RussSchultz said:
Sorry demalion, you're not going to drag me into a point by point argument.
Well, that's a bit new. Is it a virtue not to have points, or simply not to discuss them?
They end up being an argument in semantics (as this one has already become).
No, it is an argument of meaning. Semantics is a "bad word" because language meaning can obfuscate the meaning behind the language. It is also diversionary, because my "semantics" are directed at discovering meaning, not hiding it. You attempt to circumvent proving that I'm actually trying to hide something by using the label "semantics" instead of the equally valid word "meaning" because "semantics" is "automatically" bad.
Oh no, I'm doing something "bad" again.
However, I will group some things together with a common response:
Actually, I just wanted you to answer my direct question, as I could have replied briefly to that answer and a clear statement of reasons behind it.
With identical output. Which demonstrably fits ATI's situation with 3dmark, but not nVidia's.
For ATI. His description that fit nVidia's demonstrable actions in GT 4 was categorized as "completey indefensible", and the one that fits ATI's actions are "grungy".
So what? Once again, it doesn't matter one goddamn bit what NVIDIA did, with respect to his complaint.
To John Carmack, it certainly does seem to matter "one goddamn bit".
Because he specifically also discussed something that fit what nVidia did, and gave it a distinct and different description. It is irrelevant to you, but it is not irrelevant to other people, including Catalyst Maker. Your criticism of his commentary depends on ignoring that the people he cited have clearly demonstrated that they don't agree with you.
The assertion that the distinction between them doesn't matter regarding CM's complaints about web sites doing what I described is pretty amazing to me.
He was bitching that people were calling ATI cheaters when they did nothing wrong,
Actually, he did specifically say
sites, hence why I kept mentiong "sites" in my replies. I've already discussed the quote specifically with you.
and called upon JC/Sweeney as proof that they did nothing wrong.
No, he called upon JC/Sweeney to say that their optimizations were legit, because CM's statement are perfectly valid when "what" is talking about the optimization itself, though you seem singularly unable to recognize that it is remotely possible that his prior usage of "what" can possibly mean that.
JC further went on to
separately criticize that the mechanism of introducing the optimization was "grungy" in a thread about the 3dmark cheat.
Read once again what he posted, exactly:
wow im lost for words. I dont understand why certain sites still accuse of us cheating. Both Sweeney and Carmack have said what we have done is legit.
Finally, some resolution in sight...nevermind that I already addressed this in the first reply. Let's spell things out.
Did Sweeney say what ATI was doing is legitimate? NO.
when avoiding talking about benchmarks said:
Therefore, any code optimization performed on a function that does not change the resulting value of the function for any argument, is uncontroversially considered a valid optimization. Therefore, techniques such as instruction selection, instruction scheduling, dead code elimination, and load/store reordering are all acceptable. These techniques change the performance profile of the function, without affecting its extensional meaning.
When CM talks about "what", he is clearly validated at the very least as far as the substance of the optimization itself by what Tim Sweeney said, since we have it clearly demonstrated that the shader output is the same for ATI's optimization.
Is that the whole of the picture? No, and we'll cover that when we see a comment by someone not adhering to Rev's stipulation of talking only about their own code in a game. But it is also a distinct part of the picture that you are persisting in saying does not exist, so you can then say CM's commentary is him portraying himself falsely as a "martyr" because there isn't anything valid he could be referring to. It is clear that it is not different than what Tim Sweeney is talking about, however, despite your assertion to the contrary.
OK, let's cover Carmack's statements in a thread about the 3dmark cheating now.
John Carmack said:
Rewriting shaders behind an application's back in a way that changes the output under non-controlled circumstances is absolutely, positively wrong and indefensible.
Well, he seems to give a damn about the distinction between what nVidia did and ATI did, as this does not fit what what ATI did but does fit what nVidia did. Can you recognize your assertion to the contrary is false?
John Carmack said:
Rewriting a shader so that it does exactly the same thing, but in a more efficient way, is generally acceptable compiler optimization, but there is a range of defensibility from completely generic instruction scheduling that helps almost everyone, to exact shader comparisons that only help one specific application. Full shader comparisons are morally grungy, but not deeply evil.
Oh my. In combination with him specifically mentioning
sites, when there are sites that are
attacking ATI's "cheats" while
defending nVidia's, his perception of this legitimizing what ATI did doesn't exactly seem to require him trying to "falsely portrary himself as a martyr", does it?
You are dedicating yourself to the proposition of vilifying him, and you are ignoring any inconvenient distinctions or questions that get in your way.
The best endorsement either one of them gave was "morally grungy"
If you ignore that what ATI did was "reordoring the shader to do exactly the same thing", you can't validate that Carmack's label of "morally grungy" applies, since that description is part of what he said was "morally grungy".
If you don't ignore it, you also can't ignore that Tim Sweeney specifically said it was "acceptable".
Your stance depends on switching between ignoring these two assertions, as sticking to either completely invalidates the statement you just made and the other places you repeat it, and is also based on ignoring that Carmack draws any distinction between "morally grungy" and "absolutely, positively wrong"
shock
and that that distinction might have some slight relevance to CM's feeling vindicated by his words.
Sticking to the first is demonstrably fallacious based on comparing output, but if you wish to do so, my reply to that could be much briefer.
Sticking to the second validates CM's commentary to the degree that your vilification no longer has the support you assert, because it clearly shows that Tim Sweeney's label of "acceptable" is directly applicable to ATI's optimization, and only leaves the question of distinction between JC's "morally grungy" and "absolutely, positevely wrong", and leaves you proposing that it is only due to trying to portrary himself as a matryr that CM considers the combination of statements as any vindication that ATI is being treated unfairly by being the primary focus of sites' criticism (nevermind what was supposed to have happened on Tech TV
)
If my proposition is false in your opinion, state why you think so, don't just state that you "don't want to be dragged into" a discussion that actually has some relevance besides "I want to believe this, so it doesn't matter what you say".
, the other avoided the topic of what ATI did entirely and focused on a generalized response to "ignoring benchmarks, do you mind if IHVs optimize your code"?
No, it did not avoid the topic of what ATI did, it specifically addressed what ATI did while ignoring discussion of the applicability to benchmarks. We know Sweeney is talking about what ATI did, because he also has a contrasting statement "Optimization techniques which change your function into a function that extensionally differs from what you specified are generally not considered valid optimizations."
Why do I say that, you might ask?
OK, please take a moment to consider your statements: You are proposing that since "nVidia" and "ATI" were not explicitly mentioned, these statements have nothing to do with either, and that it is
completely irrelevant that these programmers made these distinctions that "happen" to exactly fit the observations about what ATI and nVidia did with regard to 3dmark.
I call that being intentionally obtuse, as I can't see what else could explain that proposal on your part at this point.
Unless, of course, you say that what futuremark exposed with regards to ATI was a generalized optimization that broke when futuremark changed something.
Based on the above fallacy, you again repeat the premise that the
only validity in the optimization would be in it being general in the first place, despite Tim Sweeney's comments
directly contradicting that.
The issue of generality is applicable to 3dmark 03 specifically, and is because it is a benchmark, and not because of the optimization itself (in the case of ATI). So
regardless of whether the the optimization was general or not now, if the
exact same optimization was generally applied in the future it would be completely legitimate.
All your statement about it possibly being a "generalized optimization that broke when futuremark changed something" would do is remove the stigma of "morally grungy" that Carmack provides right now, because then
it wouldn't have been application based detection and simple shader replacement in the first place (
this is the part that is "morally grungy",
not the optimization itself), just a fragility that got exposed by Futuremark's efforts to explore nVidia's methods of cheating.
Feel free to doubt that it was a fragility, but that doubt does not justify your vilification of CM's comments, and you depend on fallacies built on that doubt to do so instead.
If thats the case, why the hell are they removing it instead of fixing it?
Personally, I think it could conceivably be either because it was application specific shader replacement, or that their general case optimizer is fragile. With what we know of the R3xx architecture, and the presumption that there should be a general case optimizer, both possibilities seem feasible at this time. The first seems more likely, which is why it is your vilification of CM that I am criticizing, not the asking of this question, or even the label "cheat" when it isn't being used to support demonstrable fallacies (...like...you are doing with your vilification of CM and his reference to TS/JC)
Of course, this entire discourse will be lost because "i'm an nvidiot" and then somebody else will harp on "What NVIDIA did was much worse".
Well, this says that I am "harping" on "What NVIDIA did was much worse", which is not true. Yes, that's semantics, because that meaning was not as directly stated as possible due to your language usage. It was your language choice that prompted me to clarify that, instead of just responding to a quote of it. Just a friendly of example of how semantics just might not always be "bad".
Anyways, to respond to that: I am pointing out that "What NVIDIA did was much worse" is relevant to why your attack on CM is fallacious, because it is a distinction that happens to be related to the JC/TS commentary you seem to be deliberately misconstruing.
No fucking shit its much worse. But it doesn't change the fact that this guy is playing the martyr when they're not innocent,
Oh, it's a fact? Why didn't you say so at the beginning? Why'd we bother to have a discussion ?
(if they are innocent then they're stupid for removing the general case optimization.)
Yet again, I am pointing out your repeated fallacy. CM not being deserving of vilification does not depend on the optimization being general case, but on the optimization performing exactly the same calculations and providing the same output, and nVidia having done much worse is directly relevant to that distinction, because it is exactly the
way they did much worse with regards to shaders.
By simplifying things to either "completely innocent" or "deserving vilification", you "automatically" "justify" your condemnation of CM (by saying that there is no "what they did" that could make his statements valid), and it is your persistence in making that simplification that necessitates discussing the meaning to point out why your simplification is invalid (because, quite simply, JC/TS do talk about a "what ATI did" that does lend validity to CM's complaints about what websites are saying).
I'm sorry bringing up points is so offensive to you, but I happen to not believe simply circumventing recognition of other's opinions is somehow better. To avoid necessitating that you get your hands dirty with "semantics", let me state that more clearly: Yes, I'm asserting that you are circumventing recognizing that CM might have a clearly valid reason to consider the statements that JC/TS made as lending credence to ATI's 3dmark optimization, and that your protestations and logical fallacies attempt to do the same with my statements...look above for the reasons behind that assertion).
Russ, maybe you could answer the question I directly posed to you twice now if you are really trying to have a brief discussion? Hopefully, I can successfuly only have to reference this post from now on instead of just repeating everything in it over and over.
