Slashdot Mirror
1GHz Alphas
RelliK writes "news.com has a story about
1GHz Alphas demonstrated at PC Expo. They'll be available by
mid-2000. In the mean time, they'll start shipping 750 MHz
Alphas 21264 in July. " MMMmmm... alphariffic.
← Back to Stories (view on slashdot.org)
Alpha has been making extremely fast chips for years. Look up their old roadmaps and check the clock speeds.
Not that clock speed means a whole lot. However, Alpha's spec numbers have been impressive too.
Saw it today... Nice little booth that Alpha Processor had there.
:)
More interesting IMHO was the dual 750mhz system they had running Linux. 8mb L2 cache ram on those suckers. They didn't seem to have any example of how fast it was, but it was impressive. I'm not sure why they were demoing the 1gig Alpha but didn't go all out and show a dual 1ghz system. Maybe a motherboard issue? Anyone know for sure?
More impressive was some of the serious hardware that Compaq was showing, but the people they had there seemed clueless about which worked with Linux and which didn't.
I was really disappointed at the Linux Pavilion though, seemed like half-baked attempts at doing something both by RedHat and Caldera. Liked what I saw at Cygnus's booth though. Liked that the BRU folks were actually giving away knickknacks, unlike practically everyone else there.
Anyone going to PC Expo -- take a gander at the 64" HDTV at the Panasonic booth. Hmmmm... I want one.
Volia! 100% CPU utilization. Moving the mouse really fast (last time I tried) also would put the CPU to 100% effectively locking everyone out.
Compaq was demoing a (I believe) 8 system dual-processor Beowulf cluster. I didn't get a good look at it, as I was drooling over the Alpha mainframes. :)
:)
I think I spotted one or two other companies demonstrating Beowulf clusters, and I remember spotting POV running on what looked like Linux above a stack of a half-dozen machines in one of the booths (might've been Compaq's) which I assume was showing the clustering too.
Corel... er... HCC... er... rebel.com had the Netwinder RM there, which was cool too... They're not doing a very good job differentiating themselves from Corel, sitting in the middle of Corel's booth.
"lowering both cost and price"
it's an interesting story, but whoever wrote the article is an idiot.
In Capitalist America, bank robs you!
I know someone who's running fluid dynamics simulations on two machines. One alpha (21164) and one dual PII.
Both machines are about a year old. The price tags where somewhere around $70K for the alpha and $3K for the PII.
When running two simulations on the PII and one on the alpha (one sim. per cpu), the PII is marginally slower than the alpha.
Ok, this is just one case. And I haven't seen the code and don't know which compiler options where used. Still, let's assume the code could be tuned to run twice as fast on the alpha, it's still one hell of a price/performance difference !
These results are far away from the SPECint/SPECfp numbers, but they are real world results achieved by real world users. I think that's what counts.
My point is, that although the alphas are really nice machines, and a new fast alpha will be faster than a new fast intel based machine, Compaq will have to get those price tags right.
I'd definitely get an alpha as my next box, if only I could afford it. A dual EV6, mmmmmmm... But there is no way in hell I can ever get that kind of money. I could buy a farm of PIIs for those money instead.
Linus himself is now working on a quad intel machine, instead of the dual alpha he had earlier. He said something like; the alpha is nice, but the intel machine is simply faster.
It would be great to see cheap alphas below 600 MHz, and really expensive ones above. That way us ordinary people could get our dirty hands on affordable alphas, and the companies that need the higher speeds can fund compaq by buying the high end ones. Much like intel is doing today.
... by the end of the year? Does that sound like a threat to the Xeon?
It's finaly (more or less) been confirmed that the 21264 Alphas and the K7s will be interchangeable. Now all we need are mobo's with BIOSes that can support both chips. And no ads in them either please!
--
Aaron Gaudio
"The fool finds ignorance all around him.
"Every man is a mob, a chain gang of idiots." - Jonathan Nolan, Memento Mori
This is an interesting innovation - at those speeds, you can cook your food with the radiated RF energy, the dissipated heat, or both. Finally, a computer that's *really* an appliance!
I can see it now, the new CPSC/FCC/DOE microwave PC warning label:
WARNING - Do not remove this tag under penalty of law!
(This isn't a matress or pillow, we mean it.)
DANGER! Microwave Microprocessor Unit! Do not ever, ever open the case of this computer!
RF Radiation Hazard inside. Opening this computer will let cancer-causing microwave
frequency photons jump out and eat their way through your retinas on their way
to your brain, where they may impair your judgement in selecting an operating system.
- You've been warned.
- Factory sealed for your protection.
- This computer contains no user-servicable parts.
- Like you could get it apart anyway, since Compaq uses these goofy screws...
- Do NOT warm strawberry Pop-Tarts in the Zip disk drive slot.
- Coffee cups on the CD-ROM drive "cup holder" may be heated, but drive
- Digital/Compaq is NOT responsible for funny little fractal patterns on your CD's.
- Discontinue use if rash, irritation, redness, or swelling develops.
- Do not use an apostrophe to indicate the plural form.
Thank you.must be closed when no coffee cup is present, or it's your retinas, baby.
Legal Department, Digital Equipment Division of Compaq Computer Corporation, Houston, Texas.
"The future's good and the present is nothing to sneeze at." - Roblimo's last
Some high-end SPECint/SPECfp results:
PA-RISC 8500 @ 440Mhz 34/51.4
21264 Alpha @ 500Mhz 27.3/57.7
Some notes: AMD's K7, even though it has a better FP unit than P-II will 'only' get SPECfp of about 20 at 600Mhz. (don't have published info, so making guess based on that it's about 30-40% faster than P-II). Sun's next gen chip (UltraSparc-III) will apparantly get SPECint/fp of 35+/60+ at 600Mhz (no actual results yet) - it is supposedly being publicly shown at the DAC (Design Automation Conference) now, but won't ship in volume until end of the year.
Anyone know how much the 8500 costs? It has 1.5Mb of level 1 cache - it has 150M transistors, to the Alpha 21264's 'mere' 15M. It must cost loads... This cache probably skews the SPEC results quite considerably when comparing to 'real world' cases - the SPEC marks scale pretty well with cache size I've heard... I also see that the 8500 doesn't seem to scale at all well at SPECfp as you add extra processors, compared to the other chips.
Real world usage can vary immensly from the SPEC values, depending on what you're doing. I have friends who've compared various machines for high end computations (fluid dynamics) and they found the SGIs ran/scaled the best, even though they didn't have the best SPECfp results for a single chip - it's their massive data buses that do the trick. Actually, the PA-RISC 8500 doesn't have a complete Fortran compiler yet... Most people I know consider the Alpha to be let down by it's IO/bus data-rate,etc. Yes, it's better than PC, but it's not much compared to the other high-end RISC guys, especially SGI, though I expect this difference to change...
A final note, a problem you get with high-speed processors is that they become nice little microwave transmittors (at 700-800+ Mhz, I think it was) and so you really need to reduce the power (the PA-RISC 8500 consumes 85W @ 440Mhz) and up the shielding when clocking at this rate, because otherwise you'd get a REAL pizza cooker/toaster in your computer...
There's some great lines in this article. The authors should in PR or marketing.
...that did not require a special cryogenic equipment to keep it from overheating.
That guy with the mineral oil cooling system is going to be disappointed. Then again, knowing this crowd somebody going to want to overclock a 1 Ghz chip. Just make sure you use special cryogenic equipment, non-special cryogenic equipment won't work.
and puts them into a fungible package.
Okay, raise your hands. Who didn't have to look up fungible in the dictionary.
Compaq, which inherited the Alpha design team when it acquired Digital...
Oh oh, who died? I don't know about you, but if my company was bought out, I would not want to be called an inherited employee.
While universally lauded for its number-crunching prowess, the Alpha chip...
Ooooohhh, prowess. I bet that gets the chicks everytime.
I'm just being silly here today because, My new program...It's Alive! It Walks! It core dumps!
See www.alphalinux.org
--Peter
www.alphalinux.org
Well, that answered half of a longstanding question: The Alpha will be interchangable with the K7 on SlotB motherboards.
But what about heterogeneous multiprocessing on 2+ processor systems? Can I have a K7 AND an Alpha on one dual-slot board? Obviously one would have to enjoy kernel hacking (and probably BIOS hacking), but this has been a fixation of mine ever since I played with Rainbow 100 (z-80 + 8088 in the same box) and Apple II (6502 with a z-80 card). I even upgraded my Kaypro II with a 8088 daughterboard when I was in school.
Imagine the performance gains when you can predictively send ops to a processor with an architecture best suited for the operation. Sick and twisted, I know.
I think not...(*poof*)
No, it's not Xeon that's being threatened by the Alpha; that chip, or at least its architecture, being Pentium-compatable, has a long and successful life ahead of it. However, Intel should well be worried about the Merced. Even if the Merced does keep up with the speed of the Alpha, Alpha systems are bound to be much more stable and less buggy simply because the Alpha's had more than half a decade of real-world use under NT and Unix. The main incentive for going with Intel at the moment is that you stay compatable; if you're going to take the hit of moving to a new processor, that advantage is lost.
cjs
The world's most portable OS: http://www.netbsd.org.
Last week The Register posted a roadmap (leaked by Compaq, apparently) of projected SPECint95 scores for Alpha, Coppermine, and Xeon, and it don't look good for Intel. Merced rates a sorry little x at the end of the chart (turning in a SPECint95 of 46 to the 1Ghz 21264's 60); while rumor has it Willamette (Intel's all new 32-bitter) will clean up, it doesn't make the chart since it's not scheduled to arrive until Q3 next year--and that's without factoring in Intel's rather unimpressive punctuality of late. And while the SPECs weren't posted, let's not forget that Alphas have traditionally humiliated Intel chips on floating point performance. Now, take the K7^H^HAthlon (*yech*! Who came up with that one?)--with its available 2MB L2 cache and 8-way (or was it 16-way?) SMP--apparently beating Xeon performance for a whole lot less on the low end, and the new 21264 Alphas creaming them on the high end...add 64-bit Win2K for the Alpha...and finally the fact that Intel's IA-64 architecture is by all accounts late and underwhelming (at least until McKinley shows up in 2001 or 2002)...and suddenly Intel's server CPU cash cows look like dubious purchases for at least the next 2 years. As for the desktop market...the Athlon (gawd that's an awful name!) has to be the CPU of choice for cheap workstations and expensive gaming boxes; the K6 is already the fastest chip for non-FPU apps. The only reason I can think of to get an Intel CPU would be a Celeron for some cheap gaming--they still overclock a bit, but only if you're willing to run your bus at an unorthodox speed. Meanwhile, Intel's having a hell of a time getting memory companies to make Rambus DRAM that works; until then, Camino, the chipset for their new .18 micron PIII's with 256k full-speed cache, is dead in the water--it's already been pushed back until November. Oh, and nevermind that (according to Tom's Hardware at least) RDRAM's benefit over PC133 looks to be negligable. The longer they spend fitzing around with RDRAM, the longer their stuck with a 100 Mhz bus, while the K7 (so sue me) and Alpha get 200Mhz. All of which has me wondering...in a couple of months, after the K7 becomes cheap enough to get into budget boxes...what possible reason would any computer buyer anywhere have to buy an Intel CPU (at least until Willamette)?? Oh, almost forgot. They make the internet go faster.