One more worm in Intels can

[darko21]

I copied and pasted the snippit below.

It is a shame that the Intel compiler, which use to be almost the no-brainer choice if your primary concern was fast code, is now being coerced into being a marketing tool. Crippling the output for non-Intel chips may mean that some published benchmarks may end up bogusly favouring Intel over AMD, but the cost is that if you want to release fast production code I can't recommend the (unpatched) compiler. There are an awful lot of AMD machines out there!

To Intel: there is a standard mechanism out there (invented by you!) for questioning a CPU as to its capabilities. You should be using that, not checking for the presence of your trademark. I don't expect Intel to support AMD-specific extensions, and I also don't expect Intel to have to test its compiler on AMD CPUs. However, if a CPU states that it can do SSE3 or whatever then I expect the code produced by the Intel compiler to use SSE3 instructions rather than to check first if the chip was made by Intel. It was not acceptable for Microsoft to go out and deliberately cripple Windows under DR-DOS, and likewise it isn't acceptable for you to cripple a product that you sell for not inconsequential sums of money so that it won't perform properly on competitors' hardware.


<A HREF="http://www.swallowtail.org/naughty-intel.html" target="_new"> info here </A>

<A HREF="http://forums.divx.com/groupee/forums/a/tpc/f/511101651/m/635104652" target="_new"> and more here </A>

If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[darko21]

Wow!! no one here found this topic worthy of discusion?

To me this is huge, I can't say it is true but would appear it is. I'm amazed nobody has posted on it.

The ramifications of this r huge IMHO. I don't know if AMD will persue this but the way i am reading this they sure as hell should.

I would have expected this to be a hot topic. I know it's somewhat technical but its not all that hard to comprehend.


If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[theholylancer]

well i would think that intel is gona loose the lawsuit and this will indefinatly make AMD's case (if its ture)

 

[fishmahn]

This isn't the 'smoking gun' it may seem. It certainly seems like this is a blatant crippling of non-Intel CPUs (and I think it is as well), but in court, the Intel lawyers (or at least I would if I were an Intel lawyer - and I'm not even a lawyer) will likely maintain that it was done to <b><i>ensure</b></i> compatibility (which is a crock of sh!t but that's how the courts work), and the probably computer illiterate judge will most likely buy that.

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>

 

[Guest]

well since its their compiler...why not!? Its like when microsoft was asked to ship NetScape with its OS.

I do agree thought thats its not really nice. Sur they should do it as not to give any advantage to AMD(like their compiler making software that runs better on AMD because of certain features intel is lacking) but making it run WORST on the oppposition is not really good practice. But in business you cant be nice!



Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, MSI 6800Ultra stock, 2X30gig Raid0

 

[slvr_phoenix]

1) It's not illegal. It's not great coding, and it's not nice, but it's also not illegal. Therefore it's no smoking gun.

2) Unless I read something wrong, the investigation done by Mark Mackey only goes into the <i>FORTRAN</i> compiler, not their C++ compiler. Besides FORTRAN becoming a long forgotten language, last I knew most people were using HP's FORTRAN. (Not so much by choice, but because DEC bought it from M$, Compaq bought it from DEC, and HP bought Compaq.) I wasn't even aware that anyone actually used Intel's FORTRAN compiler. :O But hey.

3) It's been patched by concerned users, so it's not like AMD folks can't unleash their full potential on the Intel FORTRAN compiler.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[darko21]

I did not say it was illegal, hard to say what a judge would think though. I seem to remember a cross licencing agreement between amd intel and this seems to try and circumvent that agreement. Ok there is now a patch but are sites that test cpu performance like THG using that patch? Might not be technically illegal but might add up through monopoly rules it certainly makes intel look bad once again. I'm sure amd could do the same using 3d-now extensions but I don't think amd has the cloute to pull it off cause they are small potatos compared to intel.

If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[P4Man]

>1) It's not illegal. It's not great coding, and it's not
>nice, but it's also not illegal.

AFAICT, that remains to be seen. If intel is officially declared a monopoly, afaik, its not legal if it can be shown to be a deliberate incompatibility intended only to damage a competitor. I am not saying it IS illegal, but it could be. dont be too sure it isnt.

>2) Unless I read something wrong, the investigation done by
>Mark Mackey only goes into the FORTRAN compiler

Yep, but similar if not identical issues have been uncovered with their C compiler. there was a big article on the subject in C't a year or so ago, where they hacked intels compiler and got significant better SPEC scores on Athlons.

>Besides FORTRAN becoming a long forgotten language,

LOL.. its *by far* the most used language for things like HPC. And for HPC, performance is everything, therefore intels compiler is indeed quite popular. Why do you think AMD funded Pathscale port (Fortan only initially) to AMD64 ? Because it was essential for opterons success.

>3) It's been patched by concerned users, so it's not like
>AMD folks can't unleash their full potential on the Intel
>FORTRAN compiler.

As if anyone runs production code using a hacked compiler ! Not too mention, it might well be illegal to do under the EULA of the compiler.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[endyen]

Intel's "monopoly" does not extend to compilers, so monopoly laws would not apply. Besides Intel has reasonable deniability. It was the guy who wrote the code who made it work poorly on Amds.
The conversation
Paul "Your damn compiler works better on Amd chips than on ours, you're fired"
Coder " Sorry boss, I can fix it"
Paul "Well if it doesn't make Intel chips look better, you will never work again"
Coder (muttering to himself)" he's f..ing mad, well maybe I can make the Amd chip look like it's not way better than the pos P4."

 

[P4Man]

>Intel's "monopoly" does not extend to compilers, so monopoly
>laws would not apply

Not a given. MS' monopoly certainly didnt extend to the browser market during the facts for which is was convicted.

>Besides Intel has reasonable deniability.

They could try and claim it was to ensure compatibility, but courts arent stupid and I assume neither are AMDs lawyers.. so..->

>It was the guy who wrote the code who made it work poorly on
> Amds.

..is total rubbish.



= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[endyen]

>It was the guy who wrote the code who made it work poorly on
> Amds.

..is total rubbish.

Was ment as a total joke.
Have you seen the bad news about opteron share?
<A HREF="http://www.theregister.com/2005/07/25/amd_ten_percent/" target="_new">http://www.theregister.com/2005/07/25/amd_ten_percent/</A>

 

[endyen]

And speaking of worms, looks like Intel's roadmap hit another <A HREF="http://www.theregister.com/2005/07/25/intel_east_fork_delay/" target="_new">wormhole</A>

 

[slvr_phoenix]

I did not say it was illegal

No, but you linked to that forum thread where people there said it was. So I'm just saying that it isn't.

hard to say what a judge would think though.

Not really. It's not even remotely anti-AMD. (Let us not forget the existence of x86-compatible CPUs from the likes of VIA and Transmeta.) It's just pro-Intel. Go figure, Intel's compiler is designed to work best with Intel CPUs. And Intel even fixed any crashing problems. They can easily claim any of the following:
1) Their compilers are specifically for their CPUs.
2) The code is written that way for compatability reasons.
3) It's a bug. Bugs happen. It'll be fixed.

You'd have to have one zealot of a judge to even remotely count their compiler as illegal, and <i>if</i> it ever happened, which is highly unlikely, it'd be appealed and won.

So it's really <i>not</i> hard to say what a judge would think.

I seem to remember a cross licencing agreement between amd intel

A cross-licensing agreement to use <i>some</i> of each other's patents without litigation is <i>hardly</i> an agreement to play nice with each other. It's more like a cold war.

And for that matter I think that their APIs like SSE and x86-64 are open to all x86 CPU developers and not even a part of any cross-licensing agreement.

Ok there is now a patch but are sites that test cpu performance like THG using that patch?

Review websites are generally not even able to use a compiler patch for commercial software because they don't have the source code. The proper question is do software companies use the patch when they use <i>Intel's FORTRAN</i> compiler? **ROFL** I highly doubt there's even a significant amount of software run by reviewers that uses FORTRAN code.

it certainly makes intel look bad once again

How do you figure? That's all part of the game. AMD released x86-64 white papers detailing their API with differences from their own implimentation. Did that make AMD look bad? Hardly. These are just the games companies play.

And it's not like Intel even remotely has a monopoly on compilers. In fact I don't know many people who even <i>use</i> Intel compilers. To my knowledge most either use M$ for C++ and HP for FORTRAN because they're Windows developers, or use GCC because they're <i>not</i> strictly Windows developers. Intel is a rarity.

I'm sure amd could do the same using 3d-now extensions but I don't think amd has the cloute to pull it off cause they are small potatos compared to intel.

**ROFL** Ludicrous. First to do it AMD would have to actually <i>write</i> a compiler. :O Please. AMD doesn't expend those kinds of resources.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[slvr_phoenix]

AFAICT, that remains to be seen.

That's a pathetic argument because it can be said of <i>any</i> court case in the US, regardless of existing law. There's <i>always</i> a probability of the absurd, which is why appeals courts exist. (Aside from zealous judges who ruin the whole judicial system, though those tend to favor bad rulings by religious grounds, not by tech fanboyism.)

If intel is officially declared a monopoly, afaik, its not legal if it can be shown to be a deliberate incompatibility intended only to damage a competitor.

Even <i>if</i>, and you've just laid out the mother of all 'ifs' there, but even <i>if</i> Intel is declared a monopoly <i>in the initial trial in the US</i>, Intel can still very easily ward off it being a deliberate incompoatability aimed at AMD by claiming it necessary for compatability with some version of a VIA or Transmeta CPU to have existed and caused then problems at some point in time. And that isn't even the only defense that they can take there, but just the easiest. And <i>that's</i> even assuming that an Intel x86 compiler is somehow supposed to be responsible for executing code as efficiently as possible on <i>all</i> other x86 processors, which is a ludicrous expectation. And even then there are always apeals to be made.

So the odds of this particlular item ultimately being declared illegal are so darn small that a single electron looks gigantic by comparison, IMHO.

I am not saying it IS illegal, but it could be. dont be too sure it isnt.

In the US court system <i>nothing</i> is ever sure. It's outdated. It's crap. But because no one ever tries to make a change it's all we've got. But as far as probability goes, IMHO snowballs in Hades have a better chance of giving wings to pigs than this has of being found to be illegal.

LOL.. its *by far* the most used language for things like HPC.

**ROFL** Which is quite possibly <i>the</i> smallest possible niche that you could even find for software development. It may get big bucks, but new code there is definately few and far between compared to other software markets.

As I said, FORTRAN is becoming a long forgotten language. COBOL has pretty much already died, as has raw C. I could be wrong, but I believe that BASIC is heading out as well. Python is the new BASIC. FORTRAN has only survived this long because 1) it has been quite useful for purely formulaic programming and 2) a crap load of old FORTRAN code cas been carried forward through the years.

As if anyone runs production code using a hacked compiler !

Are you kidding me? As long as the hack can be proven sound, I don't know of a single programmer that wouldn't. And that includes representation in the US military where my old team did <i>exactly</i> that because of a floating point accuracy bug in Microsoft's FORTRAN compiler. An easy and proven hack to fix a compiler is a hell of a lot better than releasing faulty executables and/or buying a whole new set of compilers for every developer.

Not too mention, it might well be illegal to do under the EULA of the compiler.

Do you even know anything about business software production? First off, EULAs have yet to even be proven legally binding. And second off what is and isn't illegal is often pretty low on the list of things that are worried about. (<i>Especially</i> when it comes to IP laws, but certainly no one even bothers worrying about something as trifling as an EULA.)

They could try and claim it was to ensure compatibility, but courts arent stupid and I assume neither are AMDs lawyers.. so..->

So you're saying that out of all of the x86 CPUs ever produced, there were no compatability problems? Not just with AMD but including VIA and Transmeta as well? Sorry, but convincing everyone that Intel would be guilty beyond a reasonable doubt (as US courts work by) is going to be damned hard with an argument as persuasive as compatability problems floating around.

I'm not saying that I agree with how Intel has coded things, but I am saying that this is hardly a nail in Intel's coffin. It's more like a spec of dust on the coffin lid.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[RichPLS]

"It's ludicrous these mortals even attempt to enter my realm." -Mike Tyson ...

<pre><font color=red>°¤o,¸¸¸,o¤°`°¤o \\// o¤°`°¤o,¸¸¸,o¤°
And the sign says "You got to have a membership card to get inside" Huh
So I got me a pen and paper And I made up my own little sign</pre><p></font color=red>

 

[slvr_phoenix]

**ROFL**

Something like that...

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[P4Man]

>Even if, and you've just laid out the mother of all 'ifs'
>there, but even if Intel is declared a monopoly in the
>initial trial in the US

With around 90% marketshare, I cant see how anyone could claim they are NOT a monopoly. Wether or not is is abusing its dominant position, and therefore acting illegally is something else, but AFAIK, in several court cases like intel vs intergraph, FTC has always said intel has monopoly power. I dont even think intel ever challenged that, let alone successfully.

>Intel can still very easily ward off it being a deliberate
>incompoatability aimed at AMD by claiming it necessary for
>compatability with some version of a VIA or Transmeta CPU to
>have existed and caused then problems at some point in time.

Nonsense. That doesnt come close to explain why the compiler detects the cpu can execute SSEx instructions, yet generates an SSE free codepath. Intel should therefore have to prove AMDs SSEx implementation is incompatible, and can not exectute the SSE code path like Pentium can, which "forced" intel to disable it for athlons. Of course intel will never be able to prove that, as its hogwash.

>And that's even assuming that an Intel x86 compiler is
>somehow supposed to be responsible for executing code as
>efficiently as possible on all other x86 processors

There is a difference between creating a compiler that generates as efficient as possible code, and one that deliberately criples only the competitions product for no valid reason. If they can give a reason, fine, if not, AFAIK it could be considered illegal for a monopoly.

>So the odds of this particlular item ultimately being
>declared illegal are so darn small that a single electron
>looks gigantic by comparison, IMHO

Electrons *are* gigantic compared to gluons.

>**ROFL** Which is quite possibly the smallest possible niche
>that you could even find for software development

The market here is not software development, its hardware. And HPC is a *very* significant hardware market, as it is the only market that will never be saturated no matter how fast your systems are. Half a million high end cpu's (!) in the top500 alone.. and I bet the vast majority of them regulary run fortran code. And the top500 is what, maybe 10% of that market in total ? Probably not even close to that.

>As I said, FORTRAN is becoming a long forgotten language.
>COBOL has pretty much already died,

You are obviously clueless about these markets. Cobol lives forever , its only true that precious little new code is written for it. Wanna bet your bank totally depends on Cobol code ? Fortran OTOH is very much alive and kicking, just not for general purpose code. Its still the n°1 language for scientific purposes and in general HPC, incidently among the most important markets when it comes to cpu power.

Only pc programmers think either language is dead

>As long as the hack can be proven sound, I don't know of a
>single programmer that wouldn't.

Again, PC mentality/ignorance. And obviously you cant proof its sound. Mind you, that requires that the people writing the code *know* about this issue, which is statistically highly unlikely.

>Do you even know anything about business software
>production?

Where are talking about fortan/HPC, not general purpose business software.

>First off, EULAs have yet to even be proven legally binding.

So ? you think CIOs will just assume them to be illegal ? Maybe they also regulary violate licence agreements thinking they might not hold in court ?

>And second off what is and isn't illegal is often pretty low
>on the list of things that are worried about

Riiiiiiight. So you think the typical scenario is that the Fortran programmers or their CIOs KNOW about this issue, are not scared to loose their jobs by creating potentially erronous code in a way that is also likely illegal ? I doubt all that happens in more than 0.5% of the cases where fortan is compiled for the K7.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[slvr_phoenix]

Intel should therefore have to prove AMDs SSEx implementation is incompatible, and can not exectute the SSE code path like Pentium can, which "forced" intel to disable it for athlons. Of course intel will never be able to prove that, as its hogwash.

You seem to know as much about the US court system as you do about business ethics. Intel doesn't have to prove a single thing. Not only is the onus on AMD to prove that Intel violated anything, but further all that Intel has to do is create a reasonable doubt, with <i>or without</i> a single shred of proof for anything. Do you seriously believe that AMD can accomplish this? I sure as hell don't.

There is a difference between creating a compiler that generates as efficient as possible code, and one that deliberately criples only the competitions product for no valid reason.

Yeah, and that difference is whether you're the defense or the prosecution.

Electrons *are* gigantic compared to gluons.

Besides not understanding what in our conversation this applies to, I really wonder why you chose gluons out of all of your choices from particle physics? Why not a neutrino? I mean, for example, when I chose electron that was because I make electrons dance for a living.

The market here is not software development, its hardware.

Actually, the market that I was referring to is that of the humans capable of programming in FORTRAN. It's quickly becoming a forgotten language. Less and less people are learning it.

You are obviously clueless about these markets. Cobol lives forever

**ROFL** No, more like COBOL won't die! If you want to count the number of lines of legacy code that it's becoming harder and harder to find people skilled enough to maintain, then sure, COBOL still lives. That's the thing with legacy code. It can stick around forever, especially once people become afraid to touch it.

But as far as COBOL being a known language, it's dead. There will be no COBOL revival. The few programmers that actually know it will all eventually die of old age, and that will be that. Oh, sure, a few brave (or insane) souls may try to learn it anew, but their job market will continue to shrink until quite simply no one will have a reason to learn it anymore. COBOL's time has passed. It's like a chicken with its head cut off. It's dead. It's body just hasn't accepted it yet.

Fortran OTOH is very much alive and kicking, just not for general purpose code.

I'm not sure what you call alive and kicking. IMHO if you keep cutting off the limbs of a person they may very well be kicking (at least while they still have legs) but they're doing more dying than living. FORTRAN's usefulness is being negated these days, both by faster and faster processors and by newer languages being able to do the same things. It's very rare to meet anyone that can even read FORTRAN code anymore. Which is sad since it's my fourth best language.

Its still the n°1 language for scientific purposes

I'm not entirely sure if that is true or not anymore. It <i>used</i> to be, but if Python hasn't taken that title by now, it probably won't be long before it does. Python is the scientific language of the future.

Only pc programmers think either language is dead

A strange comment. On the same token however, I bet that most console programmers haven't even heard of COBOL. :O

Again, PC mentality/ignorance. And obviously you cant proof its sound.

Wow. Gibberish on both counts. Firstly I've known the embedded controls in an F-16 and ATM software to both use an unsupported compiler patch about a decade ago. I haven't known any developers <i>not</i> to use one if they can prove that it works in their target hardware. So the embedded system programmers are <i>much</i> more apt to use an unsupported compiler patch than PC programmers are.

Secondly, to obtain proof all that you have to do is test it. Obviously the theory of the patch is sound or else the author of the patch wouldn't have figured out how to patch. So the only hard part is testing, which is done easily enough.

So perhaps you should pull your head out of your bum sometime and pay attention.

Mind you, that requires that the people writing the code *know* about this issue, which is statistically highly unlikely.

Yeah, you know a good programmer never looks into why their program is crashing, <i>especially</i> when the code isn't what's at fault. They just accept that it is. I mean it's not like software engineers are highly skilled technical problem solvers that like to tinker or anything. And certainly the plethora of information shared on the internet doesn't help track down these kinds of problems either. No, you're right. No programmers know about this issue.

Hell, I don't even write FORTRAN anymore and <i>I</i> know about it.

Where are talking about fortan/HPC, not general purpose business software.

No, <i>you</i> are trying to <i>narrow it down</i> to FORTRAN/HPC. <i>I'm</i> talking about the software industry in total.

So ? you think CIOs will just assume them to be illegal ?

CIOs thinking that they're illegal? No. Thinking them to be not-legally-binding? Most definately.

Maybe they also regulary violate licence agreements thinking they might not hold in court ?

Violate an actually signed binding legal physical document compared to 'violate' an unsigned, unbinding (and even in some cases in the US <i>illegal</i> because in some states minors by law can't enter into legally binding agreements) non-physical unstored agreement? Yeah, that's almost a sane comparison.

Riiiiiiight. So you think the typical scenario is that the Fortran programmers or their CIOs KNOW about this issue, are not scared to loose their jobs by creating potentially erronous code in a way that is also likely illegal ?

You clearly have no grasp. They'd only use a patch if there were a problem. (Such as crashes as in the Intel FORTRAN 7 compiler.) So it's a choice to either <font color=red>release fully supported code with a known error that they <i>know</i> won't work</font color=red>, or <font color=blue>release theoretically unsupported code thoroughly tested and proven to work a hell of a lot better than the supported code</font color=blue>. Given that choice you'd have to be an idiot <i>not</i> to use the unsupported patch to fix your product. Some may take the 'safe' route by paying the fortune to replace their development software with compilers that don't have such a bug, whether it be a newer version or a different vendor. But that costs a company a lot of money too, so generally isn't done, at least not any time soon. Generally the patch is used first and the new compiler comes a year later.

So no, they're not scared. Firstly, nothing about it is illegal. And secondly it's <i>proven</i> to work better. They're given a pat on the back for a job well done because they solved a problem that could have cost the company money. They get a shiny gold star in their records for being innovative.

At least that's the way that it's gone every time that I've seen it, <i>especially</i> for critical softwares.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[P4Man]

>Not only is the onus on AMD to prove that Intel violated
>anything, but further all that Intel has to do is create a
>reasonable doubt, with or without a single shred of proof for
>anything. Do you seriously believe that AMD can accomplish
>this? I sure as hell don't

Im not betting, but unlike you, I certainly wouldnt guarantee they cant. The proof isnt that hard btw, much of it is linked above.

>But as far as COBOL being a known language, it's dead.

Its not dead, and I doubt we will witness its death during our lives. It has only become invisible, but as long as cobol code is actually running and doing (very!) significant things for corporations like it does today, its about as dead as the mainframes that run it in the first place

(for those that do not know: IBM makes more than $4B per year on mainframes, more than it makes on either x86 or unix/power servers! and despite popular believe that mainframes are a dead, a thing of the past, sales actually keep *increasing* even further).

>I'm not sure what you call alive and kicking.

In the context of this discussion: important enough to have a significant effect on sales of hardware used to run Fortran code (HPC market). No one can reasonably claim it doesnt matter, because it most certainly does. The hardware market for running fortran code is (still) a multibillion dollar business, so its very much "alive".

>Secondly, to obtain proof all that you have to do is test >it.

Oh really.. well, sorry, I forgot exact science like math wasnt your thing, and neither was logic. Testing provides empyrical data, not proof it will always work. It could only proof something if it *doesnt* work. Someone running aerodynamic or stresstest simulations for the next boeig/airbus whatever is not likely to trust a hacked compiler, even after some superficial testing. Especially not when the code runs just fine without the hack, like in this case.

>Yeah, you know a good programmer never looks into why their
>program is crashing

The code in this case doesnt crash. It runs slower than it potentially should/could. So no, most programmers would NOT know there is an hack that can improve performance for AMD processors only, and if they would know, chances are slim they would even bother looking into it, testing it and then applying it. Again, its safe to assume less than 1% of the intel Fortran compiler generated code running on K7/K8 is created using this hack.

>You clearly have no grasp. They'd only use a patch if there
>were a problem. (Such as crashes as in the Intel FORTRAN 7
>compiler.)

Aaah.. well apparently you have been sleeping, or havent even read the article that is being discussed here, but the issue here is *performance*, not buggy code. the intel generated code obviously works on AMD, just not as fast as it should. It would be nice if you kept with the program.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Rugger]

*Lays back in recliner*
*starts eating kettle corn*
Want some Rich?

__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>

 

[Starfishy]

I love it when they do this... but the part I like best is when someone else starts to step between these two guys and gets eviscerated. Where is Porky when we need him?

 

[slvr_phoenix]

proof isnt that hard btw, much of it is linked above.

What part of "reasonable doubt" don't you seem to comprehend? AMD doesn't have to prove that the compiler had a problem. They have to prove that Intel intended to cause harm. They have to prove beyond a reasonable doubt that it wasn't just bad coding or driven by a need for compatability with some unknown non-intel CPU.

Its not dead, and I doubt we will witness its death during our lives.

Just a matter of point of view. You see death as the lack of any code. I see death as the lack of any programmers willing to touch it.

Just because a language exists doesn't mean it's alive. Like so many human languages in our long forgotten past, we have the language, we just don't actually know how to use it. It's what linguists call a "dead language".

The hardware market for running fortran code is (still) a multibillion dollar business, so its very much "alive".

If by alive you mean constantly losing interest in the academic community then yeah, FORTRAN is alive and well. If by alive you mean the number of uses in the real world deminishing day by day as they're replaced by the use of other languages like C++ and Python, then sure, FORTRAN is alive. If by alive you mean dwindling into mere niches (though some indeed very lucrative), yeah, it's alive. Of course it would take a very unique individual to consider any of these a defenition for 'alive'.

Oh really.. well, sorry, I forgot exact science like math wasnt your thing, and neither was logic. Testing provides empyrical data, not proof it will always work.

Funny, last I knew we were talking law, not science. So using a scientific definition of legal terminology is about as logical as, well, you. Your lack logic is as obvious as your skill in law, science, mathematics, and grammar. So wow me some more, big boy. You're just on such a roll right now.

Someone running aerodynamic or stresstest simulations for the next boeig/airbus whatever is not likely to trust a hacked compiler, even after some superficial testing.

Funny, how the US Military fully backed trusted a vendor-unsupported fix to a compiler for their sickeningly accurate and highly critical takeof and landing proceedure software (AFTERPS), and for their F-16's embedded systems. Clearly you know as much about programming as you do about ... well ... <i>is</i> there anything that you know <i>something</i> about? So far it isn't looking too good for you...

Especially not when the code runs just fine without the hack, like in this case.
...
The code in this case doesnt crash. It runs slower than it potentially should/could.
...
Aaah.. well apparently you have been sleeping, or havent even read the article that is being discussed here, but the issue here is *performance*, not buggy code. the intel generated code obviously works on AMD, just not as fast as it should. It would be nice if you kept with the program.

And appearantly when I specifically said the Intel FORTRAN 7 compiler you seemed to be oblivious to the fact that even the very first paragraph in the above linked article states:

<font color=red>A while ago I upgraded my copy of the Intel Fortran Compiler for Linux (to version 7.1.040, build 20040309Z). All seemed well, and the code compiled with the new version passed all of our unit tests. However, when we moved the new executables out to our compute farm they started segfaulting all over the place. What was going on?</font color=red>

Or maybe you just don't know anything about programming, so you have no idea what segfaulting means?

Again, its safe to assume less than 1% of the intel Fortran compiler generated code running on K7/K8 is created using this hack.

I already something similar far previously. Why? Because most FORTRAN programmers are using either HP's compiler or GCC.

But please P4Man, do keep arguing here. You are being ever so entertaining.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[slvr_phoenix]

(Almost) always glad to entertain.

Can I have some of that kettle corn, please? I like kettle corn.

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[darko21]

You'd have to have one zealot of a judge to even remotely count their compiler as illegal, and if it ever happened, which is highly unlikely, it'd be appealed and won.
So it's really not hard to say what a judge would think.

Once again no-one in this thread has said its Illegal, so stop bringing that up please. This is one more worm in intels can of abuse by a monopolist trying to keep the competition down. Maybe a judge will see it that way and maybe he won't but it will be the sum of all parts together (i believe) that will come to expose Intel as a monopoly abusing its power and hurting consumers never mind amd. If Intel is found to be a monopoly abusing its power steps will be taken.

Review websites are generally not even able to use a compiler patch for commercial software because they don't have the source code. The proper question is do software companies use the patch when they use Intel's FORTRAN compiler? **ROFL** I highly doubt there's even a significant amount of software run by reviewers that uses FORTRAN code.

AHHH did u not say <<It's been patched by concerned users, so it's not like AMD folks can't unleash their full potential on the Intel FORTRAN compiler.>>

So once again, are review site like THG using that patched version of Divx to show the real power of the a64?

From the second link above in original post

"""I am informed that the Intel compiler id string is in the header in at least one release of DivX.""""

**ROFL** Ludicrous. First to do it AMD would have to actually write a compiler. :O Please. AMD doesn't expend those kinds of resources.

I did say they were small potatos so unlikly but if they can build a better cpu than Intel I'd say they are a very capable little group


BTW is not the "beyond a reasonable doubt" thing reserved for criminal court. as far as I know amd will just have to demonstrate that the sum of intels actions are abuse of a monopoly I got a feeling they will prove that. I can certainly see it, lets hope amd's lawyers can explain it to a judge or jury.

If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[Rugger]

This is entertainment!
*Passes kettle corn to Fiishy*

__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>