One of the books I was required to read back in college was Paul Ceruzzi’s “A History of Modern Computing.” If you are looking for just one book that is jammed pack full of both the history and the technical jargon for the world of computing, Mr. Ceruzzi’s book fits the bill quite well.
Among the various computing companies discussed, Digital Equipment Corporation (DEC and later rebranded to Digital.com) itself could have volumes written about its rise and fall as the second largest computing company in the world (it made databases, operating systems, CPUs, switches).
It was founded in 1957 in Maynard, Massachusetts by MIT engineer Kenneth Olsen (and his brother and a colleague), who created the company after working on the TX-2. Reportedly when Mr. Olsen went in for a bank loan, he was instructed not to say his company was planning to build computers but rather digital equipment (hence the name). I’m not sure if this is true though, because in his interview with David Allison he states that American Research Company originally invested $70,000 in his venture.
At any rate, DEC made a huge splash with their new mini-computers (as opposed to the massive multi-room variety), starting out with the PDP-1 (Programmed Data Processor, which Ceruzzi discusses in length).
Throughout the ’70s and ’80s, DEC grew at a phenomenal rate which ultimately climaxed at the end of the ’80s. With over 125,000 employees worldwide and billions of dollars in sales each year, DEC hit The Wall, fell down a stairwell into a street, run over by a car and later picked apart by a couple of crows.
Throughout the ’90s DEC attempted to reassert itself as the preeminent provider of computing technology:
1992 they unveiled the Alpha 21064, their next-generation CPU workhorse and first commercial 64-bit CPU
1993 they begin shipping 200-mhz Alpha’s in volume
1994 they sell off their RDB database software to Oracle
1995 Alta Vista goes live, originally started in the deep dungeons at DEC’s black-ops Palo Alto office as a way to index the web and for companies to use in private intranets
1996 Alta Vista is then made available online for the general public (altavista.digital.com) and DEC discontinues their home PC Starion line (catchy name, eh)
1997 they sell off their printing division to Genicom and end up being countersued by Intel
1998 to settle the Intel litigation, DEC sells its semiconductor operation to Intel. Then on January 26, 1998 DEC is acquired by Compaq.
I found this article amusing as it is written in February of 1998, a week after the DEC-Compaq merger was announced — with the central theme “will the Alpha remain the fastest chip?” If the development of the chip continued as originally planned, it would have remained the fastest chip, but today the answer is no (though it took several years for the competition to catch up).
Throughout it’s 10+ year life, several operating systems were ported or built for the Alpha. NetBSD support came along in February of 1995. OpenBSD was ported to the Alpha in 1996. FreeBSD was ported at this time as well (no exact date was readily available). AlphaLinux, a distribution of Linux for the Alpha was also created (all of these are still maintained to some extent).
In 1993 DEC managed to convince Microsoft to make a port of NT for the Alpha (no small feat), a year and a half after Compaq acquired DEC (in mid-1999), MS and the Q stopped Windows development for the RISC processor.
Internally, VMS and OpenVMS were ported to the Alpha as well. OpenVMS is simply VMS renamed (*cough* synergetically *cough* it seems) and was neither Open or new. VMS was the operating system created for VAX (the DEC workhorse prior to the Alpha). Note: Compaq developed an operating system for the Alpha (originally for others as well) called Tru64 which still has a large active community.
Michael Capellas (who was the Chairman and CEO of Compaq prior to the HP merger) let the cat out of the bag spelling the doom for the Alpha platform in this letter to OpenVMS/Tru64 users on October of 2001. Comically, the Merger link at the bottom of his letter routes you to an old ‘Vote Yes’ merger campaign page at HP. A little dated, but just another sign of the times at HP (here are a bevy of insights from the Garner Research Group regarding the merger prior to it occurring).
Last April (2002), the HP-Compaq merger was officially completed. This doomed the Alpha primarily because HP is the co-creator of the Itanium (along with Intel) and needs the Itanium to succeed financially — as you will see, HP kills the development of the Alpha shortly after the merger (HP even canabalized their own PA-RISC line which performed better than the Itanium).
Today’s Opteron has the unofficial motto of being the ”Poor Man’s Alpha.” Not only does it perform at levels similar to the 21264 from yesteryear, but several innovative designs were brought over by DEC engineer Dirk Meyer who worked on both the 21064 and 21264 (in addition to the EV6 bus which AMD licensed from DEC).
This past February, Paul DeMone continued his analysis of the 64-bit market, including the Opteron. He also discussed the current life of the Alpha and predicts what HP will do, stating:
Large scale EV7x systems will be relatively inexpensive to build (look ma, no chipset!) but expect HP to charge top dollar regardless. Since the acquisition of Compaq, HP’s official policy has been to direct all new customers to IA64 hardware and limit sales of Alpha systems to the remaining customer base. But with PA-RISC long toothless for technical computing and a number of vendors like SGI offering huge Itanium 2 based systems optimized for HPC, sheer pragmatism will likely force HP to sell EV7x gear to anyone with approved credit. Obscenely high prices will ensure the Alpha tail doesn’t wag the IA64 dog.
In the past Mr. DeMone has posted at Ace’s Hardware, which itself has discussed the EV7 several times too. Back in September of 1999 Ace’s concluded, “Provided the 21364 can be delivered in sufficient quantities, it will garner the same appearance of strength and stability, making for an easy victory.” The Itanium bombed (sux0r3d), but as Intel and HP control the majority of the Alpha intellectual property and engineering talent, they were able to easily kill off the possible contender.
Several weeks later, Johan De Galas (at Ace’s) started his series on ‘The Secret of High Performance CPUs.’ Notice just how far ahead of the pack the Alpha still is at this point in terms of rough benchmarks (not real-world) and clock speed. That trend was supposed to continue too (a 1000x performance increase in 25 years, the EV9 was targeted at ~77 GFlops and slated for production in 2005-6).
In May of 2000, Ace’s benchmarked AlphaLinux and concluded:
OpenSource and Linux is here to stay, and Linux offers a huge opportunity for Compaq to promote the excellent Alpha Platform. We URGE Compaq to be more aggressive and secure the future of the Alpha processor. Compaq should include the Alpha Linux Compiler with each entry level Alpha server and work together more vividly with Samsung.”
However, Paul DeMone’s prediction regarding the death of the Alpha in the hands of HP appears to be on the mark. As of this writing the EV7 21364 (which was supposed to come out over 2 years ago) costs a small fortune, a calculated strategy that will most certainly keep new blood from purchasing that solution. The EV79, the last planned Alpha processor release should debut sometime early next year, but don’t expect competitive pricing, it’s in ‘maintenance mode’ (preping for CPU heaven).
And for those of you that think you might be able to snag the stellar technology through other venues, PriceGrabber comes up negative for any actual Alpha. They certainly have enough listings for RAM modules, but I don’t think that’ll get the women in the sack (or whatever you try to impress).
In conclusion, the Alpha is the BeOS for the contemporary microprocessor world. It was originally launched with enthusiasm and recieved accolades for innovation, however it did not muster the necessary marketing momentum to win the hearts and minds of IT managers (though Intel FUD didn’t exactly help — obfuscating can win short-term sales but builds a shaky ground for a long-term relationship). In the end, this blogger predicts the Opteron will soundly defeat the E2K Itanium technically and with any luck, the marketplace.