9 ms·
DEC Alpha: A History in Facts and Comments (2005)
- Lammy 7y agoThe title of this submission doesn't do a good job conveying that it's about the DEC Alpha CPU architecture. It says so in the first sentence, but that doesn't matter if the headline doesn't grab you. This is one I wish mods would editorialize a little :) https://en.wikipedia.org/wiki/DEC_Alpha https://en.wikipedia.org/wiki/DEC_Alpha
- Lammy 7y agoThank you, whoever made the change!
- bsder 7y ago> Olsen refused the offer saying that the VAX architecture hadn't reached its end-of-life yet and EV4 was still a work in progress. The rumor from multiple directions is that the Olsen and Sculley actually had a deal for Alphas in Apples. And then Olsen found Sculley "in flagrante delicto" with a woman not Sculley's wife--which incensed Olsen enough that he torpedoed the whole deal.
- tyingq 7y agoYou know you're on to something when Microsoft ports Windows to your ISA. At one time, NT supported Alpha, MIPS, and PowerPC.
- ch_123 7y agoThe Alpha support was mainly because of a lawsuit, I suspect that was also why it lasted the longest of the non x86 ports of the 90s. Reference: https://www.itprotoday.com/compute-engines/windows-nt-and-vms-rest-story https://www.itprotoday.com/compute-engines/windows-nt-and-vm...
- sterwill 7y agoPerhaps it lasted longer because you could actually buy Alpha computers from DEC with Windows NT installed. Could you ever buy MIPS or PowerPC computers from a major vendor with Windows NT installed?
- tyingq 7y agoNot saying it was a raging success, but yes, you could buy MIPS machines with Windows NT installed: http://www.sensi.org/~alec/mips/acer_pica.html http://www.sensi.org/~alec/mips/acer_pica.html
- bluGill 7y agoInteresting. I was not aware of them at the time, while I was aware of Alpha. My impression was alpha competed well with the high end wintel systems, with a better price/performance ratio. There were no cheap alpha systems available though, nor where there laptops (only latter did I find out how power hungry alpha was), so those of us who didn't spend a lot of money or wanted a laptop didn't try alpha but those who bought better systems considered Alpha the system to get. (Note that I have always been a BSD or linux guy and don't know what the Windows world did) What is missing from your link is price.
- sterwill 7y agoI had never heard of the PICA-61, thanks! I owned a DEC Multia back in the late 90s, when they were already obsolete and plentiful. It had SRM firmware, not ARC, but I didn't have any interest in running NT on it anyway. Ran Linux really well and used it as my personal server for years. Later I put NetBSD on it for a short time, but I don't remember why. I remember the it frequently printing messages about unaligned memory traps to the console, but it did the job. I'm sure Microsoft was sincerely hedging their architecture with NT on MIPS/Alpha/PowerPC/i960/etc at the start, but by the time that hardware shipped with NT on it was pretty clear all the momentum wasn't leaving x86. On the 64-bit Alpha, I believe NT userspace was still 32-bit and so were all the applications. I don't remember any 64-bit Alpha NT applications ever shipping, but I admit I wasn't watching that scene very closely.
- m0llusk 7y agoThis is one of the most clear and also tragic demonstrations of a marketing to engineering ratio driven failure: "A long time ago, Kenneth Olsen, who was a founder, president and CEO of DEC until almost the end, said that well engineered products would sell themselves. So, there is no real need in advertising campaigns or other instruments of market promotion."
- robk 7y agoStill blows my mind how NT ran on so many other platforms yet no one was able to really capitalize on it. I spent one of my summer internships in Littleton MA doing QA on the NT networking stack for SMB and other lanman features. The machines kept my cube hot despite the building AC but seemed pretty performant versus my i386 home PC at the time.
- ubermonkey 7y agoIn the mid-90s, I worked at what had to be one of the last serious DEC/VMS holdouts: TeleCheck. That they had ended up on VMS was not evidence of a bad choice, but by 1995 it was pretty clear which way the market was going, and that they were about to be kinda out in the cold. They'd also managed to give themselves an ENORMOUS case of the MUMPS (1), brought on by an almost religious devotion to "not invented here" syndrome. (Part of that was justified -- they couldn't get what they needed out of a commercial DB early on, so they wrote their own. And then KEPT writing their own well past the point when they should've been evaluating migration to COTS. But that's just a background detail.) Anyway, at that point if you read any industry press, or went to any industry event, or were even vaguely aware of computing trends, you KNEW the future didn't include VMS, and that even the Alpha architecture -- which was cool and powerful! -- was doomed. But not at 5251 Westheimer; there, the Kool-aid was DEC-flavored, even to the extent of making bizarre counterfactual claims about COTS options, or other hardware, or whatever. "Sun" was a dirty word there; hatred of Unix was an article of faith. It probably won't surprise anybody to hear that turnover there was INSANE. I stayed 25 months before moving on; that was about normal. Those that did stay became more valuable to the organization, but not so much that they ever got market raises or anything. And of course by staying, they let their skills get farther and farther out of sync with the rest of the industry, thus making it harder to leave. 1: https://thedailywtf.com/articles/A_Case_of_the_MUMPS https://thedailywtf.com/articles/A_Case_of_the_MUMPS
- Rochus 7y agoWhat do you have against MUMPS (i.e. ANSI M)? It was a great technology of its time, and there are still many critical applications running with it today, and internationally recognized companies like Intersystems whose products still run ANSI M inside.
- ubermonkey 7y agoRead the link. Companies that develop their own idiosyncratic dev systems are generally making a mistake. Developers who endeavor to master those systems are using energy and time to add a skill to their resume that has zero value on the open market, which is a bad career move.
- drewg123 7y agoThis article brings back so many memories. My first job was as a sysadmin for a highly computational statistics department in 1993. The first alpha we had was a DEC 300/500X ("hot pink flamingo"). I remember adding 256MB of RAM to the box on one one of my very first days on the job. At the time, the price for the memory was well into the 5 figures, and I was thinking that I'd better be careful, because the memory was worth more than I was. In a later job, I was one of 2 people to port FreeBSD to the DEC Alpha. We never supported the older turbochannel boxes, but I did a lot of the chipset support for most of the PCI alphas. Seeing the APECs diagram brought back memories of implementing the board support for them (and most of the PCI based alphas). I had an API UP1000 running FreeBSD as my desktop up until 2003 or so. That box was given to me to do the FreeBSD platform support. Much later I was lucky enough to work with Dick Sites at Google. When I realized I was going to be in a meeting with him, I brought him my copy of the alpha architecture reference manual to autograph. I can't say enough good things about him. In addition to being a luminary in the field, he's a truly wonderful human being.
- dekhn 7y agoI truly loved the alpha but by '97 or '98 it was clear it was not a good platform for me. I remember being told to buy a basic alpha for $15K. Instead I bought an entire Linux cluster for that price, learned to manage it, and got all my scientific work done at larger scale than other people in my department.
- jacquesm 7y agoAt roughly that same time I did buy an Alpha but for one reason only: it could use more than 4G of RAM in Linux and files could be larger than 4G. It was the cheapest 64 bit platform at the time and I needed that to be able to use very large MySQL tables. Very nice hardware, super reliable and it ran for many years.
- thedance 7y agoI was buying ES40s at work around that time for the same reason. It was a lot of memory in one box, and close to the most memory you could stuff in a single rack. Unfortunately dealing with Compaq sales at the time was horrible. They basically did not know what the web was, or why anyone would want to serve it.
- dekhn 7y agoYup. At the same time I bought my linux boxes, the department next door was purchasing ES40s, assembling them into TruClusters, and running modest MySQL tables. They ran on that until the hardware and software was EOL'd. Later I worked with an 8-node TruCluster composed of monster GS1280s. It had some nice properties- especially the high throughput IO (100MB/sec!) The industrial version of the human genome was assembled on large-memory Alphas for the same reasons (lots of RAM, predictable disk IO).
- downerending 7y agoSounds ridiculous, but I still remember the lack of synchronous reporting of divide-by-zero errors, and wasn't really sad to be rid of it.