NVIDIA claims top DX9 GPU marketshare spot.

[Evildeus]

Sure Dave, but waivers enable the WHQL certification. As for FP32 on beta drivers not being able to all the element, that's possible.

 

[RussSchultz]

Ignoring all else for the time being....where again did I communicate (through written or oral, or intimation) "properly" cannot include performance?

Oh, no where.

Then why are you insisting I did?

You seem to suggest that by agreeing with Dave H, that I somehow communicated it CANNOT include. Strange, but in rereading what Dave communicated, HE didn't communicate such a thing either.

So where, again, did I communicate that "properly" cannot include performance?

(notice how I changed all my say/said to communicate/d? That isn't the point of my disagreement with your assertion of my written statements)

 

[kemosabe]

I really fail to see what the fuss is with these recent Mercury numbers. According to the second Xbit blurb ( http://www.xbitlabs.com/news/video/display/20030801102103.html ), Nvidia has allegedly lost a whopping 4% of overall market share, mostly by way of lost mobile share. Some assumptions would have to be made to explain how Intel became the main benefactor of Nvidia's losses, but the point is that this report does not indicate that Nvidia has emerged unscathed as some suggest. If you consider that 53 million units were shipped altogether, then Nvidia's shipments went down from about 16.4 million (31%) to 14.3 million (27%) units. That's a drop of about 13% in unit sales in a single quarter! Not to mention that while they maintained desktop market share, it is logical to conclude (as most already have) that they are selling proportionally more low-end/mainstream chips than in the past. Along with their 0.13u yield problems, this would be consistent with their falling profit margins.

My main contention with these numbers relates to how Nvidia has not suffered a more significant revenue hit if unit sales in its core PC business have dropped so much. Increases in XBox sales would be unlikely to compensate for that much. I believe Mercury used to omit the integrated segment from its data (or at least calculated it differently), but seems to have incorporated them this time. I'd have liked to see Peddie's numbers for comparison, but he doesn't seem to have updated his free Market Watch reports since the last quarter of 2002. Regardless, with numbers like this I can't imagine why anyone would raise the spectre of Mercury being an extension of Nvidia's PR department.

 

[Dave Baumann]

Sure Dave, but waivers enable the WHQL certification.

Thats what I said to begin with!

As for FP32 on beta drivers not being able to all the element, that's possible.

The weight of evidence (the fact that current WHQL drivers are only FP16, there are rendering erros in some apps with FP32) would actually suggest its likely.

 

[Evildeus]

Sure Dave, but waivers enable the WHQL certification.

Thats what I said to begin with!

So do i

As for FP32 on beta drivers not being able to all the element, that's possible.

The weight of evidence (the fact that current WHQL drivers are only FP16, there are rendering erros in some apps with FP32) would actually suggest its likely.

Well it could be drivers issues also, couldn't it? Did you test with 44.67 drivers?

 

[demalion]

Ignoring all else for the time being....where again did I communicate (through written or oral, or intimation) "properly" cannot include performance?

Russ, did you read my post?

Oh, no where.

I guess not?

Then why are you insisting I did?

I answered this...

You seem to suggest that by agreeing with Dave H, that I somehow communicated it CANNOT include.

Oh, you did read my post? I guess you decided to build your argument support backwards...by "saying" my support wasn't there, and then addressing the premise itself after having established that...my support wasn't there....by asking me where my support was...
For clarity(?): "asking where my support is" "communicates"/"says" that "my support wasn't there" when you are replying to a post where I proposed I was giving you support.
I guess that's a good way to avoid actually discussing what support I specified...works almost as well as going off on a tangent and ignoring it when someone specifically discusses exactly what you proposed, doesn't it? Is this description of what you are doing what you propose is an "intellectual slight", btw?

I guess I'll just have to wait for the PS 2.0 discussion?

Strange, but in rereading what Dave communicated, HE didn't communicate such a thing either.

Now, do I have to quote my reply to Dave H's post where I already provided this illustration, or can I just link to it? Look for "couldn't", "can't" and other discussion related to what I proposed he was precluding...the bolding and underlining should help you a bit. . If you find them, give the phrases "says he cannot" and "precludes him from" some careful thought, and then consider what I've provided and not provided again.

In case it helps, let me quote Dave H's comment for you, since it appears I'm going to have to go over it again:

FX5200, which we all know is an under-powered card which probably can't run DX9 properly.

And they let you review video cards?

Yes they do. In fact he does a damn good job at it. Don't shoot the guy for having an opinion

Compliance with an API is not a matter of opinion. The 5200 can run DX9 perfectly properly.

(the blue text is what your original quote dropped).

Do I really have to explain this, or are you stuck on the "literal utterance" concept again? If you aren't stuck on the idea that "cannot" has to be in the text, I suggest you focus on the phrase "not a matter of opinion" and consider how it is related to Jerky's statement in the post.

So where, again, did I communicate that "properly" cannot include performance?

You know...I did actually quote you in the prior post when I discussed why I said this. Doesn't this strike you as a slightly silly question?

(notice how I changed all my say/said to communicate/d? That isn't the point of my disagreement with your assertion of my written statements)

Maybe if we were both adhering to a common method of communication, the word change would have helped. I foresee a discussion of what "cannot" means, with unambiguous dictionary and grammar discussion being required as proof on my part, before you'll get past this item, or even address the explanation for it I already provided.

 

[RussSchultz]

And your quote of mine and Dave's does not show that I have said (which for the future you can presume that "said" means wrote and/or communicated so you can avoid filling up the screen with more feigned ignorance and mental musings about what I mean), or agreed with the idea that "properly" in that context CANNOT include performance.

You're the one here with a bit of cognitive dissonance. Yes, I agreed with the idea that I believe the statement "properly" in context MEANS some sort of measurement that does not include performance, but nowhere do I state that it CANNOT include performance.

I have evaluated the meaning in context and came to a conclusion on what I believe it means. How you can bridge the gap to it meaning "I refute any idea or notion that it means somethign different" is beyond me.

So, have we cleared that issue up enough so that we can continue?

 

[Tridam]

The register useage is a problem, but it's not the only one.

I'm talking about the performance difference between running in FP16 and FP32, that shouldn't have much relationship to its compaetitive standpoint.

Ok

You're right. But what could be this reason ? Maybe NVIDIA drop down to FP16 just the time required to make better drivers and application-specific optimisations.

Oh, that could vey well be the case. Even if there are some hardware issues, that also doesn't preclude that they will find a workaround for that in a later release and be able to offer it at full FP32 precision. But that doesn't change the fact that, at present, the only officially released WHQL drivers from NVIDIA for these boards do not meet the specification set by MS - does it?

Of course, NVIDIA has still not any official DX9 compliant drivers.

It's really strange. Does WHQL certification require DX9 compliance ? Was MS checking DX9 compliance when they passed the 44.03?

Again, I think this is the entire point of the DCT tests. Whenever a WHQL submission is ent to MS the IHV's run DCT themselves, and this drops out encrypted logs that are sent to MS. MS then asses these logs and then decide whether they are compliant or not (although there will be some dialogue on areas that are not compliant, as we've seen!).

MS says whether the drivers are compliant or not in regard to WHQL requests. But couldn't WHQL compliance and DX9 compliance be different? IE : DX8 compliant drivers can be WHQL certified.

Of course, but we've also demonstrated that these issues disappear with newer drivers.

From a recent conversation I had with NVIDIA I believe that has been done via shader replacement again. While this may not stick to FM's guidelines, it does stick to NVIDIA's policy of making the IQ the same. Thats said, these later drivers have also shown differences between the reference rasteriser and running on an NV board.

Of course, NVIDIA uses shader replacement in 3Dmark with some newer drivers. But it is not the case with every drivers. Some 'old' Quadro drivers run FP32.

In some shaders I wrote, there are precision issues with 44.03 and not with 44.10 and newer. But of course, I can't say that there are never precision issues. I can just say that I don't see them with the shaders I use.

 

[indio]

Ok here's the beef with Mercury Research . I have never seen a comment or quote from them anywhere, except in news articles that show Nvidia in a positive light . That's why I suggest a LexisNexis search to see if this is actually the case.
Maybe their numbers are correct but is there counting method correct? Is a 5200 really DX9 hardware? Who classifies it as such? The IHV? No one AFAIK has gotten one to actually perform to DX9 specifications. The crux of the situation is that the report should not be taken at face value . Given the current controversies (i.e. compliancy and cheating) and a link (albeit flimsy) can be established between the generator of the report and other questionable analysts.
Mercury Research is quoted VERY often through out many online and print news outlets. They have alot of influence on PERCEPTION. If they go and run around telling the world that the 5200 is the most prevalent DX9 hardware what do you think that will do?
The bean counters will wave this report at the production manager and say "Nvidia has the most DX9 cards according to Mercury Research" When the planning stage of a game starts they will program for these crap cards. The net effect is to drag the majority PC gaming industry down to the feature-set level of consoles

 

[Dave Baumann]

MS says whether the drivers are compliant or not in regard to WHQL requests. But couldn't WHQL compliance and DX9 compliance be different? IE : DX8 compliant drivers can be WHQL certified.

Yes, you can be compliant to DX8, but then you would only be exposing PS1.x caps. Here there driver are testing to the PS2.0 caps bits, but being allowed to run in FP16.

Of course, NVIDIA uses shader replacement in 3Dmark with some newer drivers. But it is not the case with every drivers. Some 'old' Quadro drivers run FP32.

The early NV30 drivers (pre shader replacement) also had FP32 support, and these show the 3DM issues as well.

 

[Tridam]

Of course, NVIDIA uses shader replacement in 3Dmark with some newer drivers. But it is not the case with every drivers. Some 'old' Quadro drivers run FP32.

The early NV30 drivers (pre shader replacement) also had FP32 support, and these show the 3DM issues as well.

What's exactly the issue you're talking about ?

 

[Dave Baumann]

Look towards the bottom of this page.

 

[CapsLock]

This is the first reason why Mercury Research should be doubted as having any kind of credibility: thier very own freaking website:

http://www.mercuryresearch.com/

Please note as you peruse this marvel of slick PR and technological know-how (not to mention enormous substance), that:

1. It looks like a 12 year old did it!!!!!!!

2. It looks like a 12 year old did it-in a couple of hours!!!!!

3. Where the hell is "Cave Creek, Az"?

Googling "Cave Creek" I have discovered that it is a town of an entire 3,728 people (in 1,571 homes), located 25 miles north of Phoenix. And as anybody can tell you, they know thier IT in Cave Creek. Its like Silicon Valley 2, for petes sake.

Please do not confuse this Mercury Research with this Mercury Research, because the second one, even though its based in Romania, looks extremely proffessional and legit. It does provide a nice comparison though.

(I love OUR Mercury Research's contact page especially.)

CapsLock

PS. After noting thier products proven worth and industry recognition, as clearly demonstrated by thier website, I am so ready to plunk down 10,000 USD. (Check out the beautiful order form page.)

Could Nvidia sink any lower than hiring/buying these clowns?

Wow, how come no one noticed this before?

edit: population found higher than first reported, assumed more recent and updated

 

[Bolloxoid]

And they let you review video cards?

And just recently you were in tears complaining how people here here received undue "treatment."

 

[RussSchultz]

Yes, I noticed I was a pack of people repeatedly attacking this guy, burying him in ad hominem attacks and repeatedly dragging up old discussions, until he leaves in disgust.

And, if you hadn't noticed, everybody is completly silent about how I'm going about doing it.

The parallels are SO there.

 

[demalion]

And your quote of mine and Dave's does not show that I have said (which for the future you can presume that "said" means wrote and/or communicated so you can avoid filling up the screen with more feigned ignorance and mental musings about what I mean), or agreed with the idea that "properly" in that context CANNOT include performance.

Well, actually, what I said was Dave H said that, and you agreed with it and defended it, while proposing your assertion of reviewer incompetence that depended on it being the interpretation of his words. Just a bit of clarity, as "refute" indicates you actually "made a case" for the notion instead of just "proposing" it. I don't think saying this the 3rd or 4th time will help, or simply referring to my text, so let me quote and add color for you...

...

First off, I NEVER said that performance is not possible to be included in "properly" in his sentence. Unless "safe to assume" means "not possible"? (I guess I should be prepared for the essay on how it means exactly that!?)

Now we're back to incompatibility with language again.

I provide an argument to Dave H concerning your assertion of incompetence based on (what you proposed as) the usage of "properly" by Jerky, which according to Dave H (at that time) was not a "matter of opinion". This argument by Dave H was predicated on the only interpretation for "properly" being exclusive of performance (i.e., "DX 9 compliance"). You did actually read this discussion, yes? You realize that this post was not a reply to you, and that your response to it placed you in the middle of it, right?

You replied to that post by saying:

Demalion: he (Jerky) specifically mentioned performance in his original statment, so it would be safe to assume that the second part (can't run dx9 properly) does not include performance as a criteria.

'Technically', you "said" "safe to assume" (if you define "said" as restricted to literal utterance)...

And if this is all opinion, can I therefor say that the R300 can't run Dx9 (or even dx7 or dx8) properly because the anisotropic isn't good?

...except it was predicated on proposing that the usage of properly wasn't a matter of opinion, both by this phrase challenging my assertion that it was indeed a matter of opinion, and by agreeing with what Dave H originally proposed. But...you didn't "say" it...you just "agreed" with it, said it was "safe to assume" and defended it.
...

I just quoted what I'm proposing shows exactly what you're saying I didn't provide for you, since apparently just referring to it was not enough. Does that mean that now you can spend your time actually addressing it and showing why your disagreement makes sense?

Alternatively, do I have to specifically discuss the definition of "say"? Some excerpts: "To state as one's opinion or judgment", "To report or maintain; allege".
Question: How are you not 'maintaining' something when you express your agreement with it, defend it, and express disagreement with someone's attack of it? Note, if you think I'm using "maintaining" in a "slippery" way that doesn't fit within the use the dictionary is specifying for "say", you are free to show why...AFAICS I'm picking a usage that removes the connotation of "literal utterance" it looks to me you are thinking of when you disagree with me.
Another set of questions: Do you disagree with my discussion of what Dave H said? If so, why aren't you illustrating why instead of "communicating" that it is irrelevant? Do I have to explain again (as I did in that discussion that I linked to) how Dave H said (at the time) that "properly" CANNOT include performance?

Now, if you propose that you weren't agreeing with Dave H's commentary or defending it, it would have saved us a lot of time if you'd actually addressed my assertion the first time instead of simply pretending it didn't exist and requiring me to "prove" to you what "say" means first.

Otherwise, you seem to be displaying a poor grasp of "say" and "cannot", again seemingly tied to having had to say "say" and "cannot", literally, and arguing over the specific word instead of their meaning. I.e., you'd say that "He precluded the possibility of her being right" never means "He said she could not be right".

You're the one here with a bit of cognitive dissonance.

Well, if my disagreement is simply a "cognitive dissonance", I guess you don't have to actually recognize my reasons?

Yes, I agreed with the idea that I believe the statement "properly" in context MEANS some sort of measurement that does not include performance, but nowhere do I state that it CANNOT include performance.

Let's switch to "maintain", since you so clearly indicated that you aren't tied to the "literal utterance" interpretation of "say".

Russ says: "nowhere do I maintain that it CANNOT include performance".

I'm saying that: Yes, there is somewhere you maintain that it CANNOT include performance. An example is in blue above.
Now that I've repeated the text and highlighted an answer to your question in blue in addition to referring to it after I typed it in the first place, are you capable of a reply other than repeating "nowhere did I state that" while completely ignoring my specific discussion of the words where I maintain you did exactly that?

If you are disputing the word "CANNOT", look at my prior post once again and take the opportunity to follow up on the clues I provided to aid in understanding it. That includes the recommendation of the two phrases to consider.

Alternatively, we could discuss the dictionary again.

I have evaluated the meaning in context and came to a conclusion on what I believe it means.

And responded in disagreement to my discussion of how the full quote indicated something else when it mentioned that something else explicitly. Now we're back to a discussion of "which" and how it indicates the opposite of properly having nothing to do with performance in the full quote. Well, if we're speaking the same language.

How you can bridge the gap to it meaning "I refute any idea or notion that it means somethign different" is beyond me.

I don't happen to believe you, but if it is so, it is so. I guess it would have been shorter to just discuss "which" with you using the dictionary, since my discussion without direct quotes from it didn't seem to help.

So, have we cleared that issue up enough so that we can continue?

No, but we can continue (I assume you mean with the PS 2.0 discussion) anyways. It was your decision to not do so.
You remember me asking you to continue the PS 2.0 discussion, don't you? You know what PMs are, right? I already recommended that you could confine requiring me to prove that English means what I say it does to you, for every particular, to some other location than this thread.

 

[Tridam]

Look towards the bottom of this page.

This problem is strange. Maybe it is not a PS precision related problem.

I've just finished some tests with a NV34. I can't see any precision problem. I check for range and epsilon. With and without partial precision. Everything seems OK and works like it has to work.

 

[CapsLock]

Hey, c'mon now, I am not a 3d graphics expert, true. But did anyone bother to check out Mercury Research's site? Its a complete joke.

Please compare:

http://www.mercury.ro/home_ie.html

and

http://www.mercuryresearch.com/

I am having a hard time believing the above site is at all legit, take a good look at it and you'll see what I mean. Especially when compared to the other one.

Caps

 

[RussSchultz]

Yes, I can read exactly what I said, and what you said for the 4th time, and it still does not amount to me stating/maintaining/communicating that "performance CANNOT be a criteria in properly". It simply amounts to you putting words in my mouth.

He didn't say that, and neither did I. You didn't even begin to assert that until well after my initial response. You seemed to be putting forth your interpretation of what properly meant, so I offered mine and my reasoning. Then, sometime later, you seem to be asserting that I'm suggesting that there was no other possible way to interpret what "properly" means.

Which completely boggles my mind. I said nothing of the sort. I said I don't think that is what he (jerky) meant, not that it CANNOT be what he meant. Matter of fact, I've been pretty much denying it ever since you started asserting it directly.

Its strange that for someone who's so insistant on the correct meaning of words, you're more than willing to conjure meaning into other peoples.

Just like this whole thing over "say". I didn't even begin to suggest that "say" doesn't mean "typed" or "conveyed". YOU DID, as some strange strawman to show that I'm incapable of logical thought because you can't accept that I didn't say what you're saying I did.

 

[demalion]

Yes, I can read exactly what I said, and what you said for the 4th time, and it still does not amount to me stating/maintaining/communicating that "performance CANNOT be a criteria in properly".

The impasses is that you are saying that "cannot" and "say"..."cannot" mean what I'm saying they mean.

The problem, Russ, is that you are simply wrong. I have the dictionary and discussion that seem to prove it. Do you have a response to this assertion other than simply denying it?

Where we are, and have been for pages, is at a point where you are telling me things are not so, without being required to prove that any of my arguments, proving that they are, are wrong in any way. As I described before: arguing by ego, where things are a matter of the world having to suit you without you ever being accountable to ever show "why" you are right. Simply ignoring this description of your actions does not make it go away, or reduce the accuracy of its correlation to them.

It simply amounts to you putting words in my mouth.

You keep saying this while completely ignoring my demonstration of it being the words you put there . I answer your question of "where did I say that?", and you completely ignore it, do absolutely nothing to prove me wrong, and simply say, again, that you did not.

The only way this makes any sense is if you are working on "said" meaning only to "literally utter" something...the meaning of "cannot" and "preclude" and all the other words I've mentioned show this in the actual English language. My proposing this is because the least illogical (but not least wrong) of interpretations, and I've given an explanation of why this is the case. If I'm incorrect, my discussion of it is available, with abundant specifics available for you to prove in error. Extending no effort, besides repetition of denial, to do this is, and have me change what I say is not a "right" you have. I'd point out it isn't one I have, but you seem to demonstrate that you know that by the level of detail you require for explanation and then ignore.

He didn't say that, and neither did I.

Well, I put together a discussion showing that he indeed did say that, and that you did as well. At this point, your behavior has been to consistently (completely so) simply ignore any argument you disagree with. The world (outside your head) doesn't work in such a way where you can just dismiss someone's argument at will without showing some rationale for doing so. This world is outside your head.

It's "too bad" that in this world (outside your head) everybody is capable of being wrong, not just everyone besides you. It's "too bad" that I have an argument that you are completely unwilling to address except by just saying it is wrong, period, and that I don't just agree with everything you say when you do this and abandon what I believe and have gone through the effort to support.

Russ, it's too bad, that to the standard of the objective criteria that's been brought to the conversation, that you are wrong.

But simply saying over and over that things are otherwise doesn't change, prove, demonstrate, or make that proposition "right"...except maybe in your head.

You didn't even begin to assert that until well after my initial response.

No, I asserted that this is what Dave H said in my reply to Dave H, exactly like I pointed out to you. See, when I discuss the word "properly" specifically in that post, and say explicitly that Dave H is precluding Jerky's using properly to include performance..and you tell me I never asserted that Dave H precluded properly meaning performance "until well after your initial response"...you're completely failing to make sense.

Having to dissect English for you to prove this is just an example of the lengths you are willing to go to deny the possibility of your being wrong without the bother of being accountable to reason or language itself.

Jerky: ...FX5200, which we all know is an under-powered card which probably can't run DX9 properly.

Dave H: Compliance with an API is not a matter of opinion. The 5200 can run DX9 perfectly properly.

demalion: Your taking "properly" to mean "compliance" is also an opinion. It is also...an assumption (wrong, thought it could be right in the context of a thread where compliance was the only factor being evaluated) that compliance (in DX 9 usage) is the only allowed meaning of "properly".
For instance, a set of standards could have a minimum fps criteria. "DX 9 compliance" is a specific word usage that excludes that, but only for the specific words used in the phrase. "running DX 9 properly" is not that phrase

Please note that "not a matter of opinion" and "you cannot say that" don't "look alike", but "communicate" the same thing with regard to properly having to mean compliance.

Do I have to go through and replace "compliance" with "standards excluding performance", or does my repeating the (underlined) text where I stated it the first time sufficiently show that your assertion and actuality are incompatible?

I.e., can you even admit error in the face of an unambiguous quote of something otherwise to your statement, or will you just continue to ignore it and state the opposite?

You seemed to be putting forth your interpretation of what properly meant, so I offered mine and my reasoning. Then, sometime later, you seem to be asserting that I'm suggesting that there was no other possible way to interpret what "properly" means.

This statement is based on excluding things I said, and things you ended up attacking when you directly attacked my statement of what was a "matter of opinion". What I was discussing with Dave H concerning what he said could not be matter of opinion is pretty specific. Does your simply not addressing my quotes of you here make it seem to you that you didn't say it?

Which completely boggles my mind. I said nothing of the sort. I said I don't think that is what he (jerky) meant, not that it CANNOT be what he meant.

This is sophistry. Here is why: What you did was attack and disagree with my reply indicating that it could be what he meant. That "communicates" and "maintains" exactly what I propose it does, that it "could not".

Matter of fact, I've been pretty much denying it ever since you started asserting it directly.

That's all you've been doing. Denial does not make a proof, Russ.
Here Russ, if I now say: "I'm not wrong in any particular"...is our argument over? If that doesn't work for me, why should it for you?

Its strange that for someone who's so insistant on the correct meaning of words, you're more than willing to conjure meaning into other peoples.

It's not conjuring, Russ, it's the English language. If you don't agree, I do believe I've demonstrated that you are in error. That's where we are until you actually demonstrate something other than your ability to repeat yourself, like, for example, that any of my support might be wrong. Pick something...start somewhere.

Just like this whole thing over "say". I didn't even begin to suggest that "say" doesn't mean "typed" or "conveyed".

"Typed" tends to mean literal communication, which is exactly what I propose your usage of say is tied to in order to make any coherent sense (though, say meaning more than that is what I'm discussing). So, this statement ends up not making sense.
Translating "typed" to "maintain", as would represent what I did propose you "began to suggest", all you've established again is that you consider repeating denial proof.

YOU DID, as some strange strawman to show that I'm incapable of logical thought because you can't accept that I didn't say what you're saying I did.

How is it my strawman when I'm eager to discuss something else entirely, and when I'm not proposing it as a substitute for another argument? You're the one who insists on perpetuating this simply by ignoring contrary opinion, and you're the one who dropped other discussion to do so. "I'm not wrong" can be a strawman argument, not just "you are wrong".
I certainly don't maintain that this proves anything about PS 2.0, which is why, several posts ago, I addressed the PS 2.0 argument separately. Whose decision was it to discuss this in this thread to the exclusion of that?
Sorry that the word you chose seems so inconvenient to what you use it to say...