Questionable Quad SLI review at THG

[Rollo]

I can't figure out what they were hoping to test? CPU?

1. 0X8X, and for the most part 4X8X benchmarks have no place in a Quad SLi review. (25X16 4X8X the one exception I can think of)

2. 8X? Why not 16X High Quality? The cards are more than potent enough to run it these days?

3. 91.29 drivers? How old are those? The 91.37s were deemed the beta quad drivers, and there were other revisions between 91.29 and those? Not to mention the 91.45 WHQL Quad driver were released two days after this article? I realize it takes time to write an article, but what's the point at all if you're going to use months old drivers on a product as new as quad?

As a quad SLi user who has been enjoying performance no dual card solution can provide for over a month now, I feel your article does a grave disservice to readers.

People wanting a recent useful review of where quad is would do better to look here:
http://www.firingsquad.com/hardware/geforce_7900_7950_gx2_quad_sli_update/

 

[Heyyou27]

If Tom's was using 91.29 then that is a problem, but other than that I didn't see anything too outrageous in the review. By the way, what resolution do you play on with your Quad SLI system?

 

[asdasd123123]

Well, one of the reviewers called another NV-fan, and I quote, "Lol. Look at the Nvidiots come out of the woodwork." (Cleeve)

So perhaps they are in fact ATi fanboys, hmm?

 

[TheGreatGrapeApe]

2. 8X? Why not 16X High Quality?

Probably because nV doesn't have HQ AF (regardless of driver revision).

3. 91.29 drivers? How old are those? The 91.37s were deemed the beta quad drivers, and there were other revisions between 91.29 and those? Not to mention the 91.45 WHQL Quad driver were released two days after this article? I realize it takes time to write an article, but what's the point at all if you're going to use months old drivers on a product as new as quad?

The point is nV didn't give much lead time before they were announcing official support for Quad SLi and releasing the recent drivers that caughrt most people by surprise.

As for quad, meh. :?

 

[cleeve]

Well, one of the reviewers called another NV-fan, and I quote, "Lol. Look at the Nvidiots come out of the woodwork." (Cleeve)

So perhaps they are in fact ATi fanboys, hmm?

Do your research, son. I've also called out more than my share of fanATIcs.

I can't stand any kind of fanboy.
Crappy Ati drivers from 4 years ago and the Geforce FX series are tired, obsolete arguments. Anyone who regurgitates marketing crap and old news to diss one company over the other is going to get an earful from me, I don't care what side they're on.

I review today's videocards, and both Ati's and Nvidia's drivers and hardware are fantastic for Windows XP gaming.

So I guess as long as fanboys from both sides hate me, I'm doing allright. 8)

 

[Gary_Busey]

You wanna borrow my shotgun? It has two barrels, one for each fanboy.

 

[Rollo]

Errrr how did this degenerate into a discussion of "fanboys" ?

This post has nothing to do with brand, and everything to do with whether or not reviews using very old drivers and settings far beneath what the hardware is capable of are worth publishing. (and if they do more harm than good)

http://reviews.cnet.com/Nvidia_GeForce_7950_GX2/4514-8902_7-31901901.html

CNET used the 91.29 drivers for a review of the GX2 they published JUNE 5th?
So the drivers are over two months old?

The FS review I linked to that used the beta quad drivers released last months said there were great improvements in them, and now there are WHQL drivers.

Do you really want THG posting articles with 10 week old pre- beta drivers saying "It's best just to wait. This stuff doesn't work very well" when the "waiting" has essentially been done through the passage of time and Quads performance and stability probably bear little resemblance to the article?

As far as the settings go, I would say the same about dual SLi or Crossfire-
0X8X may be a valid test for a GeForce 2, those settings are so far beneath these rigs that differences are largely irrelevant. Who spends $1000 on video cards not to use AA?

I play at 19X12 8X16X High Quality. As the FS benchmarks show, Quad Sli can easily handle that, and that is where it differentiates itself from the dual card sets- at 8XAA.

For those who think 4X is "good enough", check out this comparison at Rage3d of the 4X to the 8X modes:
http://www.rage3d.com/reviews/video/bfg7800gtocsli/index.php?p=4

(takes a bit of time to load)


For the guy that says nV "has no high quality", keep telling yourself that to justify your purchase, I guess.
I've owned a X1800XL and a X1900AIW, and while the angle dependent AF can be a bit better, it depends a lot on the game you are playing, and your monitor. In any case, this "great AF" didn't help ATI get more than 13% of the performance DX9 market in the first quarter of 2006, so people just don't seem to care about it?

 

[cleeve]

Errrr how did this degenerate into a discussion of "fanboys" ?

You can thank asdasd123123 for that, when he implied underlying motives. Had to respond to it, couldn't let that one slip by for obvious reasons. Sorry it had to be in your thread.

This post has nothing to do with brand, and everything to do with whether or not reviews using very old drivers and settings far beneath what the hardware is capable of are worth publishing. (and if they do more harm than good)

I didn't do the review so I can't comment on that directly. I can say that THG is a really busy place, sometimes reviews take a while to get posted after they're completed. That's the reality of the business, unfortunately.

If the new drivers have changed the results to the extent you think they did, this is the place to mention it and a follow up article may be considered....

 

[Rollo]

The problem is when you're reporting on a new technology it does the readers a grave disservice to use months old drivers.

For example, check out the release notes of the 91.45 drivers:
http://download.nvidia.com/Windows/91.45/91.45_ForceWare_Release_Notes.pdf

The guys at Hothardware and HardOCP have articles out about the 91.45 quad SLi experience one day after it's release, presumably THG had access to the driver as well?

If not, the 91.37s have been out several weeks?

This article should really be pulled- if someone forgot to post it a couple months back when those drivers were current, why should THG readers be mislead?

For that matter as nVidia themselves pretty much recommend 19X12 8X16X as the starting point for Quad, why give people it's not a good solution at these low settings any card can run?

As noted, the 25X16 4X8X benches are the only ones where Quad had a chance to shine, and even that was removed by the use of pre-beta drivers.

 

[Action_Man]

Aren't you that fella that goes around trolling forums for nvidia or AEG or whatever the hell its called?

 

[anton]

I don't know how any one of you people could argue what Rollo is saying. If you bother to go and have a look at the site he has linked to, you will see that there is an undeniable, substantial performance gain from a quad setup.

Clearly, these cards are so powerful that for current games to actually stress them, you need to run them at insane levels. Running them at lowish settings is like never taking your ferrari out of first gear. It's going to go the same speed as your golf, and will use more juice.

This is one of the most dissappointing reviews I've ever seen from THG, mainly because it shows a real lack of understanding of what they are trying to review. I really think that they should include a few untra-high benchies with the current driver set and see if their conclusion changes.

 

[anton]

So I see THG have just published an article explaining why the results were poor, but without benchies. Seems like they realised that they had misrepresented the target market in their previous test. 8)

It will be interesting to see what they come up with, when they go into the highest quality realms, which is clearly what quad is designed for.

Hopefully they will also be mating the quad with an overclocked 6800 extreme, because that's where they'll really start cooking. 4 of these cards together are just way too powerful to be used with something like an fx60 except at crazy resolutions.

 

[Rollo]

Aren't you that fella that goes around trolling forums for nvidia or AEG or whatever the hell its called?

I guess that depends on your perspective.

It's true I'm a member of nVidia's user group.

Do you consider it "trolling" to point out that the information contained in the review is misleading due to the age of the pre-beta drivers and settings used?

I don't normally post here, but saw this review referenced as evidence Quad SLi has no value. As a quad SLi user, who has seen other reviews contradict this, I thought I owed it to the community to ask THG if they thought this review should stand as is.

As a person who's done a lot of benchmarking, I've seen how it can be difficult to get to the relevant settings to show what a card is or is not worth. The best example of this with me in recent times was when I compared 6600GTSLi to 6800GT single.
The 6600GT SLi often had similar performance at 4X8X, but really kicked butt at high res 0X16X, which to some people might have value. The only site that mentioned this was Firing Squad.

In any case, I'd be interested in Darren's thoughts on what I've said here.

I think Quad SLi established a new set of guidelines for review due to what it does well and what it doesn't.

If THG can agree that a lot of us 24" and 30" LCD owners would like to see 8XAA benches, and the 20" crowd might be interested in 16X, I've accomplished what I hoped to.

 

[Rollo]

So I see THG have just published an article explaining why the results were poor, but without benchies. Seems like they realised that they had misrepresented the target market in their previous test. 8)

It will be interesting to see what they come up with, when they go into the highest quality realms, which is clearly what quad is designed for.

Hopefully they will also be mating the quad with an overclocked 6800 extreme, because that's where they'll really start cooking. 4 of these cards together are just way too powerful to be used with something like an fx60 except at crazy resolutions.

Excellent. Kudos Darren!

 

[TheGreatGrapeApe]

For the guy that says nV "has no high quality", keep telling yourself that to justify your purchase, I guess.

HAs nothing to do with justifying purchases it has to do with reality and truth. nV has no equivalenet level of HQ AF, it's a hardware limitation (just like FP16HDR+AA) and no magic drivers will ever fix that. Sure there are work arounds but they all involve the CPU to be equal, and thus irrelevant. So enough with your BS rhetoric. At least get your requests correct if you want to start asking for HQ tests.

I've owned a X1800XL and a X1900AIW, and while the angle dependent AF can be a bit better, it depends a lot on the game you are playing, and your monitor.

Just like everything else, there are extraneous factors, but we're talking about head 2 head so those other things you want to toss in there to make yourself feel better really aren't relevant.

In any case, this "great AF" didn't help ATI get more than 13% of the performance DX9 market in the first quarter of 2006, so people just don't seem to care about it?

What does market share have to do with this? But people obviously care or else it wouldn't keep coming up in reviews;
http://www.hardocp.com/article.html?art=MTEwMCwzLCxoZW50aHVzaWFzdA==

"In the first screenshot you can see the advantages of ATI’s high quality anisotropic filtering option. This is a highly zoomed in screenshot, as HQ AF can be a fairly subtle change at level 4X AF. As the pathway angles, NVIDIA’s angle-dependent AF just can’t keep up. ATI’s high quality anisotropic filtering mode is able to filter steep angles better. "

Really it's a checkbox feature, just like SM3.0 support on the GF6 series, and it only matters a small amount to people who care about the difference; but you wanna talk about AF then it matters, if you don't care then don't bring it up. :roll:

I don't normally post here, but saw this review

Or were directed here by the viral PR firm you work for to spread the news... :roll:

BTW, why didn't nV simply provide the drivers to all the main review sites and help out if they were so interested or is their policy of Beta drivers not working well where no reviewer gets enough time to know what version nV will finally decide to certify. I'm sure that most reviewers just don't care whether nV wants to wait or not and then gets the review done with the tools at their disposal, later they may wish to update the review which is nice, but what would be nicer would be if nV focused more on review site relations and helping them unbiasedly, instead of sending their P minions into the forums after the fact to try and spread the 'love'. Seriously, talk about getting it all backwards, more interested in Spin control than in helping to get accurate information (because not all accurate information is to their benefit so they wouldn't want that right?).

I agree that it would be great if all reviews could be current with all the latest drivers and hardware, but as long as nV and ATi aren't willing to be completely open and uncontrolling in their support, then they will be at the mercy of review sties' lead times, and while this may be old news, there's no way the reviewers could know what nV felt like doing sometime down the road once they found a driver they liked enough to certify. Since this is and investigation of something even nV didn't support at the time of writing, I'd say it's a nice effort, and hopefully tey'll update it again when they have the time, until then nV will be stuck with poor impressions of 'Quad' SLi due to their own resistance to be open and helpful to reviewers. And you can take that back to the mothership with you. :tongue:

 

[Action_Man]

Owned.

 

[Action_Man]

I've seen you post this at a number of forums now and I've seen you get banned at a number of forums as well. Fock off with this sh!t.

 

[Rollo]

I've seen you post this at a number of forums now and I've seen you get banned at a number of forums as well. Fock off with this sh!t.

Sorry, my points were valid and you haven't really contributed anything to the thread.

Not likely Darren would have issued his clarification article and intent to re-examine if I were wrong, is it?

 

[Heyyou27]

Not all of the FP16+AA workarounds involve the CPU; as far as the screenshot, they're both taken at different angles so it's really harder to compare. (I do wish I had HQ AF though )

 

[Rollo]

HAs nothing to do with justifying purchases it has to do with reality and truth. nV has no equivalenet level of HQ AF, it's a hardware limitation (just like FP16HDR+AA) and no magic drivers will ever fix that.

You mean like the magic unsupported patch that enables HDR+AA in Oblivion, and has issues with rendering shadows?
http://www.elitebastards.com/cms/index.php?option=com_content&task=view&id=98&Itemid=29

However, as we enter August, we are still left without a WHQL driver incorporating the 'Chuck' patch, nor do we have a non-WHQL release of CATALYST 6.7 incorporating this functionality.

Maybe the lack of supported FP16 HDR+AA games is the reason so few people care about this check box feature?

Just like everything else, there are extraneous factors, but we're talking about head 2 head so those other things you want to toss in there to make yourself feel better really aren't relevant.

Heh- I don't need to make myself feel better. You seem to put more stock in this stuff than many people might though?
The fact of the matter is that the image quality is pretty similar and posting a link to some 4X AF pictures on low end cards doesn't really prove your point. How about some 16X AF High Quality comparison?

What does market share have to do with this? But people obviously care or else it wouldn't keep coming up in reviews;

Like I said, the first quarter Mercury research poll showed that ATI only captured 17% of the performance DX9 market.
http://phx.corporate-ir.net/phoenix.zhtml?c=116466&p=irol-newsArticle_Print&ID=855747&highlight=
Do you honestly think if there was some huge difference in image quality 4/5 enthusiasts would be ignoring ATI? When their cards are usually cheaper? Uh huh. And they didn't just get swallowed whole by a cpu maker that needs a business chipset and some integrated graphics either....errrr...oh yeah.

Really it's a checkbox feature, just like SM3.0 support on the GF6 series, and it only matters a small amount to people who care about the difference; but you wanna talk about AF then it matters, if you don't care then don't bring it up. :roll:

Hmmm You've been here a long time. I wonder if you were this down on ATI back in the R300 days when ATI had angle dependent AF and nV 5900s had angle independent AF? Should we do a search, or do you want to fess up now?

Or were directed here by the viral PR firm you work for to spread the news... :roll:

:roll:
Oh boy. Another self proclaimed "expert" on the PR business who has no clue what he's talking about.
1. I have no connection to any PR firm of any kind, and haven't for many months. While it is true I used to get parts and software from AEG and nVidia, I only work with nVidia now. I'm a moderator on nZone and at GeForce3d, and still do beta testing for nVidia on occasion.
2. No one "sends" me anywhere, nor does nVidia necessarily approve of what I say. They don't try to control it either, because guess what? I don't work for nVidia.

BTW, why didn't nV simply provide the drivers to all the main review sites and help out if they were so interested or is their policy of Beta drivers not working well where no reviewer gets enough time to know what version nV will finally decide to certify. I'm sure that most reviewers just don't care whether nV wants to wait or not and then gets the review done with the tools at their disposal, later they may wish to update the review which is nice, but what would be nicer would be if nV focused more on review site relations and helping them unbiasedly, instead of sending their P minions into the forums after the fact to try and spread the 'love'. Seriously, talk about getting it all backwards, more interested in Spin control than in helping to get accurate information (because not all accurate information is to their benefit so they wouldn't want that right?).

Firing Squad didn't seem to have any trouble getting out a good review of Quad SLi with the 91.37 beta drivers almost two weeks before THGs article with the pre-betas?
HardOCP and HotHardware got out user impressions and a few benches of the 91.45 production drivers a few days later?
Only THG used the old pre-beta drivers lately, and frankly, I'm a bit sick of flogging that horse. They admitted that it needed to be re-examined, so I was right? :roll:
I'm sure it was an honest mistake on their part, and I can understand being busy, so it's not a big deal.
Interesting how you think me pointing it out is though? Why is that? What's it to you?

I agree that it would be great if all reviews could be current with all the latest drivers and hardware, but as long as nV and ATi aren't willing to be completely open and uncontrolling in their support, then they will be at the mercy of review sties' lead times, and while this may be old news, there's no way the reviewers could know what nV felt like doing sometime down the road once they found a driver they liked enough to certify. Since this is and investigation of something even nV didn't support at the time of writing, I'd say it's a nice effort, and hopefully tey'll update it again when they have the time, until then nV will be stuck with poor impressions of 'Quad' SLi due to their own resistance to be open and helpful to reviewers. And you can take that back to the mothership with you. :tongue:

Your speculation is all well and good, but nVidia released the Beta drivers for Quad SLi on July 17, two and a half weeks before the THG article was released. Surely as industry professionals THG was aware of this development, and had those two and a half weeks to either re-bench or post pone the article?

Anyway, I'm not going to get into that whole AEG thing again here. I've said in this thread what I plan to say about it. (especially as I haven't even talked to anyone from there since March)

ATI makes some nice hardware, and there's no doubt that on some monitors, with some games, you can see they have less shimmer in some spots. HDR+AA is a novelty in it's infancy, by the time the Chuck patch is incorporated into drivers, the G80 will be released, with the R600 on it's heels, and current HDR+AA will not be what anyone with any money will want to have.

You and I have no history, and therefore no quarrel beyond this thread.

I'm not here to start or participate in flame wars, if you want to, I'll just end up putting you on ignore and it will be one sided.

I'd rather discuss the issues without sarcasm or accusations, but I guess that is up to you.

 

[Action_Man]

Sorry, my points were valid and you haven't really contributed anything to the thread.

Mainly because its sh!t and sponsored by nvidia.

 

[niz]

Yeah it always amazes me when they still benchmark the absolute fastest GPUs at like 1024x768.

Its a total waste of time as at that res. the performance bottleneck isn't the GPU but the rest of the system. The GPU isn't even breaking a sweat so it doens't give you any clue how fast the GPU really is.

Anyway who still plays games at 1024x768? I havent done that for years...

 

[niz]

Yeah but ATI's Linux drivers are a joke. Nvidias Linux drivers are really good.

 

[Rollo]

Sorry, my points were valid and you haven't really contributed anything to the thread.

Mainly because its sh!t and sponsored by nvidia.

Again, my points were valid.

Darren and THG admitted as much after I posted this.

Who had more value to the board in this thread?

The new user who pointed out serious flaws with a review that could shape buying decisions with misinformation?

Or the board veteran who is angry the new user gets a couple free video cards a year from nVidia?

If I came here and posted, "Buy nVidia, ATI stinks!" you would have a point.

However, I came here and posted the truth about problems with a review, and you should be grateful I did because apparently no one else realized the problems with the review.

Someone with a 24" display who was looking for a new multi card setup might want to know what I posted here. Why didn't you want them to have it?

Because nVidia gives me a couple cards a year? Sort of petty on your part , isn't it?

 

[Action_Man]

Again you're totally missing the point. Fock off and do something better with your time.