Intel/AMD Battle Rages On

An anonymous reader writes "The battle between Intel and AMD has broken out of the cleanroom and literally into public view with AMD's public display CPU speed challenge to competitor Intel. Should the competition take place, the infamous chip makers will battle their best 2-way and 4-way configurations for the latest title as speed king." From the article: "AMD's proposed dual-core duel would be a live, public performance evaluation between server platforms based on the dual-core Opteron 800 Series or 200 Series processors and the corresponding Intel product. Should Intel accept AMD's challenge, the duel would take place at a public venue to be announced in the coming weeks, with testing conducted by a neutral, third-party testing lab. "

Comments

What software? What terms?

[fredistheking]

This is just marketing by AMD. There is no way that Intel and AMD would come to agreement on the benchmarking software to use. Both companies know their weak points and their strong points. Neither company is going to agree to lose.

Re:What software? What terms?

[PsychicX]

Although, I read a lot of Opteron benchmarks over the past couple months, and those suckers are blazing even at encoding tasks and all. The multiple cores humiliate Intel's HyperThreading. I don't think Intel makes any chips that can compete with Itanium...and we all know how much Itanium matters these days. *cough*

Re:What software? What terms?

[ciroknight]

It's marketing that will work. Intel won't go through with the bout, and AMD will hold it over their heads for the next five years, and everyone on slashdot will troll about it.

Truthfully, AMD could do it, even without Intel's permission. Just go grab a chip off the shelf and let loose.

Lastly, parent's completely correct. There's no way they could settle on what software to use. Intel would argue Linux is made mostly by people with AMD hardware, whereas AMD will argue that Windows has been tailored to Intel for 10 years. Intel will argue that their compiler produces accurate x86 code, AMD will argue it's inconsistancies.

The only way I could see it happening is if they ran every single possible configuration of software and averaged the results, but I'm sure someone will point out some flaw in that even.

Re:What software? What terms?

[_DangerousDwarf]

Perhaps more interesting then software, which "neutral third-party testing lab"?

Perhaps even more interesting, what is a "neutral third-party testing lab".

Most of these neutral labs are only as neutral as the people paying them are....

Re:What software? What terms?

[InvalidError]

This is supposed to be a multicore, multiprocessor and presumably multi-tasking/threaded benchmark. In this case, the Xeons will choke on the FSB and the Opterons should win most benchmarks hands-down.

After you strip games and media-encoding benchmarks, there are not many Intel-biased benchmarks left.

Re:What software? What terms?

[drudd]

Of course you know AMD has already done just that.

You don't actually think they'd challenge Intel to a contest they would lose, do you?

Doug

Re:What software? What terms?

[Dark_MadMax666]

"Strip Games"? Since when intel has any good performance in games?

Re:What software? What terms?

[whitehatlurker]

what is a "neutral third-party testing lab".

I think they mean "electrically neutral". It certain can't mean "unbiased". "Third party" is the ordinal of the shindig at which the testing lab is finally fully purchased. (Until the fourth one, I guess ...)

Re:What software? What terms?

[Michael Woodhams]

Truthfully, AMD could do it, even without Intel's permission. Just go grab a chip off the shelf and let loose.

But the credibility of the results would suffer. If Intel are producing the box, with their reputation on the line, you know they'll have the best possible motherboard, memory etc. for the purpose. If AMD built the Intel box, you don't have this confidence.

Re:What software? What terms?

[jejones]

Intel will argue that their compiler produces accurate x86 code, AMD will argue its inconsistancies.
AMD would love to have another reason to point out the way the Intel C compiler libraries test for the presence of certain features in such a way as to never detect them on AMD chips even if they're present.

Re:What software? What terms?

[Anonymous Coward]

lol what?

Re:What software? What terms?

[mapmaker]

This is just marketing by AMD.

Yes, it's very good marketing. The point of AMD's dual core challenge (which everyone here seems to be missing) is that Intel has no dual core server processors with which to compete. In the server space AMD has dual core Opterons and Intel has...nothing. The only dual core Intel processor is Smithfield, which by their own admission was a slapped-together rush job that isn't good enough for the server space.

It's like pulling a Ferrari alongside a pedestrian and saying "let's race". One side doesn't have anything to race with.

Re:What software? What terms?

[beelsebob]

You obviously don't get it -- AMD *know* this contest will never happen, because the conclusion could be called either way, no matter what the results. AMD *know* that Intel will not accept because they know no matter how it turns out it'll get spun off the planet. AMD *know* that this will be incredibly good publicity. It's as simple as that.

I'm not supporting AMD or Intel here, but I do recognise that all this is is a publicity stunt.

Re:What software? What terms?

[Anonymous Coward]

I disagree about AMD complaining about Windows, I think a high end opteron would spank any high end x86 intel chip in the server category and possibly others. Windows or linux would be irrelevant.

What I think is going on is this is AMDs way to steal Intel's thunder. I think they know that when the new gen of chips come, they're going to have some real competition so they need to milk the remaining window of opportunity for all its worth now, while they still have chips that are clearly superior.

Re:What software? What terms?

[moviepig.com]

Truthfully, AMD could do it, even without Intel's permission. Just go grab a chip off the shelf and let loose.

And conversely, Intel could do it without AMD's permission.

Moreover, if enough public attention manifests, that's exactly what will happen... with each side using the tests and conditions needed to assure its victory, and then publishing the results.

(While we wait for that entertaining debacle, let's prepare by taking the Pepsi challenge, America...)

Re:What software? What terms?

[Anonymous Coward]

You know those games, where you get this little video "Runs Great on Intel Pentium 4" when you start them? For some reason, these happen to be the games where you are likely to get a better performance with an Intel processor.

Re:What software? What terms?

[Holi]

Well since intel does not have a comparable dual core server chip I think amd knew exactly what they were doing.

Re:What software? What terms?

[mjh49746]

I'm already preparing to start trolling once AMD has been declared the winner. How else do you get mod points around here? :-)

Re:What software? What terms?

[Anonymous Coward]

The funny thing is, on those games, the FX's still beat the living shit out of the Intels.

Really, it's like me putting "Enhances DVD Playback!" on the DVD cases I'm selling on ebay right now. Sure, the DVDs _may_ enhance DVD playback, but that doesn't mean some other case doesn't do a better job.

Re:What software? What terms?

[superpulpsicle]

OMG is this the wave of the future!? Reliving old commercials again. This came straight out of the Pepsi challenge.

They are going to have 2 fancy Alienware case, and ask kids which Soda/Processor do they prefer.

Re:What software? What terms?

[anon*127.0.0.1]

Even if Intel wins all the benchmarks, AMD can still argue that they only lost by x% and their chips are much cheaper, therefore a much better deal.

Intel can't agree to the contest, because that'll be acknowledging that AMD is competitive with them. If they acknowledge that, then consumers will start considering AMD when they make their purchase decisions.

The only way Intel could come out ahead on something like this would be for them to absolutely trounce AMD on the benchmarks, or for AMD to suffer a hardware failure during the tests.

Re:What software? What terms?

[InvalidError]

The P4 vs A64 gaming benchmarks were not always as nearly one-sided as they have become over the last year.

I do not pay much attention to benchmarks since I have no plan to upgrade or get new PCs for the next 2-3 years... unless what I have now fails beyond practical repair or runs out of expansion options.

Re:What software? What terms?

[Anonymous Coward]

You troll in favour of microsoft and apple, how did you think? You really think pro-linux or pro-amd trolling gets you mod points? No wonder your karma sucks.

Re:What software? What terms?

[Anonymous Coward]

There's more info available about duel. Who knows, maybe online petitions change the world ;)

Re:What software? What terms?

[Oestergaard]

You have a point, and maybe you're right - but don't be so sure.

It would be difficult to pull of a somewhat real-world alike test where a Xeon would beat an Opteron.

Anyone could pull off a synthetic benchmark that would prove Xeon to be the faster CPU of course, but I'm pretty sure you will find it difficult to take off-the-shelf server software and make it run faster on a properly configured Xeon than on a properly configured Opteron.

Re:What software? What terms?

[Oestergaard]

Exactly.

Which is also why it is so much better to have Intel *not* compete than it is to have AMD provide an Intel box for testing.

Re:What software? What terms?

[Anonymous Coward]

I suppose since it has been referred to as a duel, it would be fair in my eyes to say that, like in olden days, if someone issues a challenge to duel, he knows that the person being challenged gets to choose the weapons.

In other words, don't challenge unless you can win in any of the possible circumstances.

Re:What software? What terms?

[edmac3]

It's "*knows*."

Re:What software? What terms?

[bogado]

AMD would love to have another reason to point out the way the Intel C compiler libraries test for the presence of certain features in such a way as to never detect them on AMD chips even if they're present.

Is this true? Is there a pointer to a more specific information on that?

Re:What software? What terms?

[Drooling Iguana]

Doesn't Intel have significantly more marketing muscle, and thus subsequently more ability to spin an ambiguous result in its favour, than AMD?

Re:What software? What terms?

[jejones]

Sorry... here is the /. article on the issue, and details can be found here.

In brief: a library routine doesn't bother to check for the SSE support if CPUID doesn't return "Genuine Intel" as the vendor string, even though one can detect SSE/SSE2 support independent of the vendor string.

Re:What software? What terms?

[Lord Kano]

If AMD were to go to a big Intel based manufacturer, like Dell and purchase their top of the line and have a neutral 3rd party conduct the testing Intel has nothing to complain about. They're not going to claim that Dell doesn't know how to properly configure a machine.

LK

AMD has a score to settle

[PsychicX]

Looks like AMD is going to really Stick It To The Man for this one. K8 has done a lot for them on all fronts, and thankfully they're not squandering what they've gained from it. If only they'd get a marketing department that wasn't completely incompetent. (When's the last time you saw an AMD ad on TV, hmm?)

I wonder though, it's interesting that this happens the same day that Intel announces the first details about their new line up. It's like they crash into each other every so often and both fire volleys of whatever they can get.

Re:AMD has a score to settle

[Anonymous Coward]

(When's the last time you saw an AMD ad on TV, hmm?)

Yeah, because I'd rather subsidize advertising over research when I buy a processor....

Re:AMD has a score to settle

[Aranth Brainfire]

'Yeah, because I'd rather subsidize advertising over research when I buy a processor....'

Yeah, I really wouldn't want the cash from those increased sales to go into the mix.

Re:AMD has a score to settle

[OK PC]

AMD don't have an incompetent marketing department they just choose not to advertise much and save the money. And lets be honest, anyone who's going to buy a CPU will know enough about computers to know of AMD.

Re:AMD has a score to settle

[ciroknight]

No, I'd say AMD's marketing department's quite good.

AMD Marketing Guy: Dudes, I just came up with the easiest way yet to get our product in the light of everyone, everywhere, and it's not gonna cost a dime.
AMD Dudes: We're listening.

AMD Marketing Guy: Let's use the media. Throw a few lawsuits at Intel that we know won't stand a chance. Start issuing public challenges to the company, even though we know they won't listen to us. It'll work because the geeks everywhere will think it's wrong of Intel not to accept, or that they knew we were superior and that they didn't want to show weakness.

AMD's ticket out of their niche, geek product is simply to look like the Little Orphan Annie being oppressed by the big, evil Intel.

AMD's got the better product. Why don't they simply market the fuck out of it, and stop acting like bitches about their product? I just hate it because everyone knows their product is better, and instead of caring, they're only interested in beating Intel.

I want a company who's willing to sell me a processor, not a company who's all about the blood sport. That's really the reason I've been buying Apple for so long; they don't give a fuck, they just want to sell their products, and make their customers happy.

Re:AMD has a score to settle

[WilliamSChips]

Yeah, I really wouldn't want the cash from those increased sales to go into the mix.

That's assuming that there would be increased sales.

Re:AMD has a score to settle

[Anonymous Coward]

I want a company who's willing to sell me a processor, not a company who's all about the blood sport. That's really the reason I've been buying Apple for so long; they don't give a fuck, they just want to sell their products, and make their customers happy.

At a certain level, everything is merely sport.

Re:AMD has a score to settle

[Skye16]

Throw a few lawsuits at Intel that we know won't stand a chance.

Seemed to work in Japan. Are you so sure it won't work here?

Re:AMD has a score to settle

[admactanium]

When's the last time you saw an AMD ad on TV, hmm?

from my recollection you're not going to see one. i worked on their advertising account briefly a couple of years ago as a creative director (i think we even suggested this public showdown idea). they've been concentrating on advertising in magazines and towards oem's mostly.

the customers they have now, they're likely to keep and increased ad spending isn't going to get them enough individual customers to make a difference to their bottom line. also, the got burned a few years back by another agency who convinced them to do some very expensive commercials with a well-known french director. the spots were quite interesting but i'm sure it had no effect whatsoever on their sales.

Re:AMD has a score to settle

[gitreel]

"(When's the last time you saw an AMD ad on TV, hmm?)" The Tour De France

Re:AMD has a score to settle

[mjh49746]

How do you think competition in business is supposed to work? Are all the competing parties supposed to play 'nicey nice' and pat each other on the back after a long day's work? Bullshit! If you and I were running competing businesses, I'm going to try and run you out of business with a better product, better pricing, better marketing, whatever it takes to do it. Naturally, I'm expecting you to work for that same goal. This is what inspires better products and cheaper prices. Not a duopoly between friends. And, for the record, you're completely delusional to think that Apple's not thinking the same thing and that they really don't give a fuck. Soon, all Apple will be is another Dell in a 'think different' package seeing that they've decided to drink the Intel Kool-Aid, and then try to go toe-to-toe with M$. (possibly the last time) And Apple's got their own foaming zealots caught in their own reality distortion field. That's why I won't have anything to do with them.

Re:AMD has a score to settle

[thriemus]

Obviously you have not kept up with the Intel vs AMD saga.

Intel started this when they approaced OEM's with a "How about you only sell Intel chips and we will give you MASSIVE discounts?" offer. This was a dirty and quite microsofty move to make. AMD is replying with a diginfied response with a fair challenge to prove who has got the best offerings on the market at the moment (and the near future IMHO)

What processor you running out of interest?

Re:AMD has a score to settle

[Anonymous Coward]

I do believe he said he was running with a PPC processor, which makes him neutral.

Re:AMD has a score to settle

[servognome]

AMD don't have an incompetent marketing department they just choose not to advertise much and save the money.

AMD spends more (~17%) on Marketing/General/Adminstrative as a % of revenue than Intel (~14%).

Re:AMD has a score to settle

[diskis]

And which company haves the higher revenue?

14% of a lot is more than 17% of a little.

Re:AMD has a score to settle

[bn-7bc]

well I can't rermember seeing an amd ad ever. One thing that is importantto remember is (at least in my expirience) is that most of the intel ads are not from intel but ads from the likes of Dell ( an inyel only shop) that says the they use intel cpus

Re:AMD has a score to settle

[Anonymous Coward]

Why just mention it when you can watch it?
http://www.amdzone.com/modules.php?op=modload&name =News&file=article&sid=3359&mode=thread&order=0&th old=0

Re:AMD has a score to settle

[servognome]

And which company haves the higher revenue? 14% of a lot is more than 17% of a little.

The point is AMD isn't choosing to not advertise to save money. They just don't have the economy of scale to get the same bang for the buck.

Public venue?

[Nuclear Elephant]

Should Intel accept AMD's challenge, the duel would take place at a public venue to be announced in the coming weeks

Place the chips in an unmarked bag and drop them in the trash on the corner of 2nd and 4th. We'll let you know when our neutral, third-party testing lab is finished with them and post their results.

Re:Public venue?

[utlemming]

Complete with Motherboards, and 4gb of DDR, and the hottest graphics card on the market too. Cases are optional.

The Battle!

[Ecko7889]

"Lets get ready to Rumble" "And in the left corner weighing in at....."

They seem...

[Anonymous Coward]

to be very confident in their processors.

Re:They seem...

[dragonp12]

It's more the fact that they know for sure that something like this would never actually happen. Intel would have everything to lose, AMD nothing.

Finally an open match

[GreatBunzinni]

Finally an open match between two independent groups who compete to prove who has the best product available and competing in a leve where the rules are set and the process is (at least somewhat) clear.

This is what has been missing in the benchmark field. I hope that this trend picks up and that from now on we see the companies battling it out on the technical field instead of the marketing field.

P.S.:yes I know. This is marketing too. But still, it is a lot better than obscure references and funny and dubious charts which show vage and misleading numbers.

Much Lucha

[Stanistani]

Nah.

It'd be more interesting if the opposing CEOs dressed up in colorful, masked outfits with capes and boots and took turns body-slamming each other in a ring, surrounded by thousands of cheering fans, and the play-by-play by a prominent Mexican wrestling announcer.

...and it would be as accurate.

Re:Much Lucha

[Anonymous Coward]

Maybe the CEO's should have a trans-atlantic swimming race. They might even beat the record set by Jon von Tetzchner (Opera) earlier this year!

Re:Much Lucha

[EnsilZah]

So basically what you are saying is that you want them to act more like Steve Ballmer?

Re:Much Lucha

[aBum]

I've used AMD processors in the last 3 computers I've assembled, but I'd be more than happy to use an Intel chip in the next one if Intel's CEO could pin AMD's CEO in three rounds of throwdown (simulcast on G4 TechTV and Spike TV of course).

who cares?

[mnemonic_]

Two computers crunching numbers next to each other, big deal.

Fistfight between executives, I'd watch.

Re:who cares?

[Anonymous Coward]

Reminds me of an ad I saw once:

People in lab coats are on top of a building, holding tennis balls.
The voice-over says:
"In repeated lab tests, our balls were shown to bounce higher than our competitors."

They release the tennis ball.
Then they grab a guy in a suit and throw him over the side :)

Re:who cares?

[Phu5ion]

Sounds like a reality TV show for Fox.

Re:who cares?

[Comatose51]

Screw that! I want to see the secretaries go at it!

Re:who cares?

[shawnce]

How about a good arm wrestle match?

One outrageous incident was his arm-wrestling showdown with the CEO of Stevens Aviation in 1992. Both Stevens and Southwest were using the advertising tagline "Plane Smart." To settle the matter, Kelleher suggested an arm-wrestling competition with the winner keeping the rights to the slogan. Kelleher lost the match, but the event generated so much good will and publicity that Stevens let Southwest continue use of the tagline.

reference

Re:who cares?

[Anonymous Coward]

>> Fistfight between executives, I'd watch

Paul would kick Hector's ass.
Then Hector would file another lawsuit.

In other news...

[Nuclear Elephant]

... global warming on the rise again.

Please tell me...

[guitaristx]

...that this competition will be measured in FLOPS and not in MHz.

Re:Please tell me...

[randyest]

Why FLOPs and not MIPs, or integer operations, or specPerf units, or . . . ?

It's all largely irrelevant and easy for either party to finegle. The point is, there is no univerally agreed-upon benchmark. Each is better than the other at something, and that's what they tout. Unless this is effectively rigged to be a tie in which each wins 50% of the "showdowns" it will never happen.

Re:Please tell me...

[Jeff Molby]

Unless this is effectively rigged to be a tie in which each wins 50% of the "showdowns" it will never happen. Your point is correct, but it wouldn't even happen under the scenario you described. There is NO way Intel wants AMD to be seen as an equal.

Re:Please tell me...

[akuma(x86)]

Uhh...flops is just as stupid as Mhz.

Intel would clearly win the FLOPs because they have higher peak floating point execution bandwidth - due to higher frequency (on P4). Or more FP execution units on Itaniums.

Performance is measured in seconds, not flops or mhz. Pick a computable task - measure how long it takes each machine to complete it.

Re:Please tell me...

[Stonehand]

...or how fast they can fry an egg, or burst into flame.

Re:Please tell me...

[dlakelan]

There *IS* a pretty decent benchmark out there which actually correlates well with processing power in terms of data processing speed.

It was called the HINT benchmark and it was developed by a guy at Iowa State University Ames Laboratory. I think they tried to commercialize it but it didn't get off the ground. It was released under the GPL. It's out there in various places.

If you can find a copy of the software then compile it up, and it produces some very interesting data. It produces a curve of "quality improvements per second" as a function of memory usage...

Re:Please tell me...

[dlakelan]

To give some more follow up information, I noticed that

This site from TU-muenchen has relatively recent HINT benchmark graphs for Opteron and P4 and Xeon processors.

Re:Please tell me...

[dsci]

flops is just as stupid

Intel would clearly win the FLOPs because they have higher peak floating point execution bandwidth

Performance is measured in seconds, not flops or mhz. Pick a computable task - measure how long it takes each machine to complete it.

UH?

If Rate(Intel) > Rate(AMD) as you assert, then how in the world can Time(Intel) NOT be < Time(AMD)?

Time = Number of Operations / Rate

So, if the Number of Operations for each CPU is the really the same (what I assume you mean by "a computable task"), the comparison of FlOPS is the same as a comparison of total compute time.

I'm assuming we are using FLOPs for some specific calculation (say matrix inversion, or what-not), so that it is really an average FLOPs over several instructions in the ISA.

Re:Please tell me...

[beelsebob]

Yes, and then try and compile something, and realise that it has nothing to do with floating point performance what so ever.

Re:Please tell me...

[akuma(x86)]

FLOPS don't include integer ops.
Also, compilers may compile things differently for different machines. For example - SSE3 or Itanium. 64 bit vs. 32 bit. etc...

Re:Please tell me...

[merreborn]

If Rate(Intel) > Rate(AMD) as you assert, then how in the world can Time(Intel) NOT be You argue that since intels processors operate at a higher frequency than AMD processors, that an intel processor will inevitably complete any task faster than an AMD. You completely neglect the fact that each instruction takes a different number of cycles to execute! A crude, and almost certainly incorrect example: Given an intel processor at 3.6 ghz and an AMD proc at say, 2.8 ghz, assume it takes the intel processor 5 clock cycles to calculate the product of "2 * 2". Assume the AMD processor can do the same multiplication in 2 clock cycles. Clearly 2 cycles at 2.8 ghz is far faster than 5 at 3.6 ghz. This becomes even more complex when you consider the more complex instructions each processor boasts: for example, the MMX instruction set. MMX provides vector math functionality, meaning you can do certain calculations in a single instruction (but multiple clock cycles) that would have otherwise required many instructions, and therefore many more clock cycles. Also, consider the Pentium M, which performs tasks twice as fast per Ghz as the Pentium 4! What it all comes down to, is Mhz/Ghz is only part of processor performance! That's part of why it's far more useful to consider things like: "How long does it take this computer to render a 3 billion poly 3d image?"

Re:Please tell me...

[akuma(x86)]

Just to elaborate:

Let's say I had some pseudo-code:

for (1 to 100 million) {
        if (test_function() == true) {
              A = A * B;
        } else {
              A = A + B
        }
}

A VLIW might compile to

LOOP:
        PREDICATE = CALL TEST_FUNC
        MUL, ADD, PREDICATE
        JMP LOOP (exit when done)

You'd do the MUL and ADD on every iteration an predicte the result based on the test. So you'd execute MUL and ADD every iteration but throw away some computes because you only take one mul or add depending on the test.

A non-predicted machine might do this:

LOOP:
        CONDITION = CALL TEST_FUNC
        JMP CONDITION, non-mul
        MUL
        jmp LOOP
non-mul:
        ADD
        jmp LOOP

You'd execute half as many floating point ops. If you have a good branch predictor (say test_func() was very predictable) you might be just as fast as the predicated machine with half as many FLOPS.

Re:Please tell me...

[dsci]

You completely neglect the fact that each instruction takes a different number of cycles to execute!

I did not neglect that at all. That's why I said use an average FlOPS number obtained over some set of real calculations. At this level, it simply is a number to compare.

Re:Please tell me...

[dsci]

Well, I spend more time COMPUTING than compiling, so FlOPS (or number crunching in general, which would include cache parameters, memory bandwidth, etc), is a better metric. I don't consider compiling a meaningful benchmark for the work my computers do.

If I spend 1-2 hours compiling, then a single calculation takes 35 hours and I must do about 20 of those calculations, the compile time is insignificant.

As always, YMMV, and the fact remains, there is no single benchmark or CPU/system best for all tasks.

Re:Please tell me...

[beelsebob]

Well, I spend my entire day compiling things... My point is not that people do not use floating point operations, it's that people use things that are not floating point as well. Compiling things is just one example of a task that uses almost no floating point processing, and a lot of branch/cache work. No test can deal with all these different things.

Re:Please tell me...

[The Warlock]

They'll measure in frames per second in Half-Life 2, of course. I mean, really. What other benchmarks matter?

Re:Please tell me...

[GaryPatterson]

I don't think there's a single metric that will cover a processor's performance.

You'd be better off to run an unoptimised set of integer and floating-point tests, then an optimised set of the exact same tests.

After that, you'd run a set of memory access tests and that should cover processor performance.

Of course, you've got different motherboards making it a bit more complex. Any tests you run depend on motherboard limitations or sub-systems, so you can't easily compare the two - you're left wondering if the motherboard choice is helping or hindering the test results.

As with all things, the simple answer of one number is bound to be a poor answer. You have to look at what you want the thing for, and then pick the tests that matter to you. For most users, application-level tests (such as Photoshop, Office, Windows Explorer or OS X Finder) give the best way to compare processors.

Neutral 3rd Party?

[yellowbkpk]

Isn't that an oxymoron by now? Every single "this was tested by a neutral 3rd party" review you hear about has a corresponding "that neutral 3rd party wasn't neutral at all" story to dupe. Even if there are "neutral 3rd parties" out there, will anyone trust them enough for AMD to be successful in this marketing campaign?

Doubt it.

Re:Neutral 3rd Party?

[__aaclcg7560]

Let the boys from id Software set up a Doom 3 server so AMD and Intel can frag it out. That should be a fair and neutral test to determine which CPU sucks down the pineapple. :P

Re:Neutral 3rd Party?

[bfree]

You know you may actually have something. Set up a major gaming contest with various suppliers of machines. Don't benchmark, but rotate the players around the machines (keep all ui hardware identical) and use statistics to see which supplier won.

Secondly you could do a rerun of the recent assisted chess competition (afair you couldn't cheat unless you managed to get away with an illegal move) except simply make it a software assisted competition. Bring whatever code you want, but you have to run it on the supplied machines. Again comparitive performance on different platforms amongst all the competitors should yield real results.

Basically set repeatable (but unique so it's a "special" definition of repeatable) challenges which require computer assistence, and see which machine/platform/vendor/os/whatever gives it's users the biggest competitive advantage.

Re:Neutral 3rd Party?

[Anonymous Coward]

Google "Microsoft vs Linux" and read all of the pro-Microsoft whitepapers and articles.

--
Proudly Remaining Neutral Through Microsoft Funding

Why should they accept?

[weedenbc]

Accepting would only hurt Intel so there is absolutely no reason why they should take the challenge. I mean come on, they are in the business of making money, not proving a community of geeks right.

AMD has scored some points with this challenge but IMHO missed a huge opportunity. They should have started an ad campaign pointing out that all the P4 class products that Intel has dumped on the world were sub-par to their own.

Intel presentations today were full of hyping a per watt performance. I would have immediately launched an ad campaign that showed exactly where Intel stood with it's current desktop and server offerings in a per watt basis.

It really pisses me off how a company can talk up its products and convince a ton of people to buy them, then turn around and say that they really sucked and they just managed to sucker people in with marketing and brand name recognition.

Re:Why should they accept?

[popo]

I couldn't agree more.

AMD has *THE WORST COMMUNICATIONS DEPARTMENT I HAVE EVER SEEN*

It blows me away that AMD doesn't run an ad campaign that says something like "AMD: Faster"

AMD's consumer messaging and advertising is hilariously bad. (Just look at their website, its like something some secretary did in Frontpage).

AMD continuously acts like they don't have the money to fight Intel's 170 Billion Dollar image. Its hilarious. AMD is an EIGHT BILLION DOLLAR COMPANY! I know 2 million dollar dot-coms that have a more savvy marketing department.

AMD should position itself as the more expensive, elite brand. Not the sucker underdog.

If you can't meet production numbers, be Mercedes. Be Ferarri. Don't be Saturn and charge a higher price for crying out loud.

My 2 cents.

Re:Why should they accept?

[mcrbids]

AMD continuously acts like they don't have the money to fight Intel's 170 Billion Dollar image. Its hilarious. AMD is an EIGHT BILLION DOLLAR COMPANY! I know 2 million dollar dot-coms that have a more savvy marketing department.

Remember that the purpose of a marketing deparmtment is to (ahem) make money. Are these two "million dollar" dot-coms making EIGHT BILLION DOLLARS?

So, who's savvy, again?

Re:Why should they accept?

[Locke2005]

It really pisses me off how a company can talk up its products and convince a ton of people to buy them, then turn around and say that they really sucked and they just managed to sucker people in with marketing and brand name recognition.

It really pisses me off too! I, mean, how many times can Microsoft sell the latest software release by telling everybody "You must upgrade now! The previous release was too buggy and insecure!"... Oh, wait, we were talking about Intel, weren't we...

Yes, all the latest Intel IDF hype can be summarized very simply as "Over the next year, our master plan is... to catch up with AMD!

Re:Why should they accept?

[Jeff DeMaagd]

I think there is still a point though, marketing plays a larger role in helping get and attain market share than geeks care to admit.

AMD and AMD fans shouldn't be complaining about market awareness when AMD barely has an ad budget. Sure, I see an occasional full-page magazine ad but I don't remember any TV ads. Really, they should at least drop a couple mil for a spot during the Superbowl, at least to show the PHBs that AMD does exist.

Re:Why should they accept?

[TexVex]

It blows me away that AMD doesn't run an ad campaign that says something like "AMD: Faster"

Yes, because we all need to take our 99.6% CPU idle time up to the next level. If you're utilizing more than 0.2% of your CPU cycles on average, you're obviously yesterday's crap.

Re:Why should they accept?

[Nasarius]

AMD already does good business in the server/workstation market, thanks to Sun and others. The truth is that AMD can't handle a huge amount of demand. They're expanding gradually, building fabs with new and better tech to crank out more processors, but they have nowhere near the production capacity of Intel.

Re:Why should they accept?

[moogleii]

Yeah, in my opinion they should ditch the color scheme, or adjust it. It comes off more like an environmental company or something. Yes I know, green doesn't always mean trees. I mean /. is green. I'm just talking initial impressions of people who glance at the logo on some bilboard, etc.

Re:Why should they accept?

[Wolfier]

You totally missed the point.

Even when we were running OS/2 on a 386, 99% CPU was idle.  I bet you don't need anything more...

What matters is, when you actually use your CPU in that 1% of the time, it runs as responsively as possible.

Plus, if you're running a high load server, you'll NOT find 99% idle CPU time.

Re:Why should they accept?

[Pranadevil2k]

Intel is screwed either way, isn't it? They won't accept it because there are plenty of benchmarks out there in the web showing AMD performance over Intel products and they don't need to have AMD put on this show to prove it. On the reverse of that, if they don't accept the challenge, they're giving AMD the community chest, so to speak. AMD will always be able to say they gave Intel a chance at proving they were faster, and Intel refused. The reasons behind their refusal are irrelevant, and AMD can say whatever they like about it.

I agree this is marketing money that could be better spent elsewhere. I don't think pointing out architecture differences is the way to go, though. Nothing gets customers to dislike you more than proving that they're stupid, and average Joe consumer doesn't know jack about processors. He might know that Mhz are good, but when you start talking about transistors and wattage and bus latency he'll not only lose interest but feel offended that they're trying to tell him something he didn't know. At least, that's how American consumers react...

Re:Why should they accept?

[GotenXiao]

Part of AMD's lure is that you can buy better CPUs for less money than you can with Intel.

AMD should position itself as the more expensive, elite brand. Not the sucker underdog.

That would be a bad thing. All they need to do is publicise the fact that, if you want good gaming performance (*cough*Dell XPS ad*cough*) the CPU to go for is AMD. If you want high performance servers, the CPU to go for is AMD. If you want a machine that will work even if you get a rogue application trying to suck 100% CPU time, has better security, performs FAR better than the equivalently clocked Intel (my S754 Athlon64 2800+ blows the Intel Celeron 1.7GHz away - both are budget CPUs), and has evidence to back it up, they'll be doing a fair bit better methinks.

Geek Pay-per-View

[jayhawk88]

Maybe we could get a few undercard matches as well?

Windows v. Linux, quickest from blank disk to running system
vi v. emacs, first to edit a 10 page document
RMS v. Bruce Perens, which person does the audience kill first

Re:Geek Pay-per-View

[Anonymous Coward]

How about: Windows v. Linux, quickest from blank disk to running system a normal person actually wants to use?

Re:Geek Pay-per-View

[Cheapy]

Nah, we want to give Linux a chance.

Re:Geek Pay-per-View

[Cannucklehead]

You missed one:

Windows v. Linux, quickest from running system to blank disk.

Re:Geek Pay-per-View

[A Brand of Fire]

You forgot the annoying drunken announcer voice... SUNDAY! SUNDAY! SUNDAAAAAAAY! AMD vs. INTEL! INTEL! INTEL! CERRHOSIS OF THE LIVER! LIVER! LIVER!

Re:Geek Pay-per-View

[Anonymous Coward]

What does Bruce Perens' entrance music sound like?

Re:Geek Pay-per-View

[dhasenan]

Depends on the Linux. Ubuntu gets decent after about three hours (add universe and multiverse repositories; add mp3 support, abiword, autoconf, automake, etc); Gentoo takes two or three days.

On the other hand, Windows...you have to install nLite and make another install disk, then reinstall with only the features and services you want or need, then reinstall, then add applications.

Re:Ladies and gentleman

[Anonymous Coward]

Proof for why Slashdot should have a "-1 Boring" moderation option.

Two Reasons

[popo]

Right now they have 2 objectives:

1) Ensure that the hundreds of millions spent on their new German production plant (set to open soon) was worth it by creating a media frenzy & consumer demand.

2) Cast a spotlight on Intel's unfair marketshare by once again proving that Intel's products are inferior and not capable of maintaining their position in the marketplace without unfair practices.

Re:What software? What terms? What Processor?

[hackstraw]

I'm not sure what "corresponding Intel product" Intel would use. If its an x86 chip, give it up Intel.

Now in terms of bang for your buck, AMD Opteron wins hands down. Now for raw performance (as if only geeks care for games I guess) I would like to see a showdown between Itanium and Opteron.

Sleeper chips...

[Jack Johnson]

I just wonder where they'll hide the n20 bottle?

Full page ad on back of A section in WSJournal

[WillAffleckUW]

I saw the full page ad for this on the back of the Wall Street Journal print edition this morning.

Let the power-saving multi-core wars begin!

Gentlemen, start your broadband!

Re:Full page ad on back of A section in WSJournal

[Anonymous Coward]

WillAffleckUW
Your signal to noise ratio is quite low!

Please refrain from posting inane comments such as these! I vow on my mod points to save the intelligence of our younger readers!

Thank You, The Readers of Slashdot

Re:Full page ad on back of A section in WSJournal

[Anonymous Coward]

Your signal to noise ratio is quite low!

No shit. Welcome to Slashdot.

Re:Full page ad on back of A section in WSJournal

[Anonymous Coward]

what mod points? I haven't seen mod points for a long time - have you?

Same binaries in use, but whose compiler?

[Anonymous Coward]

If they start benchmarking then whose compiler will they use?
http://fahwiki.fahstats.com/index.php?title=FAH_%2 6_QMD_%26_AMD64_%26_SSE2

Re:What does AMD's speed matter..

[thegamerformelyknown]

Don't they have a new plant opening soon?

McDonald's VS. Burger King?

[gabecubbage]

I don't see where Intel would enjoy any strategic advantage in participating.

Burger King has in the past been fond of touting its #2 status -- as has Pepsi.

But the big boys, McDonald's and Coke, generally like to pretend that #2 doesn't exist. After all, it would only publicly legitimize their fear of a threat by doing so. AMD gets positive publicity whether they play and win, play and lose, or if Intel refuses the contest.

Whereas Intel can only AT BEST hope to win the contest and essentially say "Hey, it's actually true that there are viable alternatives to our technology out there, but just remember that for the time being we outperform the competition by 1.23%."

Hmm...

[plug_it_in]

So let's say that AMD wins. Are people besides you and I reading this post gonna care? Is my grandma gonna care that her brand new Dell came with an Intel processor that can't perform up to par with AMD? No. AMD is definitely trying to score some points with the geeks, and I think that this competition is good for them because geeks are the people that are gonna be faced with the decision of Intel and AMD.

Re:Hmm...

[Anonymous Coward]

Exactly. Geeks spur the movement. Let's face it ma and pa buy Celeron and don't know they are celerons. It is a pc and I check my email and see the photo's of my grandkids.

Geeks say more man. Buy this man. Over clock this man. etc. etc.

Re:Hmm...

[Anonymous Coward]

Not only do geeks buy more, but who do people turn to for advice?

"I'm going to get a new Dell, they have those Intel things I've heard about on the TV"
"No grandma, you'd be better off with a... ...and it's cheaper too."

Re:Hmm...

[nzkbuk]

Obviously your grandma and mine have vastly different idea's.
Yours went to dell and purchased the hype.
Mine came to me, told me what she wanted to do and I found out some prices based on slightly different components, eg crt vrs flat screen.

That aside, while the average consumer will buy a generic brandname, most businesses will seek atleast some advice before they go and start spending money. So if AMD manage to convince enough geeks (and can put together half good pr that we can use to sell it over all the "intel inside" hype) then their market share will go up.

Oh and one other thing, the last couple of reports have shown home users aren't currently buying pc's in the numbers they were a few years ago. seems like most of those who want a have got one and don't see the point in spending more money.

Re:Hmm...

[hustlebird]

Are people besides you and I reading this post gonna care? Is my grandma gonna care that her brand new Dell came with an Intel processor that can't perform up to par with AMD?

Actually, Yes.

It's the association to AMD's other chips that matters. If joe shmoe goes into a best buy and sees two machines, the same speed (i.e. p4 3ghz vs athlon64 3000) He's going to remember that amd won with a specific chip, and the rest of the chips will look better in comparision.

It's like in cars, Chevy builds the corvette, not because they are going to sell tons and tons of them, but because by making that nice (and fast) car, all of chevy's other cars receive a big boost in rep through association.

Now that the gauntlets thrown, should they accept?

[WillAffleckUW]

because - in another article I read today in the Wall Street Journal - they just were told by the United Kingdom that all government bids can no longer specify Intel chips, but must be x86 generic specs.

They have serious market and mind share inroads they need to fight back on, or they'll be another Lotus in a few years, or Texas Instruments.

Remember them?

Relevance, evidence, simple thought?

[Cyphertube]

And your off-the-cuff remark is based on what?

I remember the height of the cola wars with the Coke/Pepsi taste tests. Yeah, Pepsi initiated them and Coke still dominates, but both companies benefitted, and neither of them was exactly young.

Where you ostensibly see immature, some of us might see refreshing.

Intel's "next-gen" CPU's show that they are behind

[t35t0r]

From the previous /. article today about Intel's "next-gen" CPU offerings in 2006, it is evident that Intel is currently behind AMD in the dual core arena. What Intel will offer in 2006 is already available from AMD (single die dual core and lower power consumption).

When pigs fly or Bush meets with Cindy.

[lupine]

This wont happen because AMD has a clear lead in server cpus. Xeons dont scale worth a darn and the Itanium has been doa. Intel is treading water and pumping up the marketing machine just to stay afloat. No real benchmarks or numbers displayed at IDF and AMD has more processing per watt right now.

What I want to see is Bush(might makes right) vs Cindy(you killed my baby) in a boxing ring. If bush beats up on a woman or gets pummeled he would lose either way.

Re:When pigs fly or Bush meets with Cindy.

[Anonymous Coward]

Army Specialist Casey A. Sheehan specifically re-enlisted in the Army after the invasion of Iraq began, and he specifically volunteered for the mission on which he was killed.

If the war was, as his mother claims, "illegal" and "immoral", well, the son she lost was a willing accomplice to that illegal and immoral war. While I can feel for the grief of any mother, I am hardly going to pay any attention to one blaming others for the death of her son at the hands of those resisting his war crimes.

If, on the other hand, Casey A. Sheehan was an innocent victim, not a willing accomplice in an "illegal" and "immoral" invasion and occupation, then the war itself was neither illegal or immoral. In that case, while it may hurt her to have lost her son, Bush did nothing morally wrong in invading Iraq, even if it may have been usound strategically. Ms. Sheehan is then owed nothing more than the condolences she already recieved the first time she met Bush.

Now, I appreciate that Ms. Sheehan is unable to logically evaluate the situation. What I do wonder is that the others who denounce the invasion as illegal cannot. If the war is illegal, then anyone who has volunteered for military service since it began has volunteered to participate in a war crime. It is inherently irrational to denounce a war crime while supporting volunteer war criminals with the argument that they are merely following orders they volunteered to follow.

If you think the war is illegal, then when you "support our troops", you must make a distinction between those who volunteered before the war and those who volunteered after. To support the latter is to support volunteer war criminals and their crimes.

Re:Grow up?

[GFPerez]

I think that the corporate part of AMD is very serious, because you can only design good chips with responsability. BUT I agrre with you in respect to their marketing department, which I agree that it's a joke.

2nd and 4th?

[Anonymous Coward]

Corner of 2nd and 4th? Whoa, there, boy! We're not in Euclidian space anymore!

Re:2nd and 4th?

[Cecil]

That would make perfect sense in Calgary. They've got a cartesian grid of streets (x) and avenues (y).

So that would be 2nd Street and 4th Ave. (in the Northeast quadrant I guess, assuming they're both intended to be positive)

Re:2nd and 4th?

[dgatwood]

New York-style, as in 2nd street and 4th avenue.

Re:2nd and 4th?

[Mornelithe]

I'm at the corner of 1st and... 1st. How can the same street intersect with itself? I must be at the nexus of the universe!

AMD could actually lose this one

[vlad_petric]

If Intel, for instance, chooses to pit a dual Itanium 2 system against the dual Opteron. Itanium 2s can have shitloads of L3 cache (like 6M, vs 1M in the Opteron), which is perhaps the most important performance feature of a server chip.

Keep in mind that server applications are a totally different beast from desktop/gaming apps/multimedia apps (things that most people here on slashdot are accustomed to). While a media application has a very high instruction throughput (say, 2 instructions retired per cycle, or more if you consider the SIMD part), server applications can be as slow as 1 instruction retired every 10 cycles. This is because they have poor cache locality, and they block on data from the main memory. In any case, for a server app you generally want as much cache as possible.

Re:AMD could actually lose this one

[lupine]

Cache helps improve perforance when your entire dataset can be loaded to cache, but I expect most realworld benchmarks would use a database much bigger than 6MB. Thats when the Dual Core nature of the dual opteron(twice the cpus) and built in memory controler would clean house.

Re:AMD could actually lose this one

[HavokDevNull]

Itanium and its "IA64" architecture, is merely able to SLOWLY emulate the myriads of 32 bit software titles we all use today, even in server applications. Unlike Opteron which can execute 32 bit x86 code natively along with x86-64. And I will take compatibility any day over the amount L3 cache a CPU has.

Re:AMD could actually lose this one

[pla]

AMD could actually lose this one
If Intel, for instance, chooses to pit a dual Itanium 2 system against the dual Opteron.

Not in x86 emulation mode, they won't.

AMD covered their butts on that one... The challenge specifically states x86, with "the corresponding Intel x86 server processors that are commercially available in volume."

Re:AMD could actually lose this one

[owlstead]

They would win the contest and loose the marketing battle. Itanium is dead in the water and Intel knows this.

Re:AMD could actually lose this one

[NatteringNabob]

AMD would still win on price/performance as Itanium chips are much more expensive and the volumes are so low that the supporting chipsets and motherboards are more expensive. Itanium would lose on performance/watt too which Intel just yesterday decided was an important metric.

Re:AMD could actually lose this one

[dascandy]

Try to match the prices up within a factor of 5, and you'll not see an itanium win.

That'd change the situation to probably an 8-way opteron versus a single Itanium 2... granted it could win, in the case of a single-threaded program...

Right Tool For the Job

[dsci]

Of course, the real answer here is AGAIN is use the CPU that best suits your application. They each have strengths and weaknesses. Any comparison is necessarily apples/oranges.

If you don't know the needs of your application, or your application is somewhat neutral, then it does not matter so much. Enter factors like cost/FlOPS and convenience of other components (mobo, etc).

Re:Right Tool For the Job

[UnrefinedLayman]

Of course, the real answer here is AGAIN is use the CPU that best suits your application. They each have strengths and weaknesses. Any comparison is necessarily apples/oranges.

Wait, let me get this straight: two x86 CPUs using the same type of memory and identical peripheral hardware is "apples/oranges"? If your comparison were a different architecture against x86, like PowerPC, then you'd be a little more on target. This is more like fuji apples/gala apples

Yeah, like...

[Nom du Keyboard]

Should Intel accept AMD's challenge,

Yeah, like this is ever going to happen.

From AMD.COM

[WndrBr3d]

Here's the actual link to the challenge issued by AMD to Intel on AMD's own website.

Much more information than the /. link.

Re:Why should they accept? Fraud?

[Nom du Keyboard]

It really pisses me off how a company can talk up its products and convince a ton of people to buy them, then turn around and say that they really sucked and they just managed to sucker people in with marketing and brand name recognition.

Can they sue for fraud?

Can the win?

Can they get paid in something other than coupons?

Nothing to see here, move along (Intel declined)

[Shazow]

AMD proposed, Intel declined.

What's next on the agenda?

- shazow

Oooh- the excitement

[PhYrE2k2]

I can see the excitement now. The suspense. THe music pumping through the crowd of geeks in a huge outdoor arena. The ushers directing you to your seats. The skantily clad women handing out deoderant (geeks... get it?) and sunscreen.

Giant projectors displaying 400ft screens, each with a progress bar from the testsuite going from 0% to 100%... slowly moving... getting closer [four hours later] almost there...

Please- this doesn't sound like a venue I'd really want to attend. I imagine the press would be just as bored as well. Good marketing effort though.

-M

Intel's Reaction

[entj1]

Here's a quote from Intel in this article

Separately Wednesday, Otellini addressed its smaller competitor, Advanced Micro Devices (nyse: AMD), which today took out full page ads in national newspapers to challenge Intel to a "dual core duel" to see whose chips are faster. Said Otellini, declining the opportunity to attack, "I think that companies and products are best judged in the marketplace."

Re:Intel's Reaction

[arose]

"...in the marketplace, without objective information, but lots of dishonest ads"

Forget Performance per Watt

[Nom du Keyboard]

AMD64 dual-core technology provides industry-leading performance, is easy to upgrade and is energy efficient,

None of these matter as much to me as Performance/$$.

Re:Forget Performance per Watt

[MerlinTheWizard]

None of these matter as much to me as Performance/$$.

Then you're obviously clueless, because power consumption equates to a lot of money. TCO (Total Cost of Ownership) is a notion that is to be considered here. You should take a look at some companies' electricity bills.

Re:Forget Performance per Watt

[KillShill]

how about not financially supporting evil companies?

to the extent possible and not too inconvenient for most people.

Intel already turned them down

[prisoner-of-enigma]

Intel's Paul Otellini has already publicly refused to take up AMD's challenge. He said he prefers such things "to be worked out in the market."

That's Intel-speak for "we know we can't beat you in any fair contest, so we're just going to outspend you ten-to-one in marketing and make everyone think we're faster, just like we've been doing for the last five years."

Yup, that's the way to do it. If you can't beat 'em, FUD 'em to death.

Re:Intel already turned them down

[Jussi K. Kojootti]

"products are best judged in the marketplace" == FUD? This is currently at +4, but I just can't understand where in Otellinis response there is Fear, Uncertainty and Doubt...

Correct me if I'm wrong but this is what I got from your post: When Intel tries to suggest they are faster it's FUD and not a fair contest. When AMD suggests (by challenging) that they are faster, the same does not apply?

To me this looks like a PR stunt, plain and simple.

Re:Intel already turned them down

[Anonymous Coward]

Seems to have worked for M$ to date

Re:Intel already turned them down

[Anonymous Coward]

>> Yup, that's the way to do it.
>> If you can't beat 'em, FUD 'em to death

Huh? I think you have confused who the party crasher is here.

AMD doesn't have ADF; so, they weasle some publicity at IDF.
AMD doesn't have a marketing dept, so they go SCO and lauch a high publicity lawsuit.

So, let me re-word you conclusion.

If you can't beat 'em, sue 'em.
If you can't beat 'em, leach some publicity at IDF.

Re:Intel already turned them down

[Anonymous Coward]

That's Intel-speak for "we know we can't beat you in any fair contest, so we're just going to outspend you ten-to-one in marketing and make everyone think we're faster, just like we've been doing for the last five years."

Actually, "faster" is meaningless to distributors if the following allegations from AMD are true (read the last quoted sentence of this post in bold for the impatient):

www.amd.com/us-en/assets/content_type/Downloadable Assets/AMD-Intel_Full_Complaint.pdf

Some choice quotes include:

73. As Gateway executives have recounted, Intel's threats beat them into "guacamole." But Gateway is not alone. Prior to its merger with HP, Compaq Computer received Intel threats every time it engaged with AMD. In late 2000, for example, Compaq's CEO, Michael Capellas, disclosed that because of the volume of business he had given to AMD, Intel withheld delivery of server chips that Compaq desperately needed. Reporting that "he had a gun to his head," Capellas informed an AMD executive that he had to stop buying AMD processors.

74. In 2002, Intel pointed its gun at NEC. Intel threatened to discontinue providing NEC with the technological roadmap of future Intel products if NEC did not convert its entire line of Value Star L computers to Intel microprocessors. Without that roadmap, NEC would be at a distinct competitive disadvantage. Predictably, NEC succumbed and eliminated AMD from the Value Star L series in 2002 and 2003.

93. Avnet Inc., one of the world's largest computer equipment distributors and an avid AMD supporter, has also received its share of Intel intimidation. Thus, Avnet cited Intel as the reason it could not distribute AMD parts to the industrial sector. And when AMD launched its Opteron server chip, Intel made clear it would make it "painful" for Avnet were it to begin distributing that chip. When Avnet did so anyway, Intel threatened to cut if off. Another distributor got even worse treatment. In retaliation for Supercom's AMD dealings in Canada, Intel pressured Supercom's customers to switch to another distributor." [bold font added for emphasis]

Literally?

[Anonymous Coward]

It's a good thing you said "literally". I was going to think that the Intel/AMD battle has moved metaphorically into public view.

The Contender...geek version.

[carlmenezes]

nuff said.

Re:What does AMD's speed matter..

[Anonymous Coward]

"Let the chips fall where they may..."

* :)

(I think this would be an INTERESTING battle to witness the results of... both CPU types have their merits/demerits vs. one another, & I hope all fronts are tested, mano-a-mano/head-to-head!)

Ought to be interesting: I, for one, look forward to seeing the results!

APK

Easy as Pi.

[Anonymous Coward]

Get both to calculate pi as far as possible until a certain time limit. The third party will be a stopwatch.

Re:Easy as Pi.

[serialdogma]

Yes, but are we to know the stopwatch does not have rounding errors?

What about Apple?

[Anonymous Coward]

I'm surprised no one has mentioned the recent Intel/Apple deal. Even if AMD does beat Intel in this competition, Intel has two very important factors in its favor: brand recognition, and the recent deal with Apple. With that, the results won't hurt Intel too much if they lose. Why? Look who's using (and will be using) Intel processors, and look at who uses Apple products.

I know someone is thinking the same thing I am.

[rkruse]

Well a battle is all well and good, but isn't CPU superiority more than just speed? I mean, of course it's one of the biggest selling points for chipmakers, but what about overclockablity, heat it produces, and electricity it consumes? Don't those qualities also make a good CPU?

Welcome to New York, now...

[FunWithHeadlines]

"Corner of 2nd and 4th? Whoa, there, boy! We're not in Euclidian space anymore!"

No, we're in New York City. You gotta problem with dat?

Re:Welcome to New York, now...

[Akiboshi]

I don't know what New York City you come from... In the one I know, 4th Avenue doesn't exist until around East 9th Street. =P

map here

Re:Welcome to New York, now...

[FunWithHeadlines]

Who said anything about 4th avenue? Try 2nd avenue and 4th street.

Re:Welcome to New York, now...

[admactanium]

Who said anything about 4th avenue? Try 2nd avenue and 4th street.

i have! actually, i lived on 3rd ave and 4th street. sure hope you don't accidentally poke yourself with any needles grabbing that bag of chips.

Re:first post ass wipes!

[Anonymous Coward]

you know you wish you could be as cool as me.

You mean cool as in FAILING IT 99.99999999999% of the time?

I'd rather be modded down as a troll.

Now, just hold on. . .

[Fantastic Lad]

Who won the last round? Coke or Pepsi?

In the end, they both sell more units to the piglets.

God, I detest corporate marketing.

I've been accused of being, 'Too Serious' and not allowing myself to just, 'Have Fun'. Fine. I can sort of see the argument. . .

After all, this is the only time in history, a window which will be open for only a few brief nano-seconds on the geologic time scale, where I can buy scratch-n-sniff stickers and scratch-n-win lottery tickets, and fizzy sugar water in a can, and pop a high-tech ecstasy pill and wear spandex and running shoes and play video games and watch movies and all of that other crazy stuff. Hey. Sure. It's all fun. This is a once in a billion chance of a lifetime to try all those funky toys out.

But pardon me for thinking there are FAR more interesting things in life than falling in line with some corporate promotional department's greedy wishes so that some millionaire can make another million off everybody's inability to resist their fascination with shiny plastic doo-dads and fake boobs.

Sure, perhaps I might seem, 'Too Serious' to the average burger-eating, cell-phone fashion zombie. --But I also have self-respect and an identity of my very own which I didn't buy at some death star mall. I take pride in not jumping whenever some corporate marketing shill tells me to get addicted to his ice cream.

And I DO NOT CARE whose microchip is faster.

But then. . , perhaps I'm just getting old. All that crap was fun when I was a teen, so to each his own. Live your life in whatever way suits you best!

-FL

Re:Now, just hold on. . .

[Anonymous Coward]

Hey! What have you got against fake boobs?

Re:Now, just hold on. . .

[Anonymous Coward]

All that crap was fun when I was a teen

Congratulations on discovering Existential Angst now that you're in your (undoubtedly early) twenties. Don't forget to pick up your black t-shirt on the way out. Thank you. You go now.

Re:Now, just hold on. . .

[KillShill]

neither coke nor pepsi won.

the corn syrup manufacturers won.

do you think that's sugar you're tasting?

Re:Now, just hold on. . .

[TheLink]

"Hey! What have you got against fake boobs?"

Maybe it's because he's never been against any... ;).

Re:Now, just hold on. . .

[Blackhalo]

Fake boobs are gross.

Re:Now, just hold on. . .

[Richard_at_work]

It is for me, we get the real deal here in the UK.

Re:Now, just hold on. . .

[Fantastic Lad]

Congratulations on discovering Existential Angst now that you're in your (undoubtedly early) twenties. Don't forget to pick up your black t-shirt on the way out.

Close but. . . Well not even close. No cigar anyway.

I'm in my mid-thirties, I don't wear black, and I skipped past Existential Angst because I've never been one who believed in wasting a lot of time being deliberately sad for no good reason.

When I say that there are FAR more interesting things in life than plastic crap, I mean things which are also FAR more interesting than wallowing in self-pity. Life is amazing and beautiful and exciting, but this doesn't mean I don't turn my nose up when a stinky cloud of corporate offal floats by. It's not because I think I'm 'better'; it's because it makes me feel annoyed and barfy. Gut reaction.

-FL

I have them both in house from hp

[agtwilight]

580 vs 585 latest generation - its not even close - the 585 (amd opteron) just flat smokes the intel 4 way zeon - top of the line for everything, SAN backend.

We only buy 585, 385 (amd opteron from hp)now for pizza boxes and moving to AMD blades

AMD ftw - and the price point is less...I do have a dual core intel I am writing this on so i can post on slashdot and check email realllly fast.

Re:I have them both in house from hp

[thesupraman]

>580 vs 585 latest generation - its not even close -
>the 585 (amd opteron) just flat smokes the intel 4 way zeon -
>top of the line for everything, SAN backend.

Wow, for someone who cannot even spell Xeon you seem to be pretty clued up! ;)

BTW everyone, the primary reason AMD is doing this now is because Intel entered the market with low-end dualcore at affordable prices, AMD entered with above-top end maximum performance chips, and they want to try and make a point before intel releases their server-class expensive dual-cores.

Remember, the CHEAPEST Opteron dual-core costs about 50% more than the MOST EXPENSIVE intel dual-core at present, and AMD specified Opteron (not Athlon X2) for this test...

Sigh, Marketing.

Fight Night

[ghostfacekilla]

I think the ceo's of both companys should just fight each other...settle this like real men...he..he

Today's Flame-o-Matic Repeat-o-posts courtesy of..

[Anonymous Coward]

Today we'd like to thank AMD and Intel for throwing the guantlets down, as well as our favorites, the wackjob religious right.

Just remember, this is slashdot, so,

Intel bad, AMD good, Bush bad, very very bad.

Draw!

[circusboy]

maya paint effects at 10 paces.

runs both linux and windows (and mac for that matter,) processor intensive.

and it makes such pretty pictures...

Re:Today's Flame-o-Matic Repeat-o-posts courtesy o

[Anonymous Coward]

Unless said Bush is about arms length below breasts. Ok, so I had to get a Breasts comment in.

Ok, but in more seriousness, Intel would probably be more willing once their latest platform has been completed, especially if they feel their latest silicon will smash AMD all the way around.

oxymoron 101

[Splork]

neutral, third-party testing lab

Re:What does AMD's speed matter..

[Anonymous Coward]

What information are you basing this opinion on? How is AMD unable to "ship enough chips to meet existing demand" when all but one major OEM (Intel-loyalist Dell) and numerous smaller OEMs and resellers are pumping out machines with their CPUs in them every day?

AMD didn't grow its server share 50% last year with magic fairy dust; it did it by selling tens of thousands of Opterons.

And the Athlon 64 isn't the #1 performance desktop CPU just because all the review sites ran benchmarks on a couple of them; gamers are buying them in droves.

Just whose demand are they not meeting?

Re:What does AMD's speed matter..

[a_nonamiss]

That's awfully strange... I just ordered 44 Athlon 64 processors and 6 Opteron processors and had them all within a week. In fact, if I go to newegg.com, every current AMD processor is listed as in stock.

Don't believe everything you read. They might be falling short on a few select processors, but as a company, they are having no difficulties meeting most demands.

Good point, but ...

[vlad_petric]

It's not actually necessary to fit the entire DB working set in the cache to get performance improvements. If the larger cache reduces the misses to memory by, let's say, 20%, it's still a great deal (because these misses easily cost 200->400 processor cycles, and out-of-order execution can't really do much in this time anyway)

Also, the Itanium instruction set allows cache placement hinting. You can tell the processor not to allocate a L1 (or L2) cache line for something that you know for sure you're not gonna need in the forseeable future. Sure, this needs heavy profiling and very careful, manual tweaking of the database server code, but trust me, when you have a benchmark there's a lot of such tweaking (read: cheating).

Re:Good point, but ...

[pkhuong]

Yes, itanium has a very large (and thus much slower than L2) L3 cache. Opterons, on the other hand, are (*iirc, i haven't googled it or anything, lector emptor or wqhateer*) closer to main memory... It could go either way, depending on how random the memory accesses really are. Also, SSE2 (maybe 3?) adds basic caching hints to the ISA, so if a program were to take advantage of those on itanium, it could use the same data on opterons.

Oh, btw, why intel will never pit Itanium VS Opteron: even if Itanium wins, cheap opterons are cheaper than cheap itaniums, and top-of-the-line opterons are cheaper than top-of-the-line itaniums. Worst, it'd only highlight the myth-itude of the MHz myth, or that P4s do little with their high clockrate. Until their next generation low-power 64bit line is out, I don't think they want to risk that.

vi vs emacs

[mkcmkc]

vi v. emacs, first to edit a 10 page document

...with novice users, using the built-in tutorial and help to get started. :-)

(As I recall, the first time I invoked 'vi', the only way I could figure out how to exit it was to unplug the machine.)

Mike

Re:vi vs emacs

[LLuthor]

and C-x C-c or whatever is more intuitive? or what? I am not a vi fan (or even a vi user) but emacs is just as cryptic for the newbie - it has better built-in documentation though.

Re:vi vs emacs

[Hal_Porter]

[ESCAPE] : q ! [RETURN]

Easy to remember

ESCAPE this
COLON thing, (The colon-thing is a nickname for vi.)
Quit and do
NOT (slang for "!". Here used in the meaning "never".)
RETURN

Re:vi vs emacs

[Anonymous Coward]

Actually, the colon thing is a nickname for ex, a line-oriented editor that came somewhere between ed and vi. That's why all the : commands are line-oriented. vi also comes with a whole host of other commands, which you don't press ESC for (the hjkl cursor keys being the first that come to mind), hence the whole reason for the different command and edit modes.

In any case, it's missing the point. Sure, exiting vi and its derivatives is easy. To someone who already knows how. Who consequently isn't a first time user.

It's probably a sign of how unintuitive this is that one of the only vi commands many people know is how to exit it.

Re:vi vs emacs

[edremy]

Mouse over File->Exit Emacs is hard? Who'd a thunk it?

Eric, who has actually used a mouse once with Emacs, but only once. I feel so dirty.

Re:vi vs emacs

[mkcmkc]

Heh. That was 20 years ago. I did eventually learn vi. Well, enough vi so that I could edit the emacs Makefile...

Re:vi vs emacs

[maxwell demon]

...with novice users, using the built-in tutorial and help to get started. :-)

vi has a tutorial and help? Maybe you're thinkin of vim instead? (This is the program which actually is started when typing "vi" on Linux)

I'm almost sure that the huge majority of "vi fans" have never used vi in their whole life, but only used vim.

Yes, I just tried to type ":help" on vi. I got told it's not a vi command.

Re:vi vs emacs

[mkcmkc]

It was a joke. GNU emacs, of course, comes with an extensive tutorial and a voluminous, well-written manual built in. With vi (circa 1986, when I was a novice), you were lucky indeed if the man page was installed, and IIRC, it only documented the command-line arguments.

Mike

Opteron?

[Kent Recal]

Does anyone have any expirience with opteron cpus for servers?
How do they perform under server-loads compared to xeons?

We are currently running dual-xeons only (heavy i/o and/or memory throughput,
webserver and java servlet engines) and I'm curious how a dual-opteron would deal with that kind of load.

Anyone know any serious, real-world benchmarks comparing the two?
Are the Opteron boards mature enough for production use, yet?

Re:Opteron?

[Anonymous Coward]

What do you mean, 'mature enough *yet*'? Opteron systems have been in production for years now, and have proven to be superior all-round to Xeon-based systems.

Re:Opteron?

[eddy]

>We are currently running dual-xeons only (heavy i/o and/or memory throughput, webserver and java servlet engines) and I'm curious how a dual-opteron would deal with that kind of load.

Umm.. the Opterons would smoke the Xeons like nothing you've seen...

Re:Opteron?

[eddy]

Working on the assumption that you weren't trolling, here's how things looked about two years ago.

Conclusion then: "the Opteron simply destroys the competition."

Today AMD has dual-core Opterons and Intel have... well, they have the same old Xeons with a miniscule FSB-bump?

Re:Opteron?

[Wiz]

There are plenty of benchmarks out there if you google for them. Anandtech certainly has a few.

You'll find Opterons are better performing, and scale better. They scale better because each Opteron has it's own memory where as dual/quad Xeon has to fight for the same amount of bandwidth. Opteron's onboard memory controller also gives very low latency to main memory, which helps a great deal.

You'd probably find a dual Opteron would easily beat a dual Xeon is almost all tasks. Most likely exception is streaming & encoding video, there it will just be "close".

Re:What does AMD's speed matter..

[a_nonamiss]

Wait... I stand corrected. AMD has approx. 80 processors at newegg.com and 3 are out of stock. Intel, on the other hand, has 62 processors listed, 8 of which are out of stock. Just how is AMD not meeting demand? Well, I guess if you want to plunk down $1350 for an Opteron 275, you'll just have to wait. That, or buy an Intel... um... wait... they don't have anything comparable. Never mind.

Re:Keep Political BS on the Blogs ASSHOLE

[Anonymous Coward]

You moron.

What does your political BS about a MAN who VOLUNTEERED for MILITARY DUTY that he knew he could be KILLED doing have to do with AMD vs INTEL?

AMD has a clear lead in server cpu's? Obviously not in sales.

Dipshit.

Think you're due for an update

[mjh49746]

Check out

Re:Think you're due for an update

[mjh49746]

wtf happened here? ;-)

Think you're due for an update, bud

[mjh49746]

Check out NewEgg or ZipZoomFly. AMD has no problems keeping them in stock these days.

Slow Down Cowboy?!? You ain't seen nothing yet!

Re:What software? What terms? What Processor?

[LLuthor]

With which CPU emulating which instruction set? Or do you seriously think that x86 and ia64 can be compared like that? Maybe you would like to compare the Opteron emulating ia64 to the Itanium emulating x86?

No, I wasn't

[FullCircle]

Overclockability?

Don't think that's high on the list for SMP systems.

Intel should have accepted!

[PhotoBoy]

Intel should hire the guy in Japan who clocked a P4 to 7.1Ghz to supply some liquid nitrogen cooling, I'm sure Intel would win comfortably then!

IT'S ON!!!

[Anonymous Coward]

it's on! it's ON, oh it's on I tell ya! AMD to INTEL - you got served!! Intel to world - they took our JOERBS!!! THEY TOK OUR JORBSS!!!

(for Southpark fans only)

quad 950 operon's here.

[jasonhamilton]

Our old box was a dual 2.4 (or was it 2.8) ghz xeon.

The new box is a quad operton 950 with 16 gigs of ram.

Both ran as web servers. To give you an idea of how much faster the opteron is (Yes, I'm aware it's 2 cpu vs 4), the xeon box with Zend's php caching is twice as slow as the opteron box without any php caching on a php application with 250,000 lines of code.

No contest at all. Even if we pulled two of the cpus, I'd be willing to bet similar performance.

Infamous

[FuzzyBad-Mofo]

AMD and Intel are not just famous, they're infamous!

Next week: Slashdot editors discuss there lack of grammer skills and generel disregard for proofreadin.

Say the magic word ... TCO

[flyingace]

I would not be surprised if somebody in Intel says the M$ magic word 'Total Cost of Ownership' next. Isnt that what M$ said after trying to run benchmarks against Linux.

I know this sounds like apples and oranges, but all they have to say is Intel processors have lesser TCO than AMD.

Its one of those things that nobody can substantiate or refute.

Re:AMD could actually lose this one - maybe not

[bunyip]

We have a bunch of 4-way Itanium systems running as servers, we tried out 4-way Opteron machines about 18 months ago and they were twice as fast on our app. We've bought a couple of hundred 4-way Opteron boxes since then and we're very happy with them.

Our code is branch intensive with low cache locality. Since Itanium can't handle out-of-order execution, memory stalls kill it, hence the need for a giant cache. Intel's compiler didn't help, we mucked with it for months. For Opteron we used gcc, compile and go, took about a day to move 500K lines of C++.

Intel could only win this on hand-coded floating point.

Alan.

Spin doctoring and word games

[LionMage]

Um, no, that's not what was said.

What is it about abysmal reading comprehension skills that gets my blood boiling? *sigh*

Anyway...

The grandparent poster was, according to my reading, suggesting that Intel's future response is to turn the volume knob to 11 on their marketing, in order to drown out the competition's claims. In other words, the FUD is yet to come. Otellini's response was pretty mealy-mouthed, if you ask me. "Let the market decide" has got to be the most overused line in this industry, and let's face it -- since Intel has engaged in anticompetitive practices, and probably will into the foreseeable future, the "market" isn't going to be a fair metric of merit. Instead, the "market" is only going to be a metric of how well Intel can keep a lock on it through exclusionary tactics.

If Intel is masterful with their PR and marketing, they can spin this refusal to accept AMD's challenge and cast doubt on AMD's claims of greater speed. Nothing suggests that Intel is going to flub this now.

Intel maintains its market position through what I consider to be unfair practices. Sure, they haven't been convicted in a court yet, but so what? Oh, wait... it seems that they did get into hot water in Japan, didn't they? And that is one of the foundations of AMD's lawsuit against Intel.

I'm surprised nobody else has really latched onto this. Intel's statement can be summarized as: "We prefer to let the market do the talking for us." Or, as Otellini put it, "Products are best judged in the marketplace." But saying this presupposes that the market is fair, and that the best products will come out on top.

The market for CPU chips is far from fair. AMD hasn't been able to grow the way Intel has because AMD has been locked out of several lucrative markets for its chips. Time and again, PC manufacturers have been pressured into using Intel-only solutions, or to relegate their AMD offerings to a small percentage of their product lines. So Otellini's words are disingenuous, and propped up by an unspoken logical fallacy.

AMD wants to compete on merit, to see whose product is "best" -- where "best" is apparently defined by whose product is the fastest overall. Intel side-steps this and says, "Hey, let the market decide." But they are shamelessly manipulating the market in subtle and not-so-subtle ways, so they're practically guaranteed to come out on top no matter what. The logical fallacy is the suggestion that greater numbers means better (or suggests a greater degree of "correctness") -- you see this fallacy all the time in a variety of guises. "There are more PCs running Windows than Linux, so Windows must be better." "My nation has more resources, so we're better than you." "More people believe in the tooth fairy than don't, so the tooth fairy must exist."

Intel isn't directly saying their chips are faster. In fact, they're not even suggesting it, at least not with words. As I said before, Intel can spin this so that their refusal to participate will play in the media (and therefore, in people's minds) as a silent rebuke to AMD's claims, and this will cause many to draw the conclusion that Intel's chips are really faster/better/whatever.

Let's be clear here. FUD stands for "Fear, Uncertainty, and Doubt." Intel can't address AMD's challenge head-on because they might lose, so the only weapon they have left is to sow -- you guessed it -- fear, uncertainty, and doubt about AMD's claims and products. So what does Intel do? Talk about the market -- the market that they know they can succeed in because they've done so for as long as they've been in competition with AMD and every other x86 knock-off vendor on the planet. Intel knows they'll win because they don't have to play on a fair, level playing field.

AMD, for its part, has nothing to lose and everything to gain by issuing its challenge, and they're going to spin this refusal as confirmation that Intel knows it can't win on merit alone. So no, technically, AMD isn't e

Re:Spin doctoring and word games

[Jussi K. Kojootti]

You're saying I didn't read the GP, but I'm not sure you read my five liner...

* I'm not claiming Intel is not crooked.
* I'm not saying Intel doesn't use FUD in general.
* I just wanted to comment that Otellinis response was the standard corporate response. No market leader is ever going to let a smaller competitor decide how the battle is fought...

In other words AMD did a nice marketing stunt and Intel responded with a standard (marketing driven) response. You may hate Intel how much you want, but this time they responded like any smart company would -- talking about FUD in this context is just stupid.

Re:Spin doctoring and word games

[prisoner-of-enigma]

I just wanted to comment that Otellinis response was the standard corporate response. No market leader is ever going to let a smaller competitor decide how the battle is fought...

The GP poster did interpret my posting properly, however. FUD can take many forms, some of them "active" and some of them "passive." Active FUD is saying things that are misleading, or downright false. Passive FUD is knowingly allowing someone to draw a false conclusion based on questionable comments. Subtle but important distinction.

Otellini's "we'll let the market decide" may be the "standard corporate response" but it's disingenous in the utmost. Otellini knows AMD would make Intel look rather pitiful in such a match up, therefore he's going to take his ball and go home. Yet it's spun as if AMD weren't worthy of competing against. Masterful politics by Otellini, but at its core it has all the spirit of the FUD even if it doesn't have the letter of the FUD.

Re:Spin doctoring and word games

[LionMage]

You're saying I didn't read the GP, but I'm not sure you read my five liner.

No, I didn't say that at all. I said you didn't comprehend (i.e., understand) what you read. Just because you read something doesn't mean you understood it.

And, for the record, I did read your "five liner," which is why I responded. Considering that the person to whom you responded, prisoner-of-enigma, confirms that my interpretation of his original post was correct (in a sibling post to this one), I think I am in the right on this.

Needs more than two

[Kris_J]

Are there any other CPU companies that could compete in such a challenge? I'd be interested if there was a chance that a small niche manufacturer might win.

AMD: Faster = lawsuit

[msimm]

Sounds like brilliant marketing until you get a pack of high paid lawers from Intel that will either A) try to sue you out of business, or more likely B) argue semantics and logistics and metrics until none of it means anything.

Meanwhile AMD seems to be building better chips and thats got to cost them some money (money you think would be better spent?).

.02

Re:AMD could actually lose this one - maybe not

[vlad_petric]

This is actually very interesting; care to give some details on the processors and server platforms you used (I realize you're under an NDA, but I guess you can say what CPUs you used) ? Also, did you use profiling for Itanium (I guess you did but just checking :) )?

What really surprises me here is that, in general, out-of-order execution isn't that powerful when you have a lot of memory accesses. The out-of-order window on a P4 is about 128 instructions, while the latency of a memory access is always > 200 cycles these days.

The way I think of FlOPs

[dsci]

I know this is not technically accurate, but as a metric of a (purely) computational platform, I think of FlOPs as a practical net quantity of computational work done. So, I'm lumping integer and floating point (and waiting for data, etc) into my misnomer'd "FlOPs."

Or, let me put it another way. The floating point calculation I want done requires some integer work by the processor (eg, loop counters, pointers, etc). Therefore, that integer work should be counted into the practical FlOPs number I am using to benchmark.

Re:The way I think of FlOPs

[akuma(x86)]

Ok, even if you use FlOPs you'd still be inaccurate.

The number of instructions used to execute a program on an Intel machine may be very different from the number of instructions used to execute the same program on an AMD machine.

This is because the compiler may choose to use a different set of instructions based on the properties of the underyling CPU. If you have a very large instruction cache, you may use a different set of instructions since you can fit more bytes in the cache - for example. A more complicated example was given with my other post about predicated VLIW.

Let's say your computable task was a matrix multiply of a large matrix. Depending on bandwidth of the memory system and execution core, the compiler will generate very different code. To use the SAME code on different machine architectures is to give an unfair advantage to the machine with the more optimal code.

Even if you had the SAME code, some machines do an internal translation (to micro-ops in the case of x86) which means that underneath the ISA, they are executing a very different program than the one the programmer told it to (but still functionally correct). Some machines may optimize instructions away, or add instructions to the stream in order to speed up overall execution time.

That's why when you run SPECfp or SPECint, you choose the best compiler for your machine. And to measure performance, you measure time and not instructions/second.

Re:The way I think of FlOPs

[dsci]

I think in a way we are saying the same thing; in my internal 'benchmarking,' I don't really count ISA ops and call it a day. All I am doing is normalizing a 'real world' calculation's actual execution time to the smallest loop. In other words, my "OP" in FlOPs is a high level op, such as "row x column in a matrix multiplication." It ain't perfect, but it encapsulates the real total run-time, just in a reduced way.

And, I also agree that the best comparison is to use fully optimized code/compiler option for each system. If the question I want to answer is "which system is better for THIS task," I sure don't want -march=athlon code running on a P4.

No matter how you slice it, getting a good comparison is extremely difficult, and I would argue very task specific.

Isn't Intel already winning?

[dtjohnson]

Most of the big computer makers are Intel-only. Most stores selling PCs are selling mostly 'Intel-Inside' machines. Most corporate PCs and servers are Intel. AMD's server share just went up to 11 percent and AMD was ecstatic about it. Seems like Intel has the business pretty well locked up so what reason do they have to respond to AMD's challenge? I predict they will just ignore it.

Re:Isn't Intel already winning?

[Anonymous Coward]

The only big computer maker that's Intel-only now is Dell (and maybe SGI for niche applications). HP, IBM, Sun, and virtually every other volume server and workstation maker on God's green earth is selling Opteron now in preference to Xeon. Even ServerWorks, which is a long-time Intel loyalist, started making Opteron products this year.

A large part of what you say about "Intel has the business pretty well locked up" is precisely the reason behind AMD's recent lawsuit against Intel. Despite AMD having a technologically superior product for over two years now (according to general reviewer consensus), it's only now being shown to what extent Intel's been using unfair strong-arm tactics to maintain their dominance. They've been found at fault in Japan already and there are pending investigations in Europe and elsewhere that may impact any US investigation, if that happens.

On the 11% issue, you're right that it's far from huge. But I'd say growing your server market share 50% (from 7 to 11 percent) is a pretty significant year-over-year change, especially when "Intel has the business pretty well locked up" as you say. The fact is Intel has no current product that can compete with a dual-core Opteron and that seems to be the point of AMD's little PR exercise.

They are clearly trying to counter Intel's announcement of their future architecture at IDF, which was also linked on /. today. At the same time, Intel appears to be trying to smokescreen IT managers into believeing they can just wait until 2007 and trust in an architecture that exists largely on paper instead of adopting AMD's Opteron solutions. The recent decline and fall of the Itanium Empire should make that a very cautionary tale for any CTO who puts their trust too highly on Intel's promises alone.

The wrong idea...

[zeno_2]

They need to make a reality show about it. Think about it.. Throw a bunch of Intel engineers in one house, AMD in the other, and for the first 6 or so episodes show them building the server. Have all the usually reality type arguements and such. It would be a hit!

Tower opterons?

[TheLink]

Seems if you want opteron servers from say IBM/HP/Sun, you can only get blades or rack mounted ones, not tower models.

Rack or 1U servers are more popular with large corps (those with datacenters, lots of racks etc). And many large corps are more brand/marketing sensitive.

Just wondering why I can't seem to find tower opterons from any of the big names. Would be nice to have a server that you can stick lots of SCSI drives into, without having to resort to expensive SAN stuff.

BTW just curious: what's the max throughput your SAN can provide?

Re:Tower opterons?

[Zemplar]

You can stuff 4 U320 drives into a Sun W2100z (dual Opteron) or a total of 5 if you use the vacant 5-1/4 drive bay.

I own a W2100z and it just plain smokes anything else I've seen in the category. Granted it is designed as a workstation, but the build quality is the best I've ever seen.

it's a walk-off!

[mumrat]

(but those intel chips are so *hot* right now!)

quakecon

[kronchev]

I was at Quakecon, which was sponsored by Intel. Saturday (i think it was, it's all a haze) an AMD guy came and in front of the center was passing out AMD shirts and tattoos. By the time he was out of shirts one of the Quakecon organizers came over and started yelling at him that he "couldn't be there", and AMD wouldn't be a future sponsor if he stayed. They went off to the side and yelled for a while. I went and got something to eat.

On the way back I see the AMD guy walking away from the center, with a huge stack of temp tattoos in his hand. I ask for one, he gives me THE WHOLE STACK and says "don't let them see these or theyll take your badge and kick you out".

Later that day I saw him and his coworker as they were leaving, they gave me about 10 more shirts, which I gave to all my friends. I personally put an AMD tattoo on my forehead and wore the shirt...my friend went and started passing tattoos out in front of the Intel booth and a guy literally ran up, pushed him, and started yelling "what the fuck are you doing, you're not allowed over here anymore!"

Good times.

Re:What does AMD's speed matter..

[Anonymous Coward]

There's a bit of a difference between you ordering 50 processors and a OEM manufacturer buying 50 million.

Easy. Pay me.

[hummassa]

I can be as unbiased as it gets.
Especially if both companies give me an equal-valued check.
Send me sealed machines, externally similar cases, preferently ship both together (one company's courier meets the other in UPS's office, they wrap the computers in unmarked boxes). Make only one distinctive mark with a Sharpie pen in one of the cases, give me a week and I'll give you the result, posted in a website:
"The marked machine performed ..."
"I put the unmarked machine in such and such situation and ..."
"Final result and conclusion: the unmarked machine wins."
Someone else can open the sealed envelop and tell the world who is who. I don't care.
(I get to keep both computers.)

VIM

[hummassa]

press [f1] for help. Following text shows:
"""
*help.txt*      For Vim version 6.3.  Last change: 2004 May 04

                        VIM - main help file
                                                                         k
      Move around:  Use the cursor keys, or "h" to go left,            h   l
                    "j" to go down, "k" to go up, "l" to go right.       j
Close this window:  Use ":q<Enter>".
   Get out of Vim:  Use ":qa!<Enter>" (careful, all changes are lost!).
"""

Can'g get much easier, can it?

Re:What does AMD's speed matter..

[a_nonamiss]

There's really no difference. In this case, newegg.com is the "OEM" in question. They order those processors in mass quantities. (I don't think anybody, not even Dell, orders 50 million processors, but let's say 50,000) If they have them in stock, that means there is no shortage, on their order at least.

Yosemite Sam

[P0pinjay]

What is this, an old west Looney Tunes bit?? "Any one of you lily-livered, bowlegged varmints care to slap leather with me?" "Oh yeah? We'll see who'll chicken out first."

Set a budget

[PurpleWizard]

They should do it with a hint of real world attachment to it. Set a budget of a few thousand pounds to build a system and say it has to run these applications. Then devise some ways to load them to see which crumbles first whilst monitoring heat, power and performance over a Le mans style period (24 hours endurance battle)

The price performance might be discernable somehow then.

Is there a doubt?

[MrCopilot]

AMD only since k5-133. Corporate always goes INTEL.

What dooes AMD have to do, Payola?

As MIS, looking to repopulate our server farm, I would be proud to secretely support such a "Marketing Strategy".
Hey it worked for SONY.

Re:AMD could actually lose this one - maybe not

[bunyip]

Vlad,

We went from HP rx5670, 4x1.5GHz Itanium 6MB cache to HP DL585 4x2.2GHz Opteron.

Out of ordeer execution is just speculation on why Opteron is faster, but it was definitely 2x on our app.

I've talked about this publicly, here's one link and a quick search on Google might find more details on what I've been up to :-)

http://www.amd.com/us-en/Corporate/VirtualPressRoo m/0,,51_104_566~91665,00.html

An important point - open source is what let use chase Moore's Law and use the fastest stuff out there. If someone comes along and knocks AMD off the top, we can move there pretty rapidly. We don't need to wait for a bunch of others to port the OS, the middleware, etc.

Alan.