Slashdot Mirror
Compaq Cutting... Alpha?
CuriousGeorge113 writes "As a result of Compaq's recent acquisition of
Digital Equipment, they plan on closing the
Salem, New Hampshire plant which employs 900
people. Most of them will be laid off. The plant currently
makes alpha-based server computers. "
This is a prime time for Compaq to jump whole-hog into the Linux arena. It will definitely keep their Alpha afloat.
Why did Compaq even bother to buy DEC? They sure seem to be killing everything that DEC had of value.
Why do you think Sundisk changed their name to Sandisk... Yep, you guessed it. Slashdot mispelled their name. ;-) Welcome Campaq.
Don't forget Intel acquired all of Digital's manufacturing (including MA FABs where the Alpha is made) about a year or so ago as part of a legal settlement. If memory serves Intel agreed to run and manage the FABs including Alpha production, however DEC retained rights on the chip design.
Since you seem to be in the know.. Where is the NX chip? I'd love to have some 'real' (64bit) hardware to test apps on, but with PIII450s selling at under $300, the price performance is just not competive with 21164a and 21254.
that's a lot of people...
And, though SGI has announced NT server, it has yet to make any.
Incorrect. They're shipping, and have been for a while now. The reviews I've seen make them out to be pretty hot boxes.
See http://www.sgi.com/products/hw_wor kstations.html
Wow! I hate to see slashdot being perveyors of such FUD. Yes, DEC had a plant in MA that produced the Alpha chips, but the plant was NOT capable of producing the smaller die sizes needed for the faster chips. The antitrust induced deal actually leaves Intel having to produce the Alpha for Compaq, and Samsung and several other companies produce the Alpha chip. People, Compaq wants others to produce the chip. The fabrication plants for the smaller die sizes are ENORMOUSLY expensive. Please, read the article you script kiddies.
As I see the world there is nothing wrong with choice. You don't need to use Linux on all machines. If your company has the funds to buy machines for >$100.000 it can surely afford on of the commercial Unix versions.
I don't understand the mentality "Linux must be better than all alternatives, everywhere, all the time !"
Right now Linux doesn't scale well. ("not very well" is a gross understatement for something that gets beaten by NT). But on the other hand this gives the benefits of simplicity and the possibility to run on lower-end HW.
Yeah, roadmaps are easy to write.
That Alpha wouldn't survive long shouldn't be a surprise to anyone.
It has long been apparent that only x86 (Merced) and perhaps some other chip would survive. Maybe Power/PowerPC ?
The reason is quite simply the steadily increasing amount of money you need to invest in reasearch and you need high volume for this.
For a time the RISC technology looked like it would reverse this trend but it only managed to delay it for a decade until all benifits had been included in x86.
So Alpha, Mips, Sparc and so on are all deadend products and now the rats are leaving the sinking ships.
With Intel about to try to get everyone to move away from x86 (and so move to Merced) Compaq would have to be very stupid to drop Alpha now. Indeed everything I've seen from Compaq and the Alpha team suggests they are getting ready for a fight just about the time that Merced arrives.
Intel says 'Move away from x86 and go to Merced'
Compaq say 'Move away from x86 and go to something which works now - Alpha'
Yes and no. Mostly no. The recession in the early 90s was due, in large part, to downsizing by a lot of companies, not just DEC. Recall also that WANG went bye-bye, Data General pretty much poofed, and Lucent/Bell Labs did some down sizing.
Things look a lot different now - the tech industry in the NH/MA area is booming, with tons of startups that are sucking talent away from Cabletron, and various others in the region. I left for a startup in January, and know of at least 1/2 dozen colleagues that have done the same since, just from my group.
Cabletron definitely has its share of problems, and the decisions that led to the revenue shortfalls were just the beginning. Now that they seem to be getting past that, they need to recover from significant losses in engineering talent. It may be that they won't, at least not in NH - the balance of power in ctron engineering seems to be shifting to the west coast with Yago, Zeitnet, etc.
yea ok but observe the crappieness of compaq's pc's, anyone i know recommends them last of all big box name vendors... they are built poorly, sure the avg user might not think so, but does the avg user know about delays on apm, and disk swapping? my biggest fear is that they'll kill alpha because they can't manage it, it was doing fine under dec, and those math libs? no big deal, the opensource market would've gotten to those.
DEC killed themselves when they killed the PDP-10s
and the PDP-11. They're just still trashing about
in their death throes. Had they never killed TOPS-10, and if Jupiter had been built, who knows
what could have happened...
Alpha will survive and so will Sparc. I'm not sure about Mips.
The research was done years ago and what about Samsung?
I do agree with your mention of delay and feel that was Digital's biggest snafu or "boyz in the back room makin deals with a certain software company". (^%;
Compaq still has a chance to be the first on the block with 64 bit based architecture, if they play their cards right.
Shhhh... and what about that 128 bit Alpha chip on the drawing board?
Compaq is deliberately killing Alpha and no amount of posturing from their "Supply Chain Management" can refute that now. EV-7 could be killed in a heartbeat and everyone involved with it knows it. Wildfire is very, very late and may prove to be the headstone. What those arrogant Texans don't seem to want to realize is that they don't know shit about building high-end systems - chances are most of them have never even seen an Alpha, let alone worked on one that was actually running, particularly with something other than Cutler's Revenge on it. They let the cat out of the bag last year when they referred to the facility as being in "Salem, Mass." in an internal memo - it shows how little they pay attention to reality. They know damned well that the bulk of the workers in Salem - those that *know* the Alpha and the products made with it over the last 7+ years - won't want to move to TaxUs or the People's Republic of California, and they will use that excuse as well as "declining sales" to shut Alpha down. In other words, it's a self-fulfilling prophecy, and the proof came last week when they announced funding for Intel's Merced start-up costs.
The fact that an EV6 is twice as powerful as an Intel chip, has exhibited better MP scaleability and that there are Operating Systems with real Clustering capabilities that only an Alpha can run is, of course, beside the point when all you can aspire to is to be is king-pin of the Wintel cartel.
Sometimes it makes economic sense to run Linux on big hardware, and for some purposes Linux outperforms commercial OS's. We measure Linux against Solaris and NT for its technical merits, not on cost, and Linux can hold its own.
.zip archive (over 50,000 files) on NT if you want to see for yourself.
Don't worry too much about Linux and SMP. It's very young, and has come along nicely. I have a 4-way Linux machine here in the lab, and BTW it handily beats NT on many real-world tasks.
NT does scale better than Linux in a few significant areas, most importantly read/write concurrency. The benchmarks we've seen are designed to exploit that. It's easy to come up with benchmarks where NT looks very bad, probably easier than doing the reverse (making NT look better than Linux)... one that comes to mind is file creates... NTFS seems to log/write the meta-information synchronously, and it is painfully slow. Try to unzip a large
Most real-world sites I've seen are compute-bound and have no real need for concurrent I/O operations... uniprocessor boxes are still good candidates for I/O-intensive stuff (look at ftp.cdrom.com for example). It would've been nice to see the Mindcraft, etc. tests repeated on a single CPU machine, just for comparison's sake.
becoming more efficient by building all of the servs in one place is not even SIMILAR to Compaq cutting alphas. From many of the responses it seems the average reader was misled. I am not complaining about slashdot, you cant pick a perfect title everytime, I am complaining about the average IQ of the readers :)
Geek Posers, hehe
Gee, and I was hoping that it was going somewhere.
hummm... i heard rumors years ago there was some other stuff going on in regard to alpha and it had to do with the release of 64 bit nt (vaporware), competition with intel, and pressure. something like that boys in the back room suggestion mentioned in a related message.
(imo) the alpha has a 'kick-ass' processor and the release of the du 'stuff' sounds promising.
i'm not from tejas, but i'd give them a little room for 'show&tell'. long live the ev6 &/7/8 when they appear!
Alphas will be produced at a different plant.
They are not laying off the people who design the
Alpha processors (they are in MA and CA). These
are people who work on the systems.
When a PeeCee Maker ended up with the Aplha, you just knew they wouldn't know what to do with it. It's floundered ever since Compaq bought DEC. Now it's being shuffled off to some out of the way facilities in TX and CA? Next you'll hear that the technology was bought by some unknown hard drive manufacturer and it will be absorbed into obscurity like the Atari 2600. Or the Alpha will just cease and be abandoned all together. Intel wins. And we'll all have to wait with bated breath for God Almighty Intel to "invent" a 64-bit processor. Of course, it still has lots of bugs in it and won't be released for several years. 21x64 whuzza? Just another page of fantastic technology locked away in the dusty warehouse of the X-files, never to be seen nor hear from again. Sad. So very sad.
Can't speak for the Compaq market as a whole but at the company I work for the mix is about 50% - 60% Tru64 Unix 20% - 25% OpenVMS and the remainder is NT. NT has probably done far more to hurt the reputation of Digital and the alpha than it ever did to help. Many Sun and HP customers shied away from Digital Unix because they believed DEC was too comitted to NT and not serious about Unix in the long term.
Fear not, Compaq is investing heavily in the Alpha side of the business. They consider it the spearhead for their advance into real enterprise territory. Bob Palmer cut alot of funding to make DEC a more attractive purchase but Compaq has reversed that.
:-)
What Compaq has done with the Alpha:
Compaq XP1000 workstation
Alphaserver VS10
Alphaserver DS10
Alphaserver DS20
" ES40 model 1 and model 2
" GS60
" GS140
Coming soon:
Wildfire - highly scalable SMP to 256 processors
New Tandem Himilayas based on the Alpha 21364.
and shame on Taco for spreading anti-alpha FUD
A) None of Compaq's quad processor "enterprise" servers are Cyrix or AMD equipped.
B) For the longest time Compaq had no real enterprise solution. They labelled there big x86 servers enterprise class but that is a joke. Those servers are used almost exclusively as file and print or departmental application servers. I have never had a single customer tell me they were betting their business on one of these machines. Sometimes they get used as front ends for big projects but there is inevitably a monster Unix system or IBM mainframe in the background doing all the real number crunching. The x86 box just puts the 'cute' on it for the end-user.
In the article, it mentions that the alpha server production is being moved to two existing plants. One in CA and the other in TX.
BTW, a server farm of alphas running linux was used to create some of the special effects for Titanic and Los Alamos has a 140 node Beowulf of alphas running linux.
Did you read the story? No where does it say that compaq is giving up on Alphaservers. It states that compaq is consolidating alphaserver assembly to two existing plants in other parts of the country.
They are consolidating all alpha manufacturing in North America. See:0 17
http://www.techweb.com/wire/story/TWB19990517S0
This is likely a cost cutting move now that Pfeiffer is out.
A few important points:
In other words, this is nothing to worry about. Alpha is not going anywhere.
Kenneth C. Schalk ( kenneth.schalk@compaq.com)
Alpha Development Group, CAD & Test Technologies
Compaq is cutting costs. Their stock dropped like a rock when they announced they weren't going to make the expected profits last quarter (from a high near 50 to about 25). These layoffs to be expected since Compaq bought both Digital and Tandem. Part of the "problem" (as seen by stockholders and analysts) was that Compaq didn't downsize Digital and Tandem as quickly as many folks had expected. They are shifting manufacturing to Houston, TX (Compaq) and Fremont, CA (Ex-Tandem) just to cut costs. They can't survive if all they do is fight with Dell and everyone else over the sliver thin margins of the low-end PC market.
Compaq is sticking with the Alpha chip and don't expect them to drop it in the near future. It goes with their plans to establish themselves even more in the high-end server market. They expect to keep Tru64 (Dec Unix), Linux, and NT around so they can take advantage of whatever develops on the high-end side.
I feel sorry for the folks who are losing their jobs out in Salem, but it was just a matter of time once Compaq bought Digital.
- 0.25u 21264 (EV67) @ 800MHz by year-end.
- First 700MHz EV67 parts are expected to spec out at 30 SPECint95/base, 60 SPECfp95/base.
- 0.18u 21264 (EV68) @ 1GHz in 1Q00, spec at 50 int/85 fp.
- 0.18u 21364 (EV7) @ 1GHz by 4Q00-1Q01, expected spec 60 int/100 fp.
- EV8 expected 2002.
- EV9 and EV10 are in development.
That doesn't sound like dropping Alpha to me.Posted by Mike@ABC:
Wow...that's big. If Compaq gives up on the Alphas to concentrate on their core PC business, then it's just ceded a big chunk of the server market to Sun, IBM and the other mega-server makers.
So here's a question: is there any way Linux can take advantage of this possibility by touting Linux alternatives? I'm curious.
Posted by d106ene5:
As for cheap x86 compatible chips, I don't see AMD or Cyrix (or Intel for that matter) playing in the enterprise,
Since you're a Compaq distributor, you may want to look inside some of their systems from time to time. Many of their "enterprise" systems are powered by Intel chips (which by default are x86 compatibles).
In fact, the overwhlming lion's share of compaq enterprise systems sold are x86 based.
Posted by d106ene5:
By the way, Digital was making $14billion a year primarily on its alpha product line before it was bought.
you're confusing revenue with profits.
here's a hint - if DEC had more profits, it wouldn't have been bailed out by compaq.
Not quite anymore, actually. SGI spun off MIPS into its own company and is now building servers based on Intel chips with SGI-style motherboards running Windows NT, if I understand correctly.
MIPS was never spun off, because it was never *absorbed*. SGI purchased MIPS in the early 90s, and only recently divested all but 20% of its stake in the company.
The CPU division responsible for the Rx-thousand series CPU is internal to SGI now.
And, though SGI has announced NT server, it has yet to make any. Most of its sales still have RISC CPUs in them.
The profit margin on Alpha based systems is far greater than anything Compaq ever saw with its intel based products. I don't think they understood what cash cows VMS and Digtal Unix on Alpha were until they saw the numbers after buying the company.
Dgital's poor earnings performance was primarily due to mismanagment and lack of vision not poor alpha sales. Without the Alpha keeping it afloat I have no doubt that Digital would have gone under years ago.
I browse with my threshold at 2 so I can't read my own comments :-)
Compaq has no plans to move the Alpha to the proliant case, the engineering costs and FCC certs are to high when they have well engineered DEC chassis to use. All of the new products are branded "Compaq Alphaserver". There will be some crossovers between the Digital storageworks product and the Compaq classic disk storage. The Compaq Alphaserver ES40 will be the first beneficiary of this with both types of disk shelves available.
I browse with my threshold at 2 so I can't read my own comments :-)
Don't put too much stock in those benchmarks. They don't actually measure scalability or enterprise readiness.
It is well known that among MCSEs that NT doesn't scale well past 3-4 CPUs. You need the far more expensive NTserver enterprise edition if you want to get up to eight on the Alpha and even then the performance scaling is nothing like linear. From the first tests coming out on the 8-way Penguin Computing systems, the 2.2.x kernel scales reasonably well on eight CPUs. The limiting factor on better scalability appears to be an I/O bottleneck after about six processors are utilized. Those tests were run using a database (Oracle?) to create the load, so perhaps with a less I/O intensive load such as rendering, the scalability would be even better.
I browse with my threshold at 2 so I can't read my own comments :-)
SGI spun off MIPS but they still use the MIPS processor in everything but their low end stuff. Sun doesn't own the SPARC plant but noone doubts their commitment to RISC. HP has always been dependent on Intel for CPU fab so its not suprising that they partnered with intel for their next generation CPU. They never had any intention of going with x86 CISC type CPUs. Most of the system Vendors have gotten out of the silicon business because the costs of remaining competitive are to high. Only companies that specialize in microprocessor fab can affor to invest $1billion everytime they want to upgrade their plant. Outsourcing your silicon fab makes economic sense, dumping your enterprise architecture for commodity consumer CPUs makes no sense at all.
If Merced were here now and faster than the Alpha, then maybe Compaq could justify making the switch, but current generation alphas are faster than Merced will be when it arrives.
As for HP's Merced moves that is a huge blunder.
They were telling their customers that in two years when Merced arrives it would be "just as fast" as the PA-RISC systems they use now. Yeah, throw out your infrastructure and buy a whole new architecture with no significant performance gains and fewr applications, that just didn't fly with the customers and HP eventually caught on.
I browse with my threshold at 2 so I can't read my own comments :-)
I work for a Compaq enterprise distributor. Compaq has increased support for and investment in Digital's Alpha product line across the board. The only products that were scuttled were low end x86 products that competed with more successful Compaq PCs. Where the Digital models were more innovative or popular they were rebranded as Compaq products.
Compaq cannot compete in the enterprise against Sun SPARCs, HP PA-RISC, IBM AS400s and RS6000s without Alpha based systems. This is why they have released eight new Alpha system models since they bought Digital and have invested heavily in the new Wildfire architecture.
By the way, Digital was making $14billion a year primarily on its alpha product line before it was bought.
As for cheap x86 compatible chips, I don't see AMD or Cyrix (or Intel for that matter) playing in the enterprise, Sun, IBM, HP and SGI all use Advanced RISC processors.
Just because you use a crummy little x86 CPU doesn't mean everyone should. That would be like Peterbilt getting out of the heavy truck business because more people buy minivans.
I browse with my threshold at 2 so I can't read my own comments :-)
Actually I believe most Alpha's run either DEC/Tru64 Unix or linux.
Certainly the vast majority of Alpha machines that DEC/CPQ sell are sold with DEC/Tru64 licences. The figure I heard (from a very reliable source) was 85%. And that's just refering to workstations. I imagine the percentage for servers is even more heavily biased to Unix.
I use Friend/Foe + mod-point modifiers as a karma/reputation system.
all y'all alarmists should take an interest in the world around you. Compaq seems to have done more with Alpha than Digital ever did.
They're even porting their high performance Fortran compilers & math libraries to AlphaLinux.
IIRC the 466MHz alphaserver is 21264-based but with reduced cache.. The microway 21264 dual processor server runs $15k, but each 500MHz cpu has 2MB cache.
According to the latest info presented to us
at a Tru64 Unix QA dept meeting, that is very
much what we are going to do. As of recently,
we consider Linux to _be_ the entry-level Tru-
64 Unix, it is a high priority to support it
with hardware and software, some even open
source, with an eye toward migrating customers
needing more than Linux being able to move up
to Tru 64. Whether and how we follow through
on that policy remains to be seen, but it's not
for nothing that Compaq employs Jon "maddog" Hall,
president of Linux International.
-- Larry Smith
Did you se they just announced a $3.500 Alpha server 466mhz
What about the workstations? I recall seeing 533-MHz Alpha workstations, for ~$3k, at Enorex, before the newest Alpha, and back when Enorex's web-site was still useful.
As for DEC/Compaq's web-site, I'm seeing primarily ix86-based workstations, and a couple of high-priced, 21264-based workstations. I'd love to go Alpha, if I can still find someone selling cheap 21164-based systems (hell, they'll probably still beat the Pentiums).
-rozzin.
The only real threat to Linux's credibility are the recent benchmarks showing Linux doesn't scale up to the really monstrous machines as well as proprietary Unices or even (dag nabbit!) NT. Frankly, I believe that; we haven't had good access to the big machines. So, besides Sun and SGI, DEC makes the most monstrous machines available. Wouldn't be nice if they got Linux to the point where it beat, say, AIX at least?
NT Alpha may be keeping the chip on life suport right now, but, even though NT is stalled in the server market, Xeons (running various OSen) are continuing to infiltrate at alarming rates. I have no serious feelings about one architecture vs. another, only that the more the better, and that the only way to protect open standards is with heterogenous networks. One way to do this is to revive Alpha, and it won't happen unless Alpha throws in its lot with Open Source. The proprietary Unices are nice, but nothing they can do will make them stand out in a twenty-year-old, saturated market.
If Alpha runs Linux and *BSD really nicely, on the other hand.... They need to do what Intel's doing, giving help to the open OSen, but in a much bigger way. I hope Compaq has the commitment to do this in their current, risk-averse environment.
Alright, nobody tell him about the Campaq thing. :) He works too hard for all of us!
Thank you.
"... I declare our city to be a free and independent state to be named Tri-Insula!" --Fernando Wood, Mayor of NYC 1861
MIPS will survive as an embedded processor.
.13micron fab. Now those would be some sweet watch-sized supercomputers...
SGI is the last of the big MIPS users. Look for them to switch to McKinley (Merced is just for show...).
Too bad HP's getting rid of the PA-RISC. Oh-well...
What I'd like to see are some copper based StrongARMs from a
--Al
I really doubt that NT is what is keeping the Alpha alive. Compaq, & Digital before it, have intel boxes that play at the level where NT sits. If anything, the Alpha has been around this long because they still need it to compete with Sun, HP and IBM in the big computer business.
Must be why they build 16-way alphas with huge I/O bandwidth. The floating point may be wasted, but it is the only chip Digital had that could play in the high end.
Alpha isn't dead. Samsung Electronics in Korea is running it full bore. In fact I just finished doing a show here with Compaq Korea that Highlights Applications for the Alpha family of servers. (I want one----I want one the heck with intels Pentium 2.5 er III) Check out these URL's for more info. http://www.samsungsemi.com/ or http://www.samsung.com/search/cgi-bin/samsung.cgi? lang=eng&page=1&where=1&query=alpha+serv er&slog=mickey&x=24&y=9 (A search result page.) Alpha isn't dead by any means. In fact Compaq is starting to work harder and harder to make Alpha a part of the Linux and FreeBSD communittee's from what I see.
-This space for Rent-
900 at this plant, 12,000 since acquiring Digital
Let me get this straight, your opinion is that the Alpha first began to flounder after Compaq acquired Digital?
I'm not Alpha expert, but I got to study the Alpha design while taking an assembly language course at UNH.
The Alpha is a really nice design. It has 32 integer registers and 32 separate floating-point registers (all 64 bits). 64 registers total. Function arguments are usually passed entirely in registers, greatly reducing function call overhead and stack usage.
It has a well-designed, compact instruction set - this is a very RISC machine. Its memory management is very elegant -- I am told it is a breath of fresh air compared to other architectures. The native CPU bus itself screams, even if the peripheral bus (PCI) is not as fast as, say, SGI's proprietary buses.
All of this means that with a good optimizing compiler, you get *outstanding* real-world performance. Your typical benchmark is designed to run on CPU itty-bitty Intel CPUs, and doesn't take advantage of all the things the Alpha can do. This makes Alpha look less good then it is. In the real world, though, all of this sweet hardware design gives you a significant performance boost. Furthermore, the Alpha tops out much higher then Intel. An Intel box, even SMP, will start to max out when an equivalent Alpha is still running at idle. At UNH, there was a dual 200 MHz CPU Alpha box that routinely had 200+ students on it, doing everything from reading mail to running emacs to compiling projects, and it never starved for CPU - it was always a lack of memory that caused it to bog (it had 348 MB, I think).
The Alpha also scales higher in an absolute sense -- you can get bigger Alpha boxes (16 CPUs, at least) then Intel boxes.
And, of course, 64-bits of address space means you can use VLM (Very Large Memory) techniques to boost performance. Throw enough RAM at a problem, and it usually goes faster.
What does this mean to us? Well, the Alpha is a great workhorse for multiuser systems running lots of jobs. Also databases, compilers, numeric and scientific programs, anything that chews on lots of data.
It does not do quite so well at pushing data off a disk and onto a network wire. It is still pretty darn good, it just is not as amazing as the rest. This has as much to do with PCI and software as anything, I'm sure. If you can afford to throw enough RAM at it, Alpha pulls ahead again by virtue of having everything in cache buffers.
On a related note, I have found Digital UNIX (or OSF/1 or Tru64 or whatever you want to call it) to be one of the better commercial UNIXes I have used.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Visit Compaq's web site at http://www.compaq.com/ and make use of the form on their " Contact Us" page.
Be sure to make the following point to them in the process: Virtually nowhere do they advertise or promote the Alpha line. No wonder it's flagging in the market! I see ads everywhere for Compaq Intel boxes but no mention of their own Alpha processor. If they would only add Alpha systems to their current ad campaigns there would be a 100% increase in ad space for Alpha.
D. Keith Higgs
CWRU. Kelvin Smith Library
My office has been taken over by iPod people.
This may be a somewhat discussion-diverting thought, but is the Alpha really a server-oriented processor at heart in the first place? I always thought that while it's got nice I/O bandwidth, that the Alpha really shines at floating point performance, which isn't a particularly critical benchmark for server operations. So even if Compaq was folding back on the Alpha (which they are not) I don't see that it would have anything to do with their server strategy. Isn't Alpha primarily for high end scientific workstations, 3-D stuff, etc?
What the hell made someone moderate down this article. The guy's absolutely right. I also really wish Cmdrtaco would read the freakin' article before he posts an alarmist headline like this. I respect his coding skillz and all, but I've begun to question his reading comprehension, or if he's not reading the linked articles, his attention span. Open-source news only works when people review the submissions.
Compaq moves the production of the Alpha processors to the factories in Fremont California.
Same place as where they produce the former Tandem NonStop Himalaya servers, because they're going to use Alpha processors in thoose instead of MIPS which is used today.
So, what this really means is, that Compaq are going to make some kickass servers with the best technologi awailable. Imagine one of thoose running Linux...
I knew most of DEC hardware was being scuttled the day they were bought.
Not really shocking - the Alpha never really has made much money for anyone.
The fastest chip does not win. This should be obvious to anyone who follows the industry for any period of time. Cheap and cheerful - this is what people want in chips. Being x86 compatible doesn't hurt either (FX32 doesn't count).
Tyhe story only says 900
No man is an island, But if you take a bunch of dead guys and tie them together, they make a pretty good raft.
I interviewed for a job at Cabletron back in Feb. I was one of two top candidates. Now I'm glad I lost out.
that was close.
Just some food for thought. Everyone who ports their 3D/2D animation software over to NT/Alpha, prefers it over Intel. Since these programs make heavy use of FP, they love the Alpha platform. Even within the limitations of the NT OS.
The company that makes the cartoon/animation software I use (Animation Stand) can't say enough good things about the Alpha. I hope Compaq continues to improve and support the Alpha chip.
The reason that intel boxes will never be as good as Risc chips from SUN SGI etc.. is that Risc chips run eerything better in terms of software written in C. I should know I wrote a compiler for the MIPS processor. If you go and use Intel boxes, things will be fine. The reason that Intel is talking about a Risc 64 architecture is because this needs to happen for their chips to be worth a damn in the future. No one wants code written in assembly anymore, no one needs this code. Without assembly program support Cisc chips are worthless, they have longer execution cycles and fewer registers. this is a bad thing for programs written in high level languages. A 64 bit structure is also helpful if one wants to run software that addresses huge drives, like terabytes and exabytes of RAM. This is not verey iumportant now, but you better watch out before you say when this will be important!
- Kill Yourself, spare us all! -
About a decade ago, Digital laid off about the same number, maybe more, at their Sanders plant in Nashua. It was one of the causes of the 1991-2 recession in New Hampshire and is said to have cost Bush a ton of momentum and helped Clinton get some name recognition. About 20% of my friends' dads were out of work as a result of that downsizing. With Cabletron on the skids as well, doesn't sound like a good time for tech in New Hampshire...
Guys - honestly. As far as I know, Salem was just a machine assembly plant - Comapq probably has 10 of them, and they want to start centralizing. The acutal Alpha chip is still made, by Intel, no less, at a large plant in the suburbs of boston. Check out http://www.unix.digital.com/ sometime. Compaq is betting on the Alpha bigtime - they want to fill the enterprise with ProLiant's and AlphaServers. They are also betting on Tru64 UNIX (read: Was Digital UNIX) big time. Compaq is just trying to save a couple bucks while firing up for the big game.
I've seen their roadmap. They have grand plans for the alpha for at least a few more years. (very cool grand plans, too!!! )