had only implemented fiull 64 bit longword access and trying to access memory that was not on a 64 bit boundary caused a fault and pal code would then convert the memory access to a 64 but boundary and extract the desired results from it. This resulted in a huge performance penalty for business programs, especially those in COBOL that deal with a lot of strings. Subsequent Alpha chip was provided with ability to access memory without such penalty.
With regards to drivers, one of the biggest challenges was that graphic card manufacturesr did not document their interfaces as they expect everyone to use their Windows drivers on x86. Digital engineers had to reverse engineer how each card worked and they needed to implement a mini x86 emulator at boot time because the card itself had x86 code that needed to run on the CPU to set registers etc. So getting drivers was NOT easy. (and to this day, Linux also has similar problems wioth graphic cards). When DEC built its own cards, it obviously knew the ful interface so was much easier for them to buidl drivers for them.
With my Alphaserver GS60, in order to avoid the non-aligned access penalty, the PCI address space appeared three times in the memory space. Firstly as "normal", then another sparse copy where each 16 bit word was aligned to a 64 bit boundary, lastly another sparse copy where every byte was aligned to a 64 bit boundary. This allows access to every combination of data type with proper alignment, at the expense of a large amount of wasted address space. Drivers need to know about this so a "simple port/recompile" of an x86 driver is insufficient in most cases.
@@thisnthat3530 Yes, this was a problem (design flaw, really) in the EV4 and soon fixed. EV4 would only read/write full words (64 bits) and VME need byte addressing. The board (not chip) designers had to perform that kluge in order to get VME to work. EV5 and beyond allowed byte addressing. And don't get me started on EV4 IEEE floating point :-).
@@gbsccfig Yeah, the EV4 was special. It also had no divide instruction so division had to be performed entirely in software. The transistor count was relatively tiny at only 1.8 million though.
Great video, loving your content!
Also fascinating reading all the comments here recounting good memories with DEC and Alpha, really seems a massive shame it all ended up like it did.
Thank you for this teleportation into my youth. Once DEC, always DEC...Felt like an IT god using VAX/VMS in those glory days you mentioned.
Unbelievable to see this video. I was on this journey on the other end porting our EAM software to every variation DEC came up with (and every other Unix variance of competing vendors) .... all the way to Windows NT Aplha..
The best Port ever was the Tru64 Alpha... stable, reliable, and the quickest.
How DEC squandered such a great system.
YES! I got my first Alpha workstation at the MIT swap meet they hold early in Cambridge MA in the late 90's. I had to lug that thing from the parking garage where they held it, to the train station, then to the bus station, then over 2 miles back to my house. 🙃
A few years later I went on to work at a place that built/specialized in Alpha Linux clusters right out of high school for $9/hr. 😜
Those were the days.
2001 I started as a computer operator for a hospital in the northwest suburbs of Chicago. I had to deal with everything your video mentioned. Miss it with my heart n soul.
Growing up, my father worked at DEC in Merrimack, NH. They moved us from Maynard, MA when the new campus opened. We had a VT100 with a “WarGames” style acoustic coupler modem, then later a Rainbow PC. It had two floppy drives and a dual boot hard drive with both CPM and MS-DOS.
I seem to remember DEC Alpha workstation ads heavily pushed in 3D and video magazines of the time. A lot of 3D animation was being done on Alphas because rendering was so much faster and I believe models could be more complex as well. I wanted one and I'm sad that DEC went kaput. I once wrote a short story where DEC won the processor war and the world now ran on DEC Delta processors.
I have heard that the graphics in the Titanic movie was produced on 200 DEC Alpha computers.Can't verify this but we were told that internally in DEC.
@@oladunk9986 Probably correct. DECs running Maya or Amigas running Lightwave (and the Video Toaster) were pretty much industry standard going by what I saw in Vancouver during the 90s. Win95 and Intel were just starting to get traction and get into the game.
I had the pleasure of restoring one of these things(Digital Personal Workstation 500au) to functional condition and I must say, the day when it booted Windows 2000 for the first time was a good day for me indeed. Love the sound hardware and its CPU is deceptively powerful - it will without too much complaint run MODERN windows programs! I also went through the trouble to track down a contemporary GPU but I can't find any drivers for it. Don't really have anything for it to do at the moment though.
Another great video! very informative, and entertaining and well made keep up the good work.
I remember salivating over the DEC Alpha workstations in the mid 90s. I wanted one so bad. Those were some amazing times.
Well done, sir! Loved the video. A major part of my career was working for DEC from 1979-1999. I was a software engineer in DECnet Engineering, RSX-11 OS engineering, the transaction processing group, DCE engineering, VMS engineering, and finally, the Tru64 Unix engineering group. The sentiments about Ken Olsen being stabbed in the back were absolutely felt by everyone I knew and the subsequent CEO, Bob Palmer, was absolutely loathed by most of the employees.
I joined DCU the day it opened in The Mill in Maynard and am still a member. Was sad when we were forced to change account numbers this year from our original DEC badge #'s ( during their transition to a new banking system).
I met my wife at DEC ( couples at DEC were typically referred to as "Digits") and we are still very close friends with many other folks we met during our careers there. As evidenced by my time there, you didn't have to leave the company to branch out into new/other technology areas you were interested in. There were so many of those available right at DEC and, under Ken, it was simply the progressive type of company just about anyone would want to work for. The talent I worked with in every group was just outstanding and incredibly diverse.
Thanks for stirring up a lot of very fond memories.
The university I worked at had a couple of Alpha 150 boxes, one running Windows NT and the other had Linux on it and both were incredibly powerful machines compared to the pc based systems. My mate who was the head of the server room got to take both machines when the uni "upgraded" to Intel's Itanium platform. His wife was a programmer and she did a bit of fiddling and got a 3dfx card working on the linux box and we all used to play Quake on it. It was an incredible experience at the time.
Thanks for this video. Informative and entertaining, a great combination.
OMG, this brings back lots of memories , has one of the best workflow envionments