Slashdot Mirror
Alpha Up For Grabs?
A number of people have been writing about Compaq selling off the Alpha processor, with some coverage from different media sources. The Inquirer cites Intel as the likely buyer, which seems odd to me considering their aversion to antitrust lawsuits. Maybe AMD? Who knows - it's too bad that the Alpha technology has never realized the same commercial success as it has technologically.
I had hoped that Compaq would aggressively market Alpha with the DEC acquisition, and would offer us a choice in the IA32-IA64 migration.
I had hoped for fast and reasonably-priced Alpha systems. These never materialized. You never even gave the architecture a chance - the marketing was nonexistent.
I've had a reasonable level of respect for Compaq equipment, but now I hear that Compaq wants to reposition itself as a services company.
Shame on you, Compaq. You are the second largest computer company in the world, but it looks to the public that you are lackeys, easily threatened and controlled by Intel and Microsoft. You could have made the market a better place, but all that you've done is make everything worse.
I guess that it's all in Sun's hands now.
In other words a lot of gifted EE's in Massachussets are getting creamated. Must say
the Alpha had the fastest memory bandwidth, even faster than AMD 760 DDR. Now I'll have to get a real VCR for recording movies like everyone else.
DEC was renowned for nothing so much as their inability to market good products.
Uh, I thought that company was IBM? ;-)
However, another question also arises. The Alpha has been around for, what, a decade or so now? Possibly the architecture is nearing the end of its life cycle, and if so no one is going to want to spend much to acquire it.
If you take a look at the market, Alpha is one of the "youngest" chips around. x86 is succesfull for more than twenty years now. And SPARC ('87) and POWER ('89?) reach their 15 yrs anniversary.
I also remember a paper from DEC with a planned lifetime for the Alpha architecture with 20 or 25 years.
I think that time has proven that you don't need a completely new archicture to keep pace with the technical advance. (It's about evolution or revolution.) AFAIR Tomshardware has a great article about (changing) chip architecture and (pretty static) instruction sets.
We did two hardware board designs with the DEC Alpha 21064 and 21066 back in 1992. The architecture and the processor were far ahead of it's competitors then (64-bit architecture, 150 MHz).
It's a pity that the Alpha had not more success!
> I would think it would be a lot of money to buy a product that would not be that useful to either company except to maybe get some ideas on improving there own chips.
Actually, most of the ideas are readily accessible in published papers. There may be some patents locked up in the Alpha, but I doubt that there are any secrets.
> I dont think there would be a large market for the Alpha being that Compaq and DEC could do nothing with the chip.
DEC was renowned for nothing so much as their inability to market good products. Dunno about Compaq, but without an NT for it I can't imagine that they would know what to do with it (which probably explains this article).
Your VIA suggestion sounds good. However, I think it would be a shrewd move for AMD too, if they're in a position to market it at a more competitive price than it traditionally has been.
Also, parts of AMD's architecture has been converging with parts of the Alpha's (even on Athlons), so AMD might be able to integrate it into the high end of a "family" of processors.
However, another question also arises. The Alpha has been around for, what, a decade or so now? Possibly the architecture is nearing the end of its life cycle, and if so no one is going to want to spend much to acquire it.
Anyone know how many more years they can squeeze hot stuff out of the Alpha? And is there any margin for cutting the price a bit?
--
Sheesh, evil *and* a jerk. -- Jade
-jhp
/. -- the Free Republic of technology.
My question is, why? I didn't believe this when I read it earlier, and still not now. The Alpha is a viable architecture, proven and powerful. Intel could gain patents and technology, but I thought these issues were, for the most part, resolved years ago in a secret settlement. Intel bought DEC's fabs, DEC dropped lawsuits against Intel over the P5 and other 'things.'
So Ace's says that the newest compilation of SPEC outperforms the Itanium (Merced). I think the Merced has a lot of potential, in the fact that it isn't the cleanest design (more of a proof of concept and a 1st attempt to learn from), and that I doubt the Intel compiler is very up to par. Yet it still gives an impressive performance, if you believe SPEC.
IA-64 and Alpha are both viable at scientific applications, and from the latest Compaq compiler, they are relatively equal in their current forms. The Alpha wouldn't die because of the 3rd party consortium (forget name: APR?). And I've read claims that many of the best engineers left when Compaq bought DEC and moved to AMD amongst others. So, what is the major gain Intel would get from this?
-----------------------------------------
"Open Source?" - Press any key to continue
If Alpha passes to Intel, it is likely to be phased out -- Intel will milk it of its best technology, or at least of the part that they find compatible with their own rigid ideas, and scratch future Alphas. They want the team, but mainly to improve on IA chips.
This is dangerous, for the same reason that other monocultures are dangerous. Plant too much of the same "improved" seed variety and diseases/pests particularly suited to that variety will erupt/spread. Spread one architecture too far and good ideas from others will get lost, and progress will slow down.
Already we've lost lots of good ideas from the early days of computing. Some of the Burroughs CPUs of the 1960s had advanced features that modern CPUs would benefit from. Multics and TENEX/TOPS-20 had features that "modern" OSs, like the many variants on 1969's Unix, lack and could benefit from. The economic benefits of spreading one design (h/w or s/w) across many units usually outweighs the benefits of a better design. At least in the short term, but then we lose the long term benefits.
That's where Alpha got clobbered. DEC had no marketing skill. Alpha was DEC's fourth in-house RISC design (after SAFE, Titan and Omega, and those are only the ones I can remember offhand) and its designers learned a lot from the weaknesses observed in SPARC, MIPS (which DEC used for a while) and other earlier designs. Alpha has unique features. It morphs into a VAX, NT or Unix machine via a code layer that other CPUs don't have; Transmeta is not quite the same idea but at least has some parallels. Its floating point processor still blows the doors off of Intel's or even the superior AMD. It's a clean architecture, unburdened with IA-32's 8080 compatibility (itself a kind of PDP-8 heritage).
But none of that mattered; Alpha never got volume, so it was always a niche machine. VMS still has strong markets (read Terry Shannon's SKC stuff, for instance) and it depends on Alpha, but that's apparently not enough these days to sustain PC-centric Compaq. How sad.
OK, then, so why not have AMD buy Alpha, if for no other reason than to deny Intel access to it?
Both companies have about the same amount of cash. AMD may even have the stronger balance sheet than INTC. What's the Alpha division really worth? (And can it be bought for less, perhaps CPQ is open to selling it at fire-sale prices in order to clear the decks for their new "We can't beat DELL when it comes to moving hardware, so we'll sell support/services" strategy?)
I can treally see why either Intel or AMD would want to spend the $$ on Alpha. They both have there own 64 bit CPU's finished (or nearly finished). I would think it would be a lot of money to buy a product that would not be that useful to either company except to maybe get some ideas on improving there own chips. I dont think there would be a large market for the Alpha being that Compaq and DEC could do nothing with the chip. IThe only company I can think that might want it would be VIA so they could make an alternative to the Intel and AMD chips like they did with the Cyrix chips.
"If ignorance is bliss, why aren't there more happy people in the world?"
Dear Compaq HPS customer,
c ement
Please see the attached note from Compaq's Area Director for North Atlantic. I will follow-up during the week to see if you have further questions.
URL for customers for Monday's webcast.
The following URL has been provided to us by Corporate for you to communicate to selected customers who would like to participate. The time of the webcast is Monday, June 25 at 9:00am Eastern time.
URL http://webevents.broadcast.com/compaq/PressAnnoun
I'm working on a super computing project up here in Canada, known as SHARC-Net. It is a group of Beowolf clusters (using Alpha's, with all the hardware supplied and serviced by Compaq) located at 3 different universities (as a side note 2 of the clusters run Linux and the other Tru-64). The project directors had a conference call with Compaq on Friday, for which they had to sign NDA's. When the Director for the University of Guelph came out of the conference call she was very unhappy with Compaq. She told us this:
1. Compaq will be announcing to the world what they discussed on Friday.
2. It won't effect us until 2004 (which agrees with the article).
3. Had they known about this earlier it could have effected their choice of supplier for the clusters.
This is very little to work with but it does agree with the article. Needless to say, may of Compaq's customers are very unhappy with them right now (including us as we are just now bringing these clusters online).
Compaq did it. It's no longer a rumor. Read the press release on Intel's web site: http://www.intel.com/pressroom/archive/releases/20 010625corp.htm
Motorola is primarily an embedded processor maker, and perhaps the Alpha can see some life there as well.
If you can't beat them, embrace and extend them.
Intel only owned the parts of Alpha that it already owned because Intel was already using Alpha technology in its chips.
Remember that Intel and DEC settled DEC's infringement suit when Intel bought the Alpha lines.
That same agreement multi-sourced Alpha at Samsung, AMD, and IBM. So there was and is no danger of Intel's monopolizing Alpha.
Compaq then bought all of DEC, and ended up owning whatever Intel didn't buy. Naturally, that sounds like an inefficiency. Compaq can't handle inefficiency. Intel is organized to mediate inefficiency and even find ways to profit from it (they build a fab for one chip partly on the premise that once that chip is done in the market they can use the fab line for less-mainstream products; they've done this for 30 years; some lines are designed knowing that their primary product--this year's desktop chip, for example--will never be enough to pay the mortgage; it's a gutsy and thoroughly pro move).
--Blair
Microsoft will buy it, turn into a hardware company, and release all their software as GPL.
--
Two witches watch two watches.
--
Two witches watched two watches.
Which witch watched which watch?
Really, what did you expect? The series 7 ('364) is designed by people that have no clue WTF they're doing. It's a '264 core with a Rambus controller on it, not much more. The '264 came out in '96, so 5 years later and they still can't release a new product? I smell incompetence.
Ahh, the series 8 ('464). Now *that's* a pretty chip. 8 instructions/cycle issued, something like 256 in flight at once, multi-threaded core. Tick for tick it has twice the performance of the '264, and it runs at twice the clock. Too bad they won't fab it in state of the art processes.
Alpha could be a good backup strategy for Intel: it's a more traditional architecture with lots of existing compiler backends.