Anandtech: then and now

Status
Not open for further replies.
LOL, hardly the same thing.
Click to expand...

Your right, he doesn't use one of the benches and yet is expected to make a special post about its validity? Take a real look at the situation at Anand's. First off he can't properly identify bilinear and trilinear filtering(+ATi), he runs benches that utilize features that have known rendering errors in certain games which are game bugs and fails to mention or acknowledge that(-ATi), he shows bench results that are nearly twice as high as other sites on some of the benches he runs(+nV), he talks about solid drivers that hard lock the most popular on line game in the world(+ATi), he doesn't mention rendering errors the require modifications to game's configuration file to fix even in the games he benches(+nV), none of these things have been acknowledged as of yet and he is supposed to make a special comment on a bench that he doesn't even use? There are several more important things that need to be looked at concerning Anand's reviews then the 3DM2K3 issue. It is of no impact to end users in terms of how it will effect their useage of the board, the issues I mentioned above all are.
 
Anand, bah. I'll never visit Anand again now that he uses Flash for graphs (I still haven't read the 5900 Ultra review because I don't use Flash. I hate Flash. I despise it with the core of my being.).

But then again, I also know sites like Anandtech actually sit there and ask themselves questions like, "Hmmm...This doesn't seem very correct. Let me investigate it." They just plop the card it, run a few timedemo's without so much as even thinking about anything relating to the game (IE how does it actually play outside the benchmark?)....
Click to expand...

Uh, you want to clear this up, Typedef?
 
oops...

What I meant to say was something like...

Sites like Anandtech pretty much plop the card in, run the timedemos, and report their findings...

Suppose, for example, the driver completely disabled AA in everything above 12x10, for some particular reason...I have absolutely no doubt whatsoever that such a bug would never be caught be these stooges because they don't do squat beyond posting framerates.
 
I suggest you email them, and let them know
Click to expand...

I've e-mailed Anand repeatedly about the Splinter cell issue in his review. The issues with both the splinter cell benches and the quake 3 benches have both been brought up in his forum. This was finally posted in his forum by Evan Lieb, one of his reviewers:

We apologize for the delayed response, we are currently looking into the issues you have mentioned here. Be patient and you will know in time.

Take care,

Evan
Click to expand...

That was posted on the 20th. Haven't heard anything since. I can understand being patient, but really, that review has been up for almost 2 weeks now. Any erroneous information in it should be corrected immediately, or at least have a disclaimer included. He's doing a disservice to his readership to keep errors in a review for so long.

To leave something like this in a review for 2 weeks when it is easily verified to be a problem with all cards that actually have multisample AA turned on is ridiculous:

The Catalyst 3.4 drivers provided us with a couple of problems, the most noticeable ones were the following:

There were serious lighting issues in Splinter Cell as well as degraded performance, an example of the lighting issues can be seen below:
Click to expand...

For him to say this is an ATI driver problem is misleading. This is something that should have happened with the nv35 also, if AA was actually working on the nv35. By the looks of his benchmark numbers it wasn't. Even worse, he says the 9800 had "degraded performance" on splinter cell. No kidding Anand, that's what happens when you actually have AA on. When the Nv35 isn't running AA it's number's do lead you to believe the 9800 has degraded performance comparitively. How bout next time you run a benchmark with AA you actually verify AA is on.

To make these mistakes is one thing. To have them still in the review 2 weeks later even after it has been brought to your attention is simply unacceptable. Fixing the errors in this review should have been a priority. Whatever future article he is working on should take a back seat to getting this one fixed.
 
Typedef Enum said:
I am disappointed that Anandtech can't see how important this is..
Click to expand...

I suggest you email them, and let them know...and tell everybody you know to do the same.

It's pathetic that the Quake3 graph is still up there right now, 2 weeks after the fact that it was utterly obvious that something wasn't right...Hell, I would go on to say that any freakin' idiot with half a brain would know that something isn't right when just looking @ the number...

But then again, I also know sites like Anandtech actually sit there and ask themselves questions like, "Hmmm...This doesn't seem very correct. Let me investigate it." They just plop the card it, run a few timedemo's without so much as even thinking about anything relating to the game (IE how does it actually play outside the benchmark?)....
Click to expand...

You are right about that. I suppose we should all be e-mailing Nvidia as well about their cheating drivers :D
 
Im curioius as to the liability that could be incurred to due misleading/false benchmarks by a website such as anands. If I purchase my card based on their numbers and then cant replicate them would they be liable?

just wondering,
epicstruggle
 
Ben, wasn't the HL ESC bug fixed with Cat 3.4?
Click to expand...

Yes, which became available at the launch of the R9800 256MB, what about before that? I listed issues that have had an actual impact on end users which the whole 3DM2K3 debacle doesn't. Anand can't cover the things that do effect his readers due to lack of depth in his articles, why should he cover something that has no impact on end users and does not influence the benches he uses? He has made direct, misleading statements in his articles on the issues I brought up above, he has remained silent on 3DM2K3 and he has yet to use the score, so why should he be looking to apologize for misinformation that other sites have provided(not on purpose mind you, just the results are invalid) when he has so much of his own misinformation that he has yet to cover?
 
He didn't benchmark Half-life so he wouldn't have found that bug, it may be a popular game but it is very old, and not used for benchmarking. Catalyst 3.1 worked fine for HL players, or run in Direct3D.

This is different, he is posting ATI driver bugs causing light shining through walls in Splinter Cell..a game that does not support MSAA.
 
I fired up HL with the 3.1 Cats (before installing the 3.4s) for several hours a few days ago--never had a problem once with ESC. No problems now with the 3.4's. As always, few bug reports are universal in scope even though it is often assumed that because some have problems everybody has problems.
 
Counterstrike didn't work, given the audience he is writing too if that didn't warrant comment how does 3DM2K3 even come close? You bring up a valid point that he doesn't use HL powered games to test, nor is he using 3DMark2K3. CS is played by hundreds of thousands of users every day, nobody 'plays' 3DM2K3 and only a small handful of people can so much as see the hack in action. On the other side of the fence, nVidia is producing numerous artifacts in SS right now with the current drivers using the default configuration files. That effects thousands of end users who read AnandTech and I haven't seen him cover that either. I listed examples from both sides where he has either skipped over issues that matter to end users or spread misinformation himself. Given that he doesn't use 3DMark2K3 why should any reasonable people expect him to pay attention to it?

This is different, he is posting ATI driver bugs causing light shining through walls in Splinter Cell..a game that does not support MSAA.
Click to expand...

To quote myself-

he runs benches that utilize features that have known rendering errors in certain games which are game bugs and fails to mention or acknowledge that(-ATi)
Click to expand...

How is covering Splinter Cell's known rendering errors with MSAA different then covering Splinter Cell's known rendering errors with MSAA? ;) Actual issues that impact end users are exactly what I'm saying he should be looking at. IF he offered up flawless reviews then I could see expanding out to cover something he doesn't use in his benches. As it stands now, he is screwing up the stuff he does run enough that he should really be focusing on that first.

Edit-

I fired up HL with the 3.1 Cats (before installing the 3.4s) for several hours a few days ago--never had a problem once with ESC. No problems now with the 3.4's.
Click to expand...

That's nice. What about the 3.2s? When one of the bodies reporting the issue is ATi themselves, it makes it carry a bit more weight then someone dancing around the drivers that have the problem ;)
 
You Seriously can't expect reviewers to tests every game for bugs, especially old ones.

You are right about the amount of people playing HL, but their survey shows that 3% are using high end cards. ;)

http://valve.speakeasy.net/

GeForce2 MX 248531 (28.6%)
NVidia TNT2 159964 (18.4%)
GeForce2 GTS 52037 (6.0%)
Intel 810 43969 (5.1%)
GeForce3 38880 (4.5%)
NVidia GeForce4 MX 30588 (3.5%)
Radeon 27851 (3.2%)
 
Render Mode:

Software 223115 (44.4%) <---- :LOL:
OpenGL 205424 (40.9%)
Direct 3D 73868 (14.7%)


Thats why so many play it, system requirements are very low...it is almost sad a DX9 card running it. :D
 
Im suprised no one has commented on anands absolutely Ridiculous comments on ATi's/Nvidias Pipeline arrangements.

http://www.anandtech.com/video/showdoc.html?i=1821&p=2

From reading his little Diatribe about pipelines and his Questionable statements about *poor* information from ATi as well as Nvidia, And his misconclusions about the whole subject all written as a *Defense* for Nvidia... Then Rolling the rest of the issues Brought up here.

It is very obvious that Nvidia had something going on in there with this Article. Perhaps he just went to Nvidia For a Visit and was full of Scmoozing? I dont know, but there is a Clear 180 from his previous few articles. Its very similar to the Sudden change at [H] as well.

Pretty Suspicious to me.
 
Btw, demalion

Thanks for the Link. I had not seen that and you did address the issues very, very well. I can relax now, people did take notice and put down the Truth Smack.

Still though.. makes it even worse that he has not updated and corrected all these *mistakes*.. if you can call them that.
 
BenSkywalker,

If 3D Mark 03 is so unimportant then why did nVidia spend so much time and energy figuring how to cheat it? That's the heart of nVidia's credibility problem with regard to this issue. I might have believed nVidia's protests about the general value of 3D Mark 03 were sincere--had the company not bothered with engineering cheats for it into their latest drivers and then brashly bragging on illusionary "performance increases" that were merely the result of those cheats in the very benchmark they seek to discredit.

What I think is at the very core of nVidia's many problems with 3D Mark is that nVidia wants to totally control how its products are compared to those of its competition. 3D Mark 03 doesn't allow that--so nVidia quit the program and cheated the benchmark in an attempt to control it regardless; an attempt which has failed and backlashed. nVidia simply doesn't want to have its products compared on a "level playing field" of a third-party's determination--and that's what all of this is about.

I don't care what kind of optimzations they do in games--that doesn't matter. But nVidia knows that whether it likes it or not people do use 3D Mark to compare products. But rather than stay in the program (which nVidia has been in for years) nVidia publicly denigrated the benchmark, resigned from the program, and cheated 3D Mark 03. Can't get much lower than that.
 
Is it me, or does this whole debacle feel like deja vu all over again? And I mean both back in terms of Quake "Quack" and in terms of some of the debates I remember back from the nVidia vs 3dfx days. I've read the threads in various forums over this, as well as some of the news coverage...and I don't want to mention names, but have seen responces such as

"I don't know which is worse, cheating or disloyalty"

followed by indication of how nVidia has always been loyal to them... I've even seen it written where IQ doesn't really matter and some people are just too obsessed with it all the while setting aside the perf gains that dropped IQ were done in a synthetic benchmark people use to compare cards, instead of in an actual game which gets played. When some people comment on the argument of IQ vs performance and what 3dfx said..."oh no, they can't be the same". Comparison of benching nv3x mode vs other modes in Doom 3 is brought up (we'll see what is done) and some people argue how this can't be at all the same as comparing Glide vs D3D scores as the IQ doesn't drop, and yet others point out that nv3x mode uses lower fp precision; albeit this would be a discussion better for another thread.

But they are... And in this case the perf increase at the expence of IQ is being done in a bench only. What I don't get is how this is supposed to be better for nVidia today then it was for ATI or others in the past. And if some people (though not all) who were in the nVidia camp for many, many years are now saying IQ is not important and so enhancing perf at all costs is good (not what I read 4 years ago), I still haven't heard a good reason for why this would apply in a synthetic benchmark only.

And I hate to say it, but at this point nVidia's responce of accusing Future Mark of changing their source code to make their GF FX look bad (which BTW didn't by their own admission sell as well as they were hoping, given in a statement alongside the NV35's release very shortly after NV30 was out the door), doesn't help their case any...

Now throw in the questions concerning the review sites and who is or is not biased...seems very familiar, though the names and faces have somewhat changed...
 
Status
Not open for further replies.
Back
Top