Slashdot Mirror
AMD Planning 1GHz CPUs
idan writes "This ZDnet article article indicates that AMD is opening a fab
that will produce 1GHz Athlon CPUs." I'm sure it's pure coincidence that AMD is making this announcement so soon after Intel announced their "real soon now" 1100 MHz "Athlon Killer". Do we get to call this one the "Athlon Killer Killer"?
An interesting point brought out in one my courses here at school... When analyzing algorythms you're not supposed to "consider" the machine it will be running on. If you think about it, it makes sense.
A really fast machine running an n^2 algorythm will still be running an n^2 algorythm. When you reach a certain (often relatively small) set of data, you will SERIOUSLY notice how laggy the system is performing. Doubling the speed of the processor doesn't mean that you can double the amount of data you give to an n^2 sort before the time it takes is greater than before. It's a mere fraction of that.
People will continue to program the way they're used to. People will program in a manner to scale, if they need to scale! If they don't need to, then they won't (why bother wasting the extra time?). I don't know about most programmers, but for me programming is an ego trip. My goal is to get the slickest smallest fastest most bug free piece of code out there. Now, I realize that there are many coders out there that don't think like that but the thing is they'd code using the n^2 algorythm anyway.
...and another diversion: in some cases an n^3 algorythm will outperform an n^2 algorythm (other examples can be made); you also have to consider the data set you use. If KNOW you're going to use a data set smaller than "1" (one is a relative term, where the two functions intersect; it may be a rather large number if one algorythm is measured in minutes and the other seconds) then then n^3 would be the better choice.
I seriously doubt this will change the way code is written...
(my apologies to The Register staff for 'deep linking'...)
--
E2 IN2 IE?
I'm sure if you asked Intel, they would tell you they'd be perfectly happy if AMD would stop making faster chips.
Or did you mean to write AMD?
Trees can't go dancing
So do them a big favor
Pretend dancing stinks!
as we need faster throughput. Face it, chipsets aren't as interesting to consumers. As long as consumers buy machines based on procesor speed, manufacturers will continue to sell junk with fast processors. Just how useful is a 1 Ghz chip when the memory runs at 100 mhz and the system is using an IDE disk controller.
some really bad bugs(by bad i mean something like the F00F bug)
The F00F bug only really was at all serious on a server which is running untrusted code. While this is certainly a bug, it's not one that will affect a large quantity of the people using this chip. Furthermore once someone smart enough thought about it, it became easy to fix/work-around in software. The Pentium's FDIV bug is probably a better example of a bad bug.
Trees can't go dancing
So do them a big favor
Pretend dancing stinks!
Have Intel been giving us their best - hardly likely - until now, no-one's pushed them, so they've been able to deliver lower cost, lower performing processors as "state-of-the-art", simply because no-one could show us, the consumers, otherwise. Now that AMD have finally "made it", they're being forced to bring out the more advanced processors - IA-32 (Williamette) launch pushed forward 9 months as an example.
AMD - without a doubt - they've been struggling to keep in the market, as a smaller, less well financed and therefor less well financed company, they've had no choice but to put forward their best just to keep up with the market - AMD chips have until recently been renowned for not being very overclockable - because they ran close to their maximum performance levels already. With the release of the Athlon, they've finally caught, and surpassed Intel - and now Intel are being forced to change tactics in order to compete with the people who could quite possibly steal the "PC clone performance kings" title from them. Much kudos to AMD for what they've managed to achieve.
As an aside, I opened up an OLD PC the other day (I can't remember what it was, but I do know it was OLD) and a large number of the chips were labelled with Intel AND AMD logos - together, on the same chips - kind of hard to comprehend considering the way Intel and AMD are now so fiercely competitive....
Didn't you just answer your own question? Re-read those two lines.
Also, how would the advent of 1GHz CPUs allow code bloat to increase at a quicker rate than it already is increasing?
-A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Good to see the old pattern is still working like a champ:
Announcement: New faster/better/bigger XYZ!
Response 1: Do we really need faster/better/bigger? Most consumers don't need it! And I'm the first person ever to ask such a profound, socially-conscious question!
Response 2: I bet that makes a lot of heat. Hyuk, hyuk!
Response 3: Huh, I bet that would make a kickass Beowulf cluster.
Repeat responses 1, 2, and 3 as necessary. Presto, automatic Slashdot conversation generator.
MJP
Don't try that "protecting the children" shit you people use to keep the tits and bad words off my TV. --Seanbaby
A micron is a 1,000th of a meter
Last I checked we called that a millimeter. Can you even image a chip done in 0.18mm?
--
FatSean mentioned that neither Gateway nor Dell sold Athlon based systems.
:(
...
... ;)
I work indirectly (through an agency) for the one of these which rhymes with Hell, and that's true -- none to be expected in the next few months, either.
Below is the text of the letter I Sent to Michael Dell a little while ago; I believe it eventually sent, but I was amused to see that it was first returned by the mailer as having "permanent fatal errors"
Subject:
processor diversity vs. Intel dependence
Date:
October 12, 1999 5:44:40 PM EDT
To:
michael@dell.com
Dear Michael:
First of all, I own a (piddling) amount of Dell stock, but none in AMD, though that might soon change. I also work for an ad agency which does a lot of Dell work. [note: deleted the name of agency. tl]
Now: As far as I know, Dell uses Intel chips in every computer it builds. If that is not true, then the rest of this message is based on false premises and you can stop reading.
However, if Dell really uses no processors other than Intel, I think the company is worth less to me (and you) than it would if it also built systems with AMD chips, or even Cyrix chips.
Dell was screwed as much as anyone with the sudden *un*release of the anticipated 820 / Camino chipset; that fact alone should be enough evidence that being in bed with a sole provider is chancey. In the case of some other PC makers, though, some of their higher-end systems would be unaffected, because they are based on the AMD Athlon.
Dell finally preloads Linux (thank you!) at least on some systems, and even with a premium. You wouldn't stick with a single hard drive manufacturer or memory supplier, so why do it with the driving point of your systems, the
CPU?
Cordially,
Timothy Lord
No response, so far
timothy
jrnl: http://tinyurl.com/c2l8yr / foes: http://tinyurl.com/ckjno5
Reminds me of the BBS quote wars (remember?)
> this
> > is
> > > getting
> > > > out
> > > > > of
> > > > > > hand
> > > > > > > !
:-)
--
Who is it that falls for these "announcements." Of course AMD is working on a 1GHz chip, that's what chip manufacturers do.
I can understand, a little anyway, why it makes sense for software manufacturers to promote vapor ware. After all, they are trying to keep you from buying into their competitors completely incompatible system. AMD and Intel, on the other hand are making products that are essentially drop in replacements for each other.
Does the average consumer care where his wheat was grown? Heck no. Soon they won't care who made their processor either. It will all be about speed and price.
I'd like to write some cool TV ad's for AMD. maybe some lame ass dudes in bunny suits driving some weird, silly auto being blown by some hot sports car with a guy looking back at the bunny suits, shrugs and thinks "Freaks..." Seriously, I'd be REALLY nice to see arrogant Intel put in their place for once. AMD has worked hard and deserves the recognition from both the media, HW manufacturers (HELLO, ASUS??) and ultimately, the consumers. GO AMD!
Everyone already knew this. It's been known that
the Athlon scales *VERY* beautifully, and that once the Dresden Fab30 comes online, they'll easily ramp up to 1Ghz.. it's why Intel is scrambling. Kryotech has been saying they'll sell thier SuperG 1Ghz (cooled) Athlon by december, and Fab30 is due to be online by the first week in 2000. So where's the news?
Now, alls AMD needs to do is make a better CHIPSET (or VIA, whichever comes first), one that supports SMP and more than 512k cache. I'm a supercomputer/scientific researcher. And i write tight code no matter how fast the processor goes.
What do I see in my future? Clusters of Athlons and Alpha's for now, and multi-threaded hardware beyond that. Funny, unless EPIC really surprises me, Intel is nowhere in my future... hmmm..
--ps, i still think AMD should buy the Alpha and it's designers.
PianoMan8
- --
"I Hate Quotes" -- Samuel L. Clemens
I'm always a little wary of this sort of stuff. Faster processors *do not* always mean greater productivity or even speed. Coders will become even more slack as a result of this, Im afraid.
:)
There is really something to be said for developing on a slow machine and spending a fair amount of time *optimising*... I hope the Linux kernal hackers dont get caught up in this and start bloating the kernal (ack!).
Anyone who has seen recent (1998-1999) Commodore 64 demo will know what I'm saying. You wouldn't believe what they do with a 1Mhz processor these days, and its all due to *optimising*...
Now, running well optimised code on a 1Ghz processor, well, thats something different
Simon
The real linux_penguin has Slashdot ID 101961. Anyone else is an impostor. Including Bruce Perens.
Has anyone noticed that when you use programs that let you create programs or web pages from templates such as front page (shudder) the code created is mainly filled with useless babble. Because computers are becoming bigger and faster Software can be written really sloppy and people don't notice because the machines are now fast enough so it dosn't make much difference. It would be interseting to see how much programming code has deteriated since the days when people had to hack away at pragrams at MIT in the 60's.
I have also heard from informed sources that they have had *very* good results with the preliminary Athlon batches from the Dresden fab. One of the later reports claimed that AMD engineers had sucessfully run multiple CPUs in the 900Mhz range with nothing more than large heatsink and fan cooling units from "rough" pre-production wafers.
;)
These comments are about a month and a half old and I've heard no more specifics. If this is an indication of production quantity yield, then I think AMD will finally get some of the rewards they have worked for so long and hard.
p.s. For the folks that whine that people don't really need a 1Ghz cpu, I still have a fully functional 486-33 with 4megs ram and a 200meg HD that was quite fast when it was new. I'll trade it to you for that P-III 500 that is way too fast for your applications. Then you can run your apps at a more relaxed pace. Watch the apps grow in size and complexity for a couple of cpu generations and you won't be confused about the need or desire for faster systems...
Kindness is the language which the deaf can hear and the blind can see. - Mark Twain
I've got this dual Celeron 550 compressing MJPA video with both processors maxing out the local buss. With 1100 MIPS they'll do only 15 frames/sec of 640x480. The problem isn't the MIPS but this 100Mhz buss. We need at least a 300Mhz bus before CPUs become useful. Anyone else benchmarking video capture on their Athlons?
Even The Register who first reported the 1.1GHz Willamette noted it is a rumour, and likely to be untrue. Besides, The Register had the 1GHz Athlon story before the Willamette, so there is no way it would have been sparked by the Willamette "announcement".
Even if the Willamette thing was true, I'm not sure it would be good idea. Shipping a CPU nine months early probably means that it has not been tested very througly, and will therefore contain a lot bugs. No CPU is 100% bug free, but insufficient testing could mean that even some really bad bugs(by bad i mean something like the F00F bug) might slip trough.
Sun's SPARC provides nice evidence of this; they are selling lots of systems for high end database and web applications not because the SPARC architecture is vastly superior to its competitors, but because the rest of the system is fast.
On a PC, the real "critical component" is the motherboard, as that tends to be a determinant of such things as:
If you're not part of the solution, you're part of the precipitate.
What do I see in my future? Clusters of Athlons and Alpha's for now, and multi-threaded hardware beyond that. Funny, unless EPIC really surprises me, Intel is nowhere in my future...
Not so fast there - put 2 + 2 together. Multi-threaded hardware, right? That means SMP on a chip, right? That means: transistors/mip matters. Well, as far as I know, the crown for best transistors/mip rating in the business goes to ARM - guess what Intel is heavily involved in?
Life's a bitch but somebody's gotta do it.
It has been observed that exponential-time algorithms (requiring, say 2^n steps on input of size n) can be executed in linear time by the following procedure:
Wait 1.5*n years, while the speed of computers increases by 2^n. Execute the algorithm in unit time.
In a lot of ways, it is like the time when Japanese cars quickly replaced those produced by complacent American car companies. Maybe now, the processor market will see a jump in quality and a dip in price.
Then again, its just as likely that we will just see a rise in quality along with a corresponding rise in price. But hey, I guess supply and DEMAND is part of capitalism too.
14 digits of Pi are all we need.
More megahertz is more polygons, pure and simple. Floating point speed is the key limitation to what we can do in games these days, and limits all aspects -- from doing transforms for polys to nifty effects like lighting.
:)
The real point of offloading generic tasks like geometry and lighting onto the graphics board is to be able to do crazy, highly specific features on the CPU. Games will only get wilder and wilder. Great things are to come!
...
Oh, you wanted to get work done? I can give you a 386 for free that will run LaTeX, lynx, mutt, gcc/g++, gdb, and everything else you need to be productive
Does the world really need 1Ghz processors?
;)
Well, i know *i'd* like one, but in reality, what does the average user need with 1 GHz (atleast at the moment anyway)?
While i am sure game designers are leaping for joy at the excessively fast calculations these processors wil do (and the resulting impact on gaming), we as yet don't *need* that power. Afterall, Some of the best looking games coming out will run nicely on 300Mhz
Obviously certain roles are perfect for these processors.....servers spring to mind....but for the average home user, it really isn't needed (once again, however, i wouldn't complain to owning one
The rumour mill is feeding this post, so take it as you will.
.18um and copper interconnects right now). They've been producing samples of K6-2/K6-3's using a .18um process using copper interconnects for the last few months.
.25um process (I find it remarkable that they were able to get to 700 air cooled).
.25um chips; it is VERY reasonable to expect AMD to be able to produce 1ghz chips soon after bringing Dresden online.
AMD has been working on their Germany plant for quite some time (last couple of years). From the moment I've heard of it, it was always AMD's goal to produce chips in huge quantities using state of the art technology (being
Within the last few weeks, rumour had it that they had been producing sample K7's the Dresden plant and sent stuff back to Austin for "verification" (ie: look over each nanometer [or whatever they do] to make sure everything is good).
To me, this article seems to indicate that everything is looking good in the verification process, and they're confident enough to start ramping up to full production (or begin preparations to ramp up).
Word is soon after the 733mhz cuMine process is released AMD will drop prices (which I think they just did actually...) and release a 750 mhz version. This, incidentally, is still on the
Kryotech has systems running at 900mhz using current
My 2c.
If you build it, they will use it.
In reality I bet a lot of us have Celerons or PII, possibly PIII processors. You non-Intel users don't feel left out, you know what I mean, right? Anyway, I don't really *need* a Celeron, but it sure runs a lot nicer than my P200, ya know?
In short, in order for progress to be made, you have to progress.
Eagles may soar, but weasels don't get sucked into jet engines.
Celeron is a perfect example of a company taking there time, optimizing design and fab, and turning out a cooler running/tighter chip(read:over clockers delight). I just get the feeling that the GHz milestone is going to be so tempting that some companies will be rushing them out the door before the work is done.
The last time a company shipped there chips to early, we ended up with a 5 volt 60MHz Pentium that wasn't pin compatible with any other Pentium that could actualy do math. In light of these mis-steps, I hope AMD and Intel have there eye on quality first.
When it comes to new technology, the early bird gets the worm, but the second mouse gets the cheese.
Actually a micron is one millionth of a meter, or one thousandth of a millimeter, being a contraction of micrometer, where micro is 1 * 10^-6. (that was one times ten to the minus six).
You are not me, therefore you are not important