Well, we all got what we waited for.. a review sample of the NV30, covered by several sources (not Beyond3D or 3DGPU yet, unfortunately..), so forums everywhere are amass with queries, generalizations, hypothesis and even a few humorous photoshop images devised by the more clever people in our midsts.
An interesting note- in the months prior, we had discussions as towards how to properly test/review a videocard. There were a number of angles, with Rev/Brian/gamers seeming to prefer some sort of baseline be used when comparing rival products, yet the more tech savvy prefer more apples to apples, technological/synthetic results from which maybe some form of performance/delivery can be derrived. Both can offer value, but can also create debate over decisions made (qualitative/subjective opinion of what is "equal" versus comparing 2x IHV A with 2x IHV B where the two resultant outputs may be totally different).
Interestingly enough, HardOCP and Anand's deviated from previous history and provided some very detailed analysis of image quality. They picked a criteria of AA + AF to build an assumed expectation that people deciding between the 9700 Pro and the GeforceFX will use, and illustrated the target resultant image quality as best they could through subjective commentary, screenshots and zoomed regions of output. I think most people found a lot of value from their reviews, albeit there is going to be some debate on whether the given output is valid/accurate (i.e. post-filter concerns, what may or may not be in screenshots, or whatnot).
Other sources, like hexus and Tom's did blind benchmarks- more true to tradition, and provided commentary with little substance or illustration to tie back to the commentary. This is another form of providing reviews and bases the opinion as a matter of trust to the reviewer rather than forcing the reviewer to provide evidence or substance. This is perfectly valid for sources with a large degree of trust built up from the past (i.e. kinda like folks that "trust" everything John Carmack has to say. If your following of his dialogue in the past 3+ years you feel has never steered you wrong, you might be more inclined to believe his suggestions without substantial proof or reasoning included.)
Now on to the NV30 itself. There has been a large degree of whining, complaining, missed expectations, crushed dreams or whatever else flowing everywhere on the internet. I think a lot of it really forgets past history that has insisted:
1) Reviews only tell half the story. With *EVERY* product release and review sample, it's what reviews DONT tell you which can often times be more weighty than what they do tell you. This is generally why people tend to prefer getting their information from 3, 4 or more sources that are deemed "reliable" in their microcosm of the internet. It's a practice that hopes to "fill in the cracks" from the pieces of various sources to get a better overall view.
I see that with xS antialiasing modes- adding texture sampling holds additional value at a higher cost, yet the results of that cost cannot be illustrated in still screenshots. Yet there has been little to no commentary on whether or not it indeed enjoys the benefits it pays the price for. There is a list a mile long of such points.
2) It has already been stated here the Geforce FX Ultra has no DX9.0 capabilities at this time. That's right- the first NVIDIA DX9.0 card and the review samples are floating around running apparently hacked DX8.1 GF4 drivers. This hasnt been noted by the sites involved, but has been at least somewhat confirmed here. Wouldn't this be an important thing to notate?
3) No single source has given ANY truly subjective, overall testimony to the products resultant "in practice" impressions. As they had only 2 days to hurridly run a suite of benchmarks, this is impossible to obtain. What will hold a ton of value are the commentary of a reviewer that has used this card over an extended period of time, throwing their stack of 3D games and applications at it.. and revealing how things went. How smooth it ran, what quirks or problems they encountered, what effects may really shine in ways no other product may have experienced, etc.etc.
4) On the size, heat, power consumption and other physical concerns- someone really needs to get some form of official NVIDIA confirmation that the final shelf product will be similar in these regards. Although most people feel it's pretty safe to assume the review samples are probably 90% close to the final target retail product, it sure wouldnt hurt to get an official word to end the debate or speculations of a physical design change between now and when it hits the shelf.
Taking all things above into consideration, I don't personally feel we are any closer to "unlocking the secrets of the beast" than we were 2-3 months ago. All I see are the efforts of a number of people that got something last week, burned the midnight oil as hard as they could to get as much tangible and related data posted by monday morning, and with no assurances or official word to the degree of similarity that what they reviewed/labored over has to the final product.
There's a lot to discuss here, and I hope such discussions can remain as untroll-worthy as possible, thanks.
An interesting note- in the months prior, we had discussions as towards how to properly test/review a videocard. There were a number of angles, with Rev/Brian/gamers seeming to prefer some sort of baseline be used when comparing rival products, yet the more tech savvy prefer more apples to apples, technological/synthetic results from which maybe some form of performance/delivery can be derrived. Both can offer value, but can also create debate over decisions made (qualitative/subjective opinion of what is "equal" versus comparing 2x IHV A with 2x IHV B where the two resultant outputs may be totally different).
Interestingly enough, HardOCP and Anand's deviated from previous history and provided some very detailed analysis of image quality. They picked a criteria of AA + AF to build an assumed expectation that people deciding between the 9700 Pro and the GeforceFX will use, and illustrated the target resultant image quality as best they could through subjective commentary, screenshots and zoomed regions of output. I think most people found a lot of value from their reviews, albeit there is going to be some debate on whether the given output is valid/accurate (i.e. post-filter concerns, what may or may not be in screenshots, or whatnot).
Other sources, like hexus and Tom's did blind benchmarks- more true to tradition, and provided commentary with little substance or illustration to tie back to the commentary. This is another form of providing reviews and bases the opinion as a matter of trust to the reviewer rather than forcing the reviewer to provide evidence or substance. This is perfectly valid for sources with a large degree of trust built up from the past (i.e. kinda like folks that "trust" everything John Carmack has to say. If your following of his dialogue in the past 3+ years you feel has never steered you wrong, you might be more inclined to believe his suggestions without substantial proof or reasoning included.)
Now on to the NV30 itself. There has been a large degree of whining, complaining, missed expectations, crushed dreams or whatever else flowing everywhere on the internet. I think a lot of it really forgets past history that has insisted:
1) Reviews only tell half the story. With *EVERY* product release and review sample, it's what reviews DONT tell you which can often times be more weighty than what they do tell you. This is generally why people tend to prefer getting their information from 3, 4 or more sources that are deemed "reliable" in their microcosm of the internet. It's a practice that hopes to "fill in the cracks" from the pieces of various sources to get a better overall view.
I see that with xS antialiasing modes- adding texture sampling holds additional value at a higher cost, yet the results of that cost cannot be illustrated in still screenshots. Yet there has been little to no commentary on whether or not it indeed enjoys the benefits it pays the price for. There is a list a mile long of such points.
2) It has already been stated here the Geforce FX Ultra has no DX9.0 capabilities at this time. That's right- the first NVIDIA DX9.0 card and the review samples are floating around running apparently hacked DX8.1 GF4 drivers. This hasnt been noted by the sites involved, but has been at least somewhat confirmed here. Wouldn't this be an important thing to notate?
3) No single source has given ANY truly subjective, overall testimony to the products resultant "in practice" impressions. As they had only 2 days to hurridly run a suite of benchmarks, this is impossible to obtain. What will hold a ton of value are the commentary of a reviewer that has used this card over an extended period of time, throwing their stack of 3D games and applications at it.. and revealing how things went. How smooth it ran, what quirks or problems they encountered, what effects may really shine in ways no other product may have experienced, etc.etc.
4) On the size, heat, power consumption and other physical concerns- someone really needs to get some form of official NVIDIA confirmation that the final shelf product will be similar in these regards. Although most people feel it's pretty safe to assume the review samples are probably 90% close to the final target retail product, it sure wouldnt hurt to get an official word to end the debate or speculations of a physical design change between now and when it hits the shelf.
Taking all things above into consideration, I don't personally feel we are any closer to "unlocking the secrets of the beast" than we were 2-3 months ago. All I see are the efforts of a number of people that got something last week, burned the midnight oil as hard as they could to get as much tangible and related data posted by monday morning, and with no assurances or official word to the degree of similarity that what they reviewed/labored over has to the final product.
There's a lot to discuss here, and I hope such discussions can remain as untroll-worthy as possible, thanks.
