One more worm in Intels can

[P4Man]

>What part of "reasonable doubt" don't you seem to comprehend?

For there to be doubt that the crippeling is not intentional, one would need arguments for what reason the crippeling was done. You have not given any half plausable reason, and though I trust intels lawyers will at least achieve a more credible attempt, until I have seen it, unlike you, I will not state as fact that nothing illegal is going on. The jury is still out there so to speak.

>Of course it would take a very unique individual to consider
>any of these a defenition for 'alive'.

Lets cut the semantics; the issue was wether or not Fortran performance was still relevant to AMD or Intel, and my point is that clearly it is. Elsewhere in this thread you said that AMD should make their own compiler, and actually, they did, they hired Pathscale to do a 64bit compiler for them. Guess what, they didnt ask them for C++, Python or any other "hot" language, but FORTRAN. You tell me *why* when fortran is as dead as you claim ? Are they idiots, or perhaps would they know what matters in the real world ?

>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

No, we where talking compatibility, proof something works. you claimed some testing would proof a hacked compiler would work, when clearly at best it can give a good indication. Through testing you can only prove a negative, ie, if it doesnt work. But Ill admit that is sematics too.

>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)

unless you can provide some evidence to back up these claims, Ill just dismiss them as hot air, possibly SVFF's. As an argument such claims have zero value. Futher more your example is barely relevant, since you are talking about an entirely different situation, where you produce standard software that can be (and will) be fully tested, debugged and qualified with controlled releases. That is very much unlike the typical fortran situation where code is often created or modifiled and then compiled on the fly by the researchers/scientists/whomever and excessive ammounts of time can not be spent to test the binaries ad absurdum. finally, and most importantly, I hope you are not claiming this unsupported patch was applied to gain a couple percent performance rather than to get around an essential and obvious bug.

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

If you read the article,you should uderstand the issue. The code the compiler generated was not broken, they experimented with optimization flags which gave unexpected results since they assumed forcing SSE support would generate code that would run on athlons too. But using the normal flags, the code ran just fine, even on the athlons, just slower than optimal, and that is the issue we are discussing here. They didnt need to hack it to get it working, they hacked it to get better performance.

>Because most FORTRAN programmers are using either HP's
>compiler or GCC.

or pathscale. and most C developpers use MSVC or GCC, but who the hell cares ? The issue is the intel compiler, not HPs or GCC.

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

 

[HansGruber]

Someone need to revive the SPANISH INQUISITION !

They would have their hands full with Intel executives and engineers, not forgetting sales personnel..

:lol:

<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[slvr_phoenix]

Once again no-one in this thread has said its Illegal, so stop bringing that up please.

1) In your very first post you yourself provided the second link to the forum in which it was claimed to be illegal. If you didn't want to discuss the same topics in that forum then you shouldn't have linked to it.

2) Again, you yourself started talking about what a judge thinks. Well gee, other than legal or illegal, there isn't much to a judge's thoughts now, are there?

3) As long as you keep referencing this compiler issue as a part of (in your own words) "the sum of intels actions are abuse of a monopoly" then you <i>are</i> talking about it being illegal. And when you keep evincing a tendancy towards AMD's side in this, you also keep inferring that it <i>is</i> illegal. You can't infer something so clearly and then not expect someone to argue the other side.

So if you want <i>me</i> to stop discussing the legality aspect, then the first thing that <i>you</i> should do is stop bringing it up. :O

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?

You're really jumping the gun here. First, is any part of the FORTRAN code even affected by the issue? Second, what FORTRAN compiler are the Divx folks actually using? Only after answering these questions can we even ask the question: Does DivX benefit from the Intel FORTRAN 7 compiler patch? And only if <i>that</i> answer is yes can we even ask if there <i>is</i> a patched version of DivX <i>to</i> use. And only <i>then</i> does it make sense to ask if review sites are using it ... if it even exists.

After all, if review sites aren't even using software affected by Intel's FORTRAN compiler quirk, then the A64 <i>has</i> been shown in a fair light. (Well, at least until a review site's bias is taken into account. But that's another subject entirely.)

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

I won't argue that. They're obviously capable folks. Just as they're also obviously a very resource limited company.

BTW is not the "beyond a reasonable doubt" thing reserved for criminal court.

Nope. It applies to all things. That's why it's always the onus of the prosecution to provide evidence of the wrongdoing, and it's always the tactic of the defence to create uncertainty. The defense doesn't have to prove anything. They just have to make the prosecution's side seem weak. Sometimes they use their own proof. Sometimes they just spead enough FUD to cover their arse. This applies in criminal <i>and</i> civil cases.

As far as Intel's actual guilt or innocence goes, I don't doubt that Intel is guilty of pushing the line to its utmost edge. Even Intel readily admits that they push right up to that line. After all, that's not abuse, and therefore not illegal. So I don't think anyone doubts that.

To go further, I personally wouldn't even doubt if Intel had crossed that line from time to time.

The real question however is if <i>AMD</i> can actually <i>prove</i> that Intel <i>crossed</i> the line of legality without leaving any gray areas for Intel to weasel their way out of. And <i>that's</i> where I have doubts of AMD's abilities.

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

 

[mozzartusm]

I was screaming about this months ago. Of course i only got flammed for even bringing it up. As for the what AMD has in store for Intel, from what im hearing they still have some tricks to play on Intel. When it rains it pours.

ASUS P5WD2 Premium
Intel 3.73 EE @ 5.6Ghz
XMS2 DDR2 @ 1180Mhz

<A HREF="http://valid.x86-secret.com/records.php?PHPSESSID=792e8f49d5d9b8a4d1ad6f40ca029756" target="_new">#2 CPUZ</A>
SuperPI 25secs

 

[slvr_phoenix]

unlike you, I will not state as fact that nothing illegal is going on. The jury is still out there so to speak.

I've never said otherwise. I've merely stated my opinion and given my argument for it. Hell, I've even repeatedly talked about probability and appellate court.

But to take thing even further, did I ever say, "It is a fact that"? Did I ever list it as a "given"? Did I list it as "Exibit A"? In case you weren't aware of this, this is a <i>forum</i>. It's a place where people <i>discuss their opinions</i>. So unless stated otherwise, <i>everything</i> said is <i>an opinion</i>. Honestly, how can you even expect people to discuss things with you when you can't grasp such basic precepts?

Lets cut the semantics; the issue was wether or not Fortran performance was still relevant to AMD or Intel, and my point is that clearly it is.

Actually, if you're going to shorten this greatly by cutting so many of the semantics, then you might as well at least do it right. <i>The</i> question was whether or not Intel's FORTRAN compiler performance on AMD CPUs was done intentionally to harm AMD or was uninentional and/or a necessity. <i>Any</i> questions beyond that <i>are</i> just semantics.

Elsewhere in this thread you said that AMD should make their own compiler, and actually, they did, they hired Pathscale to do a 64bit compiler for them.

Wow. Even with facts in black and white you can't get them right. First, I never said any such thing. I said that AMD can't possibly even be in a position to do a similar thing to Intel until AMD writes their own compiler. And second, AMD didn't write a compiler. They didn't even hire Pathscale to. What AMD <i>did</i> do was <i>partially fund</i> Pathscale's <i><b>64-bit</b> port</i>.

You tell me *why* when fortran is as dead as you claim ? Are they idiots, or perhaps would they know what matters in the real world ?

I never said that FORTRAN is dead. I'm saying that it's dying. It's on its way out. The market for FORTRAN is very much shrinking, and will continue to do so. So if you're going to argue semantics even after you say that you don't want to, at least get my arguments straight.

So FORTRAN is alive and well in HPC. I've never argued that. Yippee. It's still dying everywhere else; some markets more than others. AMD funded this specifically to increase their Opteron market share. And this matters exactly to what percentage of the total number of computer users in the "real world" again?

No, we where talking compatibility, proof something works. you claimed some testing would proof a hacked compiler would work, when clearly at best it can give a good indication. Through testing you can only prove a negative, ie, if it doesnt work. But Ill admit that is sematics too.

Actually, <i>I</i>was talking about compatability primarily in regards to critical softwares run on specific platforms. And when you know the exact platforms that you have to support, then "proving the negative" for <i>each and all</i> of those platforms proves the positive as well. It's all about scope.

And hell, for that matter if we leave fields with such defined scop and move into general programming, we've not made <i>everything</i> about probability. Because in general programming you <i>don't</i> have a clearly defined scope of hardware to target. All that you can do is test on as many configurations as is economicly feasable to maximize the probability of the target consumers.

So not only does testing work just fine as proof for embedded markets and similar markets with clearly defined hardwares to target, but it also works just fine in general programming where it's all about probability.

unless you can provide some evidence to back up these claims, Ill just dismiss them as hot air, possibly SVFF's. As an argument such claims have zero value.

To you they have zero value because you don't have any trust in the validity of my word. Others however may feel differently. Of course I cannot prove such things. The military is well known for restricting access to things that aren't even remotely classified or sensitive information.

Futher more your example is barely relevant, since you are talking about an entirely different situation, where you produce standard software that can be (and will) be fully tested, debugged and qualified with controlled releases. That is very much unlike the typical fortran situation where code is often created or modifiled and then compiled on the fly by the researchers/scientists/whomever and excessive ammounts of time can not be spent to test the binaries ad absurdum.

Wait. Now who's making a barely relevant argument? You've been trying to peg this into HPC specifically all along, where hardware platforms are fairly solidly defined and bugs are less acceptable than in general programming, and now suddenly constantly trying to dramaticly leave your own scope and enter <i>mine</i> just so that you can call my example barely relevant?

Do you even realize how irrelevant your whole argument is? Whether we're talking about specialized computing fields or general programming, either way you're arguments are failing. Your logic is flawed. Just admit it and be done with it already.

I hope you are not claiming this unsupported patch was applied to gain a couple percent performance rather than to get around an essential and obvious bug.

Not only did I <i>not</i> say any such thing, but I even explicitely stated <i>repeatedly</i> that this is about bugs and crashing, not performance. Why you would even suggest something like this only shows how desperate you are.

If you read the article,you should uderstand the issue. The code the compiler generated was not broken, they experimented with optimization flags which gave unexpected results since they assumed forcing SSE support would generate code that would run on athlons too. But using the normal flags, the code ran just fine, even on the athlons

I'm beginning to think that you didn't even read that article. You've pretty much got it all bass ackwards. The code that the compiler was generating <i>was</i> broken. I don't think anyone in the world would call code that segfaults not-broken.

Further they experimented with the <i>standardly used</i> optimization flags and found that it was the specific flag to <i>enable</i> SSE support that caused segfaults on AMD CPU's. Only by adjusting the compiler flags to permanently turn off SSE for <i>all</i> processors could they produce safe executables. Not being happy with an all or nothing thing (and who <i>would</i> be happy with permanently disabling <i>all</i> SSE), they produced a patch to the compiler which allowed them to still use SSE optimizations without segfaults for Athlons.

They didnt need to hack it to get it working, they hacked it to get better performance.

They needed it to use SSE. Whether you want to consider fixing something to be able to use significant features of CPUs just getting "better performance" or actually a fix to get it working is your opinion of course. But for someone so concerned with HPC, I can't imagine even just "better performance" being a small thing.

or pathscale.

Right. Or Pathscale. I keep forgetting to include it, but honestly I don't mean to. :O

and most C developpers use MSVC or GCC, but who the hell cares ? The issue is the intel compiler, not HPs or GCC.

The point is that if no one is using the Intel compiler, what difference does it really make either way? Either way almost no one is affected. It's kind of silly to get up in arms over something that matters so little, no? It would be like ranting about a bug in BeOS. Who really cares? **ROFL**

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]

1)In your very first post you yourself provided the second link to the forum in which it was claimed to be illegal. If you didn't want to discuss the same topics in that forum then you shouldn't have linked to it.
2) Again, you yourself started talking about what a judge thinks. Well gee, other than legal or illegal, there isn't much to a judge's thoughts now, are there?
3) As long as you keep referencing this compiler issue as a part of (in your own words) "the sum of intels actions are abuse of a monopoly" then you are talking about it being illegal. And when you keep evincing a tendancy towards AMD's side in this, you also keep inferring that it is illegal. You can't infer something so clearly and then not expect someone to argue the other side.

So if you want me to stop discussing the legality aspect, then the first thing that you should do is stop bringing it up. :O

Nice play on words, if you want to argue that its not illegal maybe you should sign up on that Divx forum cause no one here is arguing it <b>IS</b> illegal. You however are stating it is legal and you are only guessing about that, something you seem to do a lot of. It is now illegal for M$ to bundle window with its media played and have it uninstall able in Europe. Was it 2 years ago? So this compiler thing COULD very well be ruled illegal. We here are saying this might be ruled illegal . I am saying or guessing that this appears wrong morally very wrong. Now a judge and or jury will decide if this along with all the other amd complaints will add up to a monopolist being disciplined. So the real answer is it could be ruled illegal.

You're really jumping the gun here. First, is any part of the FORTRAN code even affected by the issue? Second, what FORTRAN compiler are the Divx folks actually using? Only after answering these questions can we even ask the question: Does DivX benefit from the Intel FORTRAN 7 compiler patch? And only if that answer is yes can we even ask if there is a patched version of DivX to use. And only then does it make sense to ask if review sites are using it ... if it even exists.

After all, if review sites aren't even using software affected by Intel's FORTRAN compiler quirk, then the A64 has been shown in a fair light. (Well, at least until a review site's bias is taken into account. But that's another subject entirely.)

Jumping the gun, the answer is was simple. We don’t know what version Dvix is being used at review sites and how benchmarks are being affected. If Intel has any input into the testing I’d bet my left nut a version that ignores amds optimizations is being used.

I won't argue that. They're obviously capable folks. Just as they're also obviously a very resource limited company.

That is good because looks like p4man showed they did make a compiler or hired people to do it.

Nope. It applies to all things. That's why it's always the onus of the prosecution to provide evidence of the wrongdoing, and it's always the tactic of the defence to create uncertainty. The defense doesn't have to prove anything. They just have to make the prosecution's side seem weak. Sometimes they use their own proof. Sometimes they just spead enough FUD to cover their arse. This applies in criminal and civil cases.

As far as Intel's actual guilt or innocence goes, I don't doubt that Intel is guilty of pushing the line to its utmost edge. Even Intel readily admits that they push right up to that line. After all, that's not abuse, and therefore not illegal. So I don't think anyone doubts that.

To go further, I personally wouldn't even doubt if Intel had crossed that line from time to time.

The real question however is if AMD can actually prove that Intel crossed the line of legality without leaving any gray areas for Intel to weasel their way out of. And that's where I have doubts of AMD's abilities.

You are guessing again, and you know what they say about assumptions.
I was not 100% sure but I looked it up and it would appear I was right. You are making to big a reply gets off topic like the are cpu review site using this code thing.

anyway enjoy


Balance of probabilities
Also known as preponderance of the evidence, this is the standard required in most civil cases. The standard is met if the likelihood that the proposition is true is more likely than it not being true. Effectively, the standard is satisfied if there is more than 50% chance that the proposition is true. Lord Denning in Miller v. Minister of Pension described it simply as "more probable than not."


Beyond a reasonable doubt
This is the standard required in most criminal cases. This means that the proposition must be proven to the extent that there is no "reasonable doubt" in the mind of a reasonable person (usually this means the mind of the judge or jury). There can still be a doubt, but only to the extent that it would be "unreasonable" to assume the falsity of the proposition. The precise meaning of words such as "reasonable" and "doubt" are usually defined within jurisprudence of the applicable country.

The difference between the criminal and civil standards of proof has raised some interesting cases, most notably of O.J. Simpson. Cleared by the criminal trial of murder, the civil trial later ordered substantial damages against him due to the lower standard of proof.

preponderance of the evidence
Preponderance of the evidence is the level of burden of persuasion typically employed in the civil procedure and administrative law. In plain English, the phrase roughly means "more likely than not."


<A HREF="http://www.answers.com/topic/burden-of-proof" target="_new"> Beyond a reasonable doubt is for criminal court </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.

 

[slvr_phoenix]

cause no one here is arguing it IS illegal

Irrelevant. As long as you keep bringing the topic up, regardless of stance (or lack thereof), keep expecting the conversation to exist.

Basically, if you want to stop talking about it <b>then stop talking about it</b>!

**ROFL**

Get it yet?

You however are stating it is legal

Correction, I'm stating my <i>opinion</i> that it is legal. Obviously the courts will ultimately decide.

It is now illegal for M$ to bundle window with its media played and have it uninstall able in Europe.

First off, that's Europe, not the US. You'll note that the US has no qualms about this. Second off, It's still pretty sad that even happened in Europe. And third, those are totally different cases. So if this is the basis of your reasoning, good luck with that.

So the real answer is it could be ruled illegal.

No one has ever argued that. Even I have specifically mentioned appeals.

We don’t know what version Dvix is being used at review sites

Umm ... actually, we do. Any good review site lists the version number.

and how benchmarks are being affected

That we don't know, unless someone takes the time to ask maybe. You could always email DivX if this is a concern to you. They might even tell you.

If Intel has any input into the testing I’d bet my left nut a version that ignores amds optimizations is being used.

You know, the funny thing is that I've seen more review sites that will listen to AMD when AMD tells them to use certain settings or even specific non-standard DLLs than they will listen to Intel. But whatever. That's what comparing against other sites and other products is for.

And honestly, what difference does it even make whether or not something was compiled to run faster on one platform than another? A smart shopper buys their CPU based on comparing benchmark results for what softwares they use. So as long as the review site is using the same software as that which is sold commercially (IE. no cheater libraries or special non-release builds) then you're getting a fair comparison.

Granted, ultimately it may not be fair that Application X runs worse on one CPU than on another for no good reason, but that's a gripe to take up with the producers of Application X, not with review sites.

That is good because looks like p4man showed they did make a compiler or hired people to do it.

Again, AMD didn't actually do either. They <i>partially funded</i> the <i>64-bit port</i> from a software producer that <i>already</i> produced their own compilers and would have done this port on their own anyway.

Balance of probabilities
Also known as preponderance of the evidence, this is the standard required in most civil cases.

God I was hoping that you wouldn't actually look this crap up, but you did, so now I have to deal with yet more idiocy. You're right that I did generalize. Frankly, unless you actually study this stuff, it's just easier to understand that way.

But yes, you're right, in <i>civil</i> cases, the actual probability of guilt to prove is a little more reasonable. It's not so much beyond an unreasonable doubt as it is just beyond doubt. That aside, the plaintif still has the initial burden of proof, and the defendant still can use any number of tricks to create doubt instead of providing actual proof.

And thusly noted, that still doesn't change anything that I said. As far as I know AMD is bringing this to a criminal court, not a civil court. And even if it were a civil matter, AMD still has to prove that Intel did something illegal beyond any FUD that Intel can defend itself with. So far AMD has only circumstantial evidence, and even their own witnesses aren't supporting them. (To the point where Dixons is even considering legal action against AMD for being named by AMD.) So if anecdotal and circumstancial evidence is the best that AMD can come up with, then this is clearly about getting press, not about winning a trial.

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]

Irrelevant. As long as you keep bringing the topic up, regardless of stance (or lack thereof), keep expecting the conversation to exist.

Basically, if you want to stop talking about it then stop talking about it!

**ROFL**

Get it yet?

slvr_phoenix actually now that you have back tracked it into merely your opinion I get it just fine. You know your very first words in this thread was “its not illegal” how many time did you state it as not illegal before changing to maybe its not illegal. You should try writing things like (I don’t believe its illegal) or (IMHO its not illegal) (probably not illegal) at what point did you change it to merely your opinion? I guess the reasonable doubt thing was merly your opinion as well proven wrong BTW. I’ll go into that later on.
When I asked you to stop bringing it up PLEASE I was under the impression you were stating “not illegal” as factual thus the request, as its not factual. So yes slvr_phoenix feel free (now that we know its just one more opinion of yours) to scream about it till the cows come home. I’m just fine with it

Correction, I'm stating my opinion that it is legal. Obviously the courts will ultimately decide.

Nice, very good now you are learning.

First off, that's Europe, not the US. You'll note that the US has no qualms about this. Second off, It's still pretty sad that even happened in Europe. And third, those are totally different cases. So if this is the basis of your reasoning, good luck with that.

I did only mention Europe other wise I’d have stated US. Anyway the US has no qualms YET maybe they will maybe they wont. I brought it up as an example of how things become illegal. Don’t try and tell me that blew right over your head.

No one has ever argued that. Even I have specifically mentioned appeals.

I dunno when reading your first few posts it sure looked like you were arguing that it’s not illegal. Glad that has now been clarified.

Umm ... actually, we do. Any good review site lists the version number.
Well a better way to put it is we don’t know what versions of DivX have the bug.

You know slvr_phoenix when I originally asked that question I really was not expecting an answer to it. You gave a huge answer to a question that could have been ignored (cause its obvious or should be if you read both links) or answered as, no one knows what benchmarks are affected by this bug.

That we don't know, unless someone takes the time to ask maybe. You could always email DivX if this is a concern to you. They might even tell you.

I think it should be a concern for amd as the public reads benchmarks to influence their purchasing decisions. Also be nice for people with amd sytems who are unaware of this problem to get their monys worth when using DviX..

You know, the funny thing is that I've seen more review sites that will listen to AMD when AMD tells them to use certain settings or even specific non-standard DLLs than they will listen to Intel. But whatever. That's what comparing against other sites and other products is for.And honestly, what difference does it even make whether or not something was compiled to run faster on one platform than another? A smart shopper buys their CPU based on comparing benchmark results for what softwares they use. So as long as the review site is using the same software as that which is sold commercially (IE. no cheater libraries or special non-release builds) then you're getting a fair comparison.
Granted, ultimately it may not be fair that Application X runs worse on one CPU than on another for no good reason, but that's a gripe to take up with the producers of Application X, not with review sites.

Good post, but I see this as a much bigger step in the wrong direction.

Again, AMD didn't actually do either. They partially funded the 64-bit port from a software producer that already produced their own compilers and would have done this port on their own anyway.

Sounds like your nit picking to me. I originally said ( 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. ) to which you replied (with the usual rolling on the floor laughing) As if totally impossible,,, well looks like it could have been possible just unlikely.

God I was hoping that you wouldn't actually look this crap up, but you did, so now I have to deal with yet more idiocy. You're right that I did generalize. Frankly, unless you actually study this stuff, it's just easier to understand that way.

Look slvr_phoenix the statement amd had to prove beyond a reasonable doubt that you kept brining up was wrong. I was not 100%, I bought it up nicely because I was under the impression beyond a reasonable doubt was reserved for criminal trials. Civil trial do not need the same degree of proof. Well turns out that is true.
Problem was you answered my question with a big NOPE. You could have admitted you made a mistake or if that is too much for you said alright wrong words now stop nit picking. I looked it up cause you gave a big NOPE implying it was correct to say amd has to prove this beyond a reasonable doubt. Fair enough.


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]

As always, you reason in circles.

>'ve never said otherwise. I've merely stated my opinion and
>given my argument for it.

Oh, so this is not a quote of yours then:" It's not illegal. It's not great coding, and it's not nice, but it's also not illegal." Someone must have hacked your account.

>Actually, if you're going to shorten this greatly by cutting
>so many of the semantics, then you might as well at least do
>it right. The question was whether or not Intel's FORTRAN
>compiler performance on AMD CPUs was done intentionally to
>harm AMD or was uninentional and/or a necessity. Any
>questions beyond that are just semantics.

Nonono, dont cut out your own false arguments. You where the one who brought up the "fact" that Fortan was a dying language.. surely you did so to prove this whole issue was irrelevant (if not, why bring it up?). My point is that is FAR from irrelevant, and Fortran performance is one of the key sales argument in a multibillion/year market. It doesnt get more relevant than that. No semantics there. The issue brought up might well be illegal, and so far, still is relevant.

> They didn't even hire Pathscale to. What AMD did do was
>partially fund Pathscale's 64-bit port.

Talking about semantics.. and just how does this change the argument ? Why didnt AMD first "partially fund" an AMD64 port of their C compiler ? Or Python or any other "hot language" ? Maybe AMD thought Fortan was indeed more important to them.

>AMD funded this specifically to increase their Opteron
>market share. And this matters exactly to what percentage of
>the total number of computer users in the "real world"
>again?

its a <A HREF="http://www.askwebhosting.com/story/871/IDC_Reports_Thirty_Percent_Year-Over-Year_Gain_in_High_Performance_Technical_Servers.html" target="_new">$7.5 billion per year market (2004)</A>. This ammounts to ~50% of the overall x86 server market.

>Not only did I not say any such thing, but I even
>explicitely stated repeatedly that this is about bugs and
>crashing, not performance. Why you would even suggest
>something like this only shows how desperate you are.

No, it only shows how irrelevant even your single (and unprovable) example of someone once hacking a compiler is to the matter at hand. Please stop acting like a braindead.

>The code that the compiler was generating was broken. I
>don't think anyone in the world would call code that
>segfaults not-broken.

if you compile with a flag that forces SSE or MMX, the resulting code wont run a 486 either, would you say the code is broken ? The only thing "broken" was the detection of the cpu capabilities, where indeed you might expect a binary with forced SSE code to run on an athlon, it didnt, which is of course the heart of the issue. The code as such ran just fine, as has been shown (not proven) by hacking the detection routine. But this is semantics again, which I even admitted.

>The point is that if no one is using the Intel compiler,
>what difference does it really make either way? Either way
>almost no one is affected.

And of course you have some numbers to back this up ?

Look, its simple. in your initial reply to this thread you made 3 points:
1) you flat out claimed it was NOT illegal behavour. My point was it might well be, which you now seem to admit (and even claim you never said anything else. Case closed
2) You insinuated Fortran was irrelevant, I showed you it was extremely relevant to AMD, and its at the heart of a $7+B market. Wether or not you want to claim the language is dying, fine, it is however, NOT irrelevant. Case closed.
3) You claimed since it was patched, the issue was irrelevant, which I still do not agree with, as 99% of the users will either no know of the patch and/or be reluctant to use it and/or it might even be illegal to do. So the patch doesnt solve a thing for AMD or the vast majority of its users, it only proves there is indeed an issue. And unless you can provide some compelling evidence this patch is widely spread among Fortran coders on AMD machines, I will stand by this point, but feel free to disagree and live happily in your parallel universe where every programmer out there reads <A HREF="http://www.swallowtail.org/" target="_new">http://www.swallowtail.org/</A> and happily hacks compilers even for mission critical stuff on supercomputer clusters.



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

 

[slvr_phoenix]

slvr_phoenix actually now that you have back tracked it into merely your opinion I get it just fine.

Backtrack nothing. In case you weren't aware of this very simple fact: <i>This is a <b>forum</b></i>. A <i>forum</i> is a place to discuss opinions. And as such, unless stated otherwise, everything said here is an opinion. That is the basic premise of a forum. No one has to say before every single statement, "this is my opinion", because that's already a given. That's how it works.

how many time did you state it as not illegal before changing to maybe its not illegal

First, as already stated, unless otherwise noted <i>everything</i> that <i>anyone</i> says here is opinion. And second, if you can read, you'll see that my very <i>second</i> post in this thread is already talking about probability, judges not always being reasonable, and appeals. So whatever has gotten up your arse is beyond me.

Don’t try and tell me that blew right over your head.

What blows right over my head is how a Microsoft case in Europe with actual hard evidence which even then was a sketchy ruling at best IMHO can even remotely apply to an Intel case in the US with what little evidence existing being circumstancial and hearsay. They're totally different cases that have almost nothing in common.

I think it should be a concern for amd as the public reads benchmarks to influence their purchasing decisions. Also be nice for people with amd sytems who are unaware of this problem to get their monys worth when using DviX..

Then these people should ask DivX. Proper research should be part of any purchase, and if a customer doesn't research, then they have no excuse for making a bad decision. So I really don't get what your point is here.

I originally said ( 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. ) to which you replied (with the usual rolling on the floor laughing) As if totally impossible,,, well looks like it could have been possible just unlikely.

AMD aided an independant 3rd party financially. That's extremely different from having done it themselves. It's not like AMD provided code or anything that could have done that. So to have done it they would have had to bribe the developers and the developers would have had to been happy to ruin their integrity. As if. The probability of it being even remotely possible go down extremely, which is why it is quite laughable that AMD could have done it.

Had AMD written their own code then it would have been easy as pie to have put in something like that. But in this case AMD would have had to (in some way) convince a major compiler company to completely destroy their credability. <i>That's</i> the difference that you don't seem to get.

You could have admitted you made a mistake

I didn't make a mistake. I intentionally lied because the lie is far easier to understand than the truth, especially since to my knowledge civil court isn't even applicable to this situation in the first place. **ROFL** But either way, you're right that what I said was wrong. I've already said as much, without hesitation. What more do you want?

یί∫υєг ρђœŋίχ
<font color=red><i>The Devil himself is good, when he is pleased.</i></font color=red>
@ 194K of 200K!

 

[Guest]

Ohhh tag team on SLVR_ !!!
anyone dare to back him up?

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

 

[slvr_phoenix]

As always, you reason in circles.

When you keep bringing up the same arguments, what else is there? It's not like the answer is going to change any. Stop arguing in circles if you want the circles to stop.

Oh, so this is not a quote of yours then

Honestly, what part of forum don't you grasp? I even explained it to you earlier. You'd think that you and darko21 had never heard of a forum before. Sheesh!

You where the one who brought up the "fact" that Fortan was a dying language.

No, I was the one who gave my <i>opinion</i> that FORTRAN is a dying language.

surely you did so to prove this whole issue was irrelevant

Do you even see the color grey? I swear, everything with you is either black or white. The reason I made that point was part of the process of mitigating the issue of how an <i>Intel FORTRAN compiler</i> affects the whole of AMD as represented by their customers. Proving complete irrelevancy had never been my ambition. Frankly, I'm not even sure how you ever thought it was.

Maybe AMD thought Fortan was indeed more important to them.

I never said that it wasn't. In fact, as I already stated, AMD surely felt it important for increasing Opteron's market share. It was a marketing trick. Beyond that however, it close to useless. The port would have happened anyway regardless. And the market still represents a very small number of programmers.

its a $7.5 billion per year market (2004). This ammounts to ~50% of the overall x86 server market.

Whoop-dee-doo. That still doesn't prove that there are by far less and less programmers learning FORTRAN every generation. It just proves that the niche is lucrative to AMD. Like I had <i>ever</i> argued that?

your single (and unprovable) example

I love your math skills. One plus one plus one equals one, eh? Sorry, I've seen it done <i>three</i> times, one of which was directly in my own department. (The other two I know about because those other departments were asking mine about how to do what we did.)

if you compile with a flag that forces SSE or MMX, the resulting code wont run a 486 either, would you say the code is broken ?

No, because a 486 doesn't have SSE or MMX. But the Athlons that it was crashing on <i>did</i> have SSE1. Therefore, <i>broken</i>.

But this is semantics again, which I even admitted.

Funny how you don't seem to be able to admit when you're wrong though. As you once said "<font color=red>The code in this case doesnt crash. It runs slower than it potentially should/could.</font color=red>" Yet clearly it was crashing, which you don't even argue <i>now</i>.

2) You insinuated Fortran was irrelevant

No, I argued that antyhing in the Intel FORTRAN compiler mattering to the whole of AMD customers is minimal.

You claimed since it was patched, the issue was irrelevant

If that is what I claimed verbatum, then I'm sorry, because that wasn't my intent. If not... My intent is that since the issue is patched, those developers supporting AMD with an Intel FORTRAN compiler (which are extremely few) that absolutely need the speed difference (even fewer) have a working option. It's their choice to use it or not.

and live happily in your parallel universe where every programmer out there reads http://www.swallowtail.org/ and happily hacks compilers even for mission critical stuff on supercomputer clusters.

1) I never said they read that site. I merely pointed out that every programmer I've ever known who has run into a segfault and can't find a reason for it in their code has googled it. In which case, the afforementioned site (amongst many others) would have explained the problem.

2) From my experience the US military is <i>especially</i> concerned with patching mission critical systems. Which would you rather have? A devision by zero bug that shouldn't be possible causing crashes of an F-16's flight systems, or a working product? It's a no-brainer. Even discounting the examples that I've seen, to <i>anyone</i> in such a situation, it's a freaking no-brainer. An unsupported patch that <i>should</i> work is still <i>way</i> better than a <i>known</i>-to-crash product.

یί∫υєг ρђœŋίχ
<font color=red><i>The Devil himself is good, when he is pleased.</i></font color=red>
@ 194K of 200K!

 

[mozzartusm]

I think that there are 3 clear cut cases of OCD disorder lurking around in here.

ASUS P5WD2 Premium
Intel 3.73 EE @ 5.6Ghz
XMS2 DDR2 @ 1180Mhz

<A HREF="http://valid.x86-secret.com/records.php?PHPSESSID=792e8f49d5d9b8a4d1ad6f40ca029756" target="_new">#2 CPUZ</A>
SuperPI 25secs

 

[darko21]

Backtrack nothing. In case you weren't aware of this very simple fact: This is a forum. A forum is a place to discuss opinions. And as such, unless stated otherwise, everything said here is an opinion. That is the basic premise of a forum. No one has to say before every single statement, "this is my opinion", because that's already a given. That's how it works.

WOW covered your azz there, …. I can’t be wrong cause it’s only an opinion even though I never stated it’s an opinion. That’s great Slvr leaves you tons of wiggle room in future debates…. You can state something as fact (just by saying it is so) and then just say it was merely my opinion so I was still correct. How convenient!

First, as already stated, unless otherwise noted everything that anyone says here is opinion.

Is this another one of your opinions?

And second, if you can read, you'll see that my very second post in this thread is already talking about probability

Not in your second post to me. Even so still looks like a feeble attempt at back tracking to me but that’s just my opinion. Something I state as an opinion even though I believe it to be true.

judges not always being reasonable, and appeals. So whatever has gotten up your arse is beyond me.

Ok if anyone in this thread is justified in using ROTFL it’s me to that post….. Slvr I have always stated it’s up to a judge AND or not illegal – legal…. Slvr you are the one who initially claimed it was not illegal then back tracked. Or was that just an opinion? I keep forgetting what you consider fact and or fantasy.

What blows right over my head is how a Microsoft case in Europe with actual hard evidence which even then was a sketchy ruling at best IMHO can even remotely apply to an Intel case in the US with what little evidence existing being circumstancial and hearsay. They're totally different cases that have almost nothing in common.

You are responding….. But not at the same time. You only quoted this “{Don’t try and tell me that blew right over your head.}” when what I said was this……
{{I did only mention Europe other wise I’d have stated US. Anyway the US has no qualms YET maybe they will maybe they wont. I brought it up as an example of how things become illegal. Don’t try and tell me that blew right over your head.}}

So slvr like I said in the post that you choose not to include so much of…...
I did only mention Europe other wise I’d have stated US. Anyway the US has no qualms YET maybe they will maybe they wont. I brought it up as an example of how things become illegal.
Don’t try and tell me that blew right over your head.

Did all that blow over your head yet once again?

Then these people should ask DivX.

What I said was… Also be nice for people with amd sytems who are <b>unaware of this problem</b> Just what percent of the population are aware of this problem in your opinion?

and if a customer doesn't research, then they have no excuse for making a bad decision. So I really don't get what your point is here.

Oh ok so if a customer is mislead or does not know what we know that’s ok with you? Slvr I got a car I’d love too sell to you.

AMD aided an independant 3rd party financially. That's extremely different from having done it themselves. It's not like AMD provided code or anything that could have done that. So to have done it they would have had to bribe the developers and the developers would have had to been happy to ruin their integrity. As if. The probability of it being even remotely possible go down extremely, which is why it is quite laughable that AMD could have done it.

They could very well have done it and that was the point unlikely as it was. Just like intel pays off benchmark software like baptco. The real issue was getting Dvix not to bitch about it and implement it, costly I’m sure…. AMD clearly could have just would be they would have no legs too stand on TODAY if they had.

Course this is just an opinion…. So I get a get out of jail free card courtesy of you.

Had AMD written their own code then it would have been easy as pie to have put in something like that. But in this case AMD would have had to (in some way) convince a major compiler company to completely destroy their credability. That's the difference that you don't seem to get.

Had AMD paid enough…. code would be written to whatever the hell AMD said! That is the difference you slvr don’t seem to get.

I didn't make a mistake.

I say you did!! You can’t argue with me,,, as you admitted anything you say is an opinion at best. I say you did make a mistake.. and I say that is a fact! wile you slvr can not produce a fact. You can only guess. How can you prove me wrong when you can only guess?

I intentionally lied because the lie is far easier to understand than the truth

Oh my god, how too answer that. Well I guess everyone reading this is thinking WTF were you talking about….. So slvr now its an opinion and when that opinion is proven wrong you will say you intentionally lied? Am I getting this?

Tell you what slvr don’t ever lie to me again. I feel I am quite capable of understanding the truth.

You ok with that?

especially since to my knowledge civil court isn't even applicable to this situation in the first place.

Me neither, but that is no excuse for admitting you merely made a simple mistake.

**ROFL** But either way, you're right that what I said was wrong.

Yes this is true, but I don't get why you are always ROFL, should be humbling really.

I've already said as much, without hesitation. What more do you want?

I want a true debate and no petty excuses. If you make a mistake I'd like you to own up to it. not answer with a big nope and 80 words of garbage. am I asking too much?

Slvr look if you cannot debate without lame excuse about it was only an opinion after the fact or I intentionally lied therefore I was not wrong.... Then I don't care to debate with you.

I got a feeling this is my last post in this thread but then again that's only my opinion or maybe I am intentionally lying. Go figure.

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.