19167 3dmark03 Score on the Hall of Fame (FM)

For developers, to check for bottlenecks when attempting to optimize performance. I would imagine an update to 3DMark should detect when the null PS mode is enabled.

 

Ok, so there is a valid reason for it and its something that can only be enabled by a registry hack or coolbits....I still don't like it.

 

labled Performance analysis. it must be a cheat!!!!!

 

Then you are missing things – I’ve already said it shouldn’t be on as the default, IMO, and I’ve fed that back to ATI. However, this is largely a driver element as opposed to “architecturalâ€; architecturally the capabilities were known to be in RV350 beforehand.

But, texturing capabilities are one such example – we were told that, architecturally, NV40 features similar, full, AF capabilities to GF 3/4/FX however as yet we’ve yet to see it; is it just a driver feature the pattern that its in at the moment? Are we ever going to see the full one? If not, why tell us that it’s capable?

Another couple of examples I can think of off the top of my head:

When trying to establish the arrangement of the ALU’s in the pipeline I queried if they were instruction duplicates of one another, to which the response was along the lines of “nearlyâ€, however it would see that two ALU’s are more of a split of instructions from than they are an overlap.

When talking about the number of colour writes, they’ve yet to mention that, so it seems, only 8 of the 16 colour writing ROP’s have blending capabilities. And this would apparently result the ability of only blending 4 pixels per clock with FP blending. (And I pushed for the fill-rate capabilities with FP filtering and blending without a response)

They have been much more open than they had been before, which was a very welcome relief from the information they gave for the FX series. However, ultimately all this stuff just boils down to marketing – the proof of the pudding is putting it all to the test and see what sticks and what else falls out. The vast majority of the people in the technology briefings didn’t understand what was being told to them – I sat on a table with a bunch of journo’s who all freely admitted that they didn’t understand much of it, nor the questions I asked, but the fact of the matter is that they just don’t need to know, since their readers are only interested in how it plays the games.

The architectural details dolled out, in both cases in many ways, are meant to make things sound impressive and give the illusion of technical superiority which will make games play better – but the numbers will show that and that’s all the majority of readers actually need to know, hence that’s all the majority of reviewers need to care about. The detailed architectural information (not necessarily what “features†it is capable of, but how it does it) is worthless to those that are just interested in playing games; however we, and a number of other publications, have a larger quotient of developers and coder who are interested in the structure and of the pipelines since it gives them an understanding of how their code will work – its far more important to have accurate information here.

 

Hey, it's you/Beyond3D NVIDIA has to deal with... what do you expect?

 

Actually, that should read ".. both have been fairly guarded..." since naturally there are always some elements you will always want to de-emphasize because, as I point out, all of this boils down to marketting in reality.

My suggestions to both IHV's after their editors days was to break out the real tech sessions from the main "editors day", or have concurrent running sessions - let the fewer editors that really see the value in the lower leve technical details get time with the architects and get rid of some of the marketting gloss, whilst the others can have sessions more relevant to their readers.

But, yes, point taken

 

My issue with this is that we have no evidence of Nvidia activating this option. Wouldn't it be obvious by the image quality degradation if it was activated in some common benchmark or game?

If I was a developer I would basically say it's none of your damn business what switches / flags / options I include in my code if they are never activated by the driver during normal usage.

If some guy wants to use it to post some bogus 3dmark score whose fault is that?

 

Hmmm, mebbe I should change the title of me story to "nVidiot cheating, again?"...

 

Just add an update and admit like a man that it was an "ooops".

 

jesus christ people this option has been in the drivers since EARLY 50 forcewares.

 

Well some people like to jump on anything

 

yea some people are slow in the head

 

Why not make him stop posting "news" at EB's front page all together? It's clear he can't handle it.
And congrats, EB made it to W2S again.
What a scoop ey?

Let Hanners post the news on EB and you'll be fine. 8)

 

And some people are slow in the head AND like to jump on anything!

 

like you

 

No, I like pissing off nVidiots. 8)

 

So thats how you bring in news at EB huh? wow...

 

Sometimes? Yup. Sometimes I do. This time I labelled it wrongly, but I still count it as news since there is no bloody reason to include it for developers.

Developers can put in their own switch in the program, unless they're complete morons or something...there is no need for this switch.

 

k just checking