“To the outside observer, improvements in PC architecture are evolutionary but logical. Processors advance inevitably in speed and performance, in happy accordance with Moore’s Law. For Nebojsa Novakovic, a consultant in high-end computing systems, that’s hardly the case. The demise of the DEC Alpha processor is a case in point. A performance leader was killed off by corporate whim.”
This will be a series I’ll be following.
The story I always heard was that Alpha was killed off because Intell sold good crack..aka offering to license a new and improved design to HP (Itanium?).
The story I always heard was that Alpha was killed off because Intell sold good crack..
Not really true. When Compaq bought DEC in 1998 they didn’t really know what to do with the Alpha (or with any of DEC, really). And when HP merged with Compaq in 2001 and thus inherited the Alpha, the Itanium alliance had been going for years already, so it was surplus to requirements.
At the time Alpha came out I was in Lake Tahoe watching TV. This comcerial came on of random things fly across the screen. It ending with a cave full of bats flying past the camera at really high speed screaming. And ended with the words “DEC Alpha, like a bat out of hell”. It was one of the craziest comcerials I have ever seen
I was excited when Alpha came out, and sad when it died.
I was a contract employee for Intel and was there the night DEC Fab 5 became Intel Fab 17. As part of the FTC agreement, Alpha was to manufactured for seven years. The failure of Alpha was not in its technology; it was the ability of DEC to manufacture it in volume at a profit. You cannot look at the history of DEC without taking into account the history of the Alpha.
Its a shame really; for its time, indeed even today, the Alpha was/is an amazing product.
Intel uses it in their chipsets. So in that sense, HyperTransport has proven to be a very good board-level interconnect standard…
Doesnt he mean Nvidia uses it in Intel chipsets? AFAIK Nvidia have been using HT since they started producing mainboard chipsets (the first Xbox and the first Nforce chipsets for AMD socket A), but this is the first Ive heard of Intel actually using it.
Edited 2006-01-24 01:01
The person interviewed must have buggered that up, because as far as I know, Intel skipped Hypertransport when they dreamed up the idea of the one connection to allow Xeon <-> Itanium swapping on the same motherboard, they came up with a competitor to called CSI – going by Intel’s early hype, it was meant to blow Hypertransport out of the water.
The sad part, however, Intel has pushed CSI to the back in favour of the old crusty FSB architecture – this, unfortunately will end up damaging their sales in the server market, where there are more knowledgable customers, AMD has a good competitor, and can delivery in volume such a product to its customers.
Ideally what AMD need are another 2 fabs ontop of what they already have, as they as struggling to keep up with demand for their consumer grade CPU’s.
I am very sad to see that these two architectures are almost dead. SGI and HP have the whole responsibility. It is a same that x86/power is the only winners-players. Capitalism is so DULL, no innovation. Hopefully BLX is gonna have some cool MIPS-like chips. I can’t wait. And Alpha is open, I cannot understand why nobody cares to make a new version. The power of Linux/BSD is that you have a standard base of apps that can be ported to both platforms. HyperTransport is an improved SysAD bus. I have a question to all in the thread since I am not an expert. If someone uses UBOOT or LinuxBIOS can the H/Transport on nforce chipsets be used with PMCSierra’s Hypertransport MIPS64 ??? Is it so difficult to adapt it? I cannot come to a conclusion. What is your opinion? Hopefully Chinese make also desktop Alphas and MIPS64. European and American market are so old and lazy. By the way I feel very good that people like BBRV try to pinpoint the necessity of alternatives. Even ARM is under crisis to the desktop.
No innovation? Now we’ve got x86 which are now as powerful as the Alpha were at a fraction of the price, that’s an innovation.
I’m not sure that Alpha is opened, SPARC is, Alpha I don’t think so.
> I cannot understand why nobody cares to make a new version
Well you’ve stated the reason yourself above: capitalism! Sure the Alpha ISA is beautiful but if you’ve made a new version of the Alpha, you’ll sell so few of them that the performance/price ratio wouldn’t be good..
Besides it would also compete with IBM’s POWER which has much more momentum..
Why would Chinese or anyone would make desktop with Alphas or MIPS??
If you look at the history it’s pretty clear that x86-64 will rule the desktop and probably also the server in the future, likely forever, probably with some modification though (for multicore or perhaps more radically for coprocessors like in the Cell).
@renon
I almost agree with you. Sorry for my mistake, I thought that Alpha is open source. By the way BLX is trying to make a MIPS-LIKE chip which would differentiate them in the market. So , chinese DO WANT to make a desktop with MIPS64. This is a benefit. Hopefully we see a sparc from some other vendor since it is open. I just believ that many different ideas should find their way to the desktop. How would you feel if there were only one brand for potato chips at the super-market? Actuually I consume non-mainstream brands because of quality and less salt. And I think that we need alpha/mips/parisc/sparc/arm besides x86 / x86_64.
I think that many other companies should live and innovate, not only Intel/AMD. In my PERSONAL view, someone chooses a job because he likes it and he wants to offer something new with his/her personal signature. Money is a secondary motiv. So, x86 only business will lead other to extinct and will fuel with greed the investors of the big companies. So, we need other processors. I use x86 every day and I am saving for my (although slower) ODW. Hopefully we will see a low cost SparcStation and possibly a MIPS-Station. I cannot understand why PMC-Sierra keeps the good stuff for itself.