Tom's Hardware

[X-Caliber]

http://www6.tomshardware.com/graphic/02q2/020514/index.html

[Nuno]

Why can´t FAA-16 be used when stencil buffer is also being used?

[mattsteg]

From what I read it doesn't see the edges produced by the stencil buffer as an edge, according to Tom's at least.

(I trust Tom's Hardware about as far as I can throw a ton of bricks. Sometimes they just say weird things with NO basis in reality. My personal favorite was when ther reviewed the Western Digital WD1200JB. The title claimed it outperformed SCSI drives, the sidebar said it performed similarly to SCSI drives, and the actual article did not include a single benchmark of a SCSI drive. http://www4.tomshardware.com/storage...305/index.html)

Take anything Tom and company say with a pillar of salt. He may be correct, he may be making stuff up, or it may be a result of beta drivers. Wait for someone credible to review the card, and read some of the other previews from more credible sources.

[ChronosZero]

Okay. I hear time and time again that Tom's Hardware just isn't very reliable when it comes to accurate reporting and research. Personally that isn't the only problem I have with the way they run the site. I'm starting up a new hardware site, as my signature so unabashedly proclaims. Since we're still getting going, I thought I would pose a question to all of you here:

What is lacking from the major hardware sites?

(BTW, no... we don't have a preview of the parhelia-512 yet, because the site is too new and small to get an NDA from Matrox and much of my time over the past week has been occupied by tending to an unexpected death in the family. We hope to have something comprehensive in a couple days.)

[Dr Mordrid]

Toms Hardware: this preview was written by the editor in chief of RivaNation.com...a pro-NVIDIA site. No independence there.

Anandtech: Anand does a good job at times, but misses the boat ofen enough to make me question his methods. In this case he noted the absense of Z-culling while missing totally the depth adaptive tesselation, depth acceleration unit and other features that make culling less essential than with other cards.

Dr. Mordrid

[Joel]

Unbiased reporting with conclusions being based on fact and not opinion.

Joel

[Dr Mordrid]

That about sums it up. The major problem is that many of them have hidden agenda's by way of support from interested parties.

That said some sites do try to be fair. Hardwarezone.com seems to have given the Parhelia a fair shot in its preview.

Dr. Mordrid

[ChronosZero]

Thanks for the early feedback. It sounds like we've really hit a nerve here. I was beginning to think that I was the only one who thought these things.

[Dr Mordrid]

I think that as benchmarks appear showing Parhelia to be VERY fast in newer and advanced games you'll hear even more of this squealing and yelling.

Both the NVIDIA and ATI camps will be bitching like hell about how this or that feature isn't so hot and that in 3/6/9 months they'll surpass it by a mile.

It'll be a massive, if not Clintonesque, damage control operation

Dr. Mordrid

[ChronosZero]

Such is the nature of the game. It's all about spin and control. Of course there will always be fan boys who are stuck on frame rates which don't give them real world results. 300 fps in Quake III Arena sounds spiffy, but it doesn't look any better. It's like the "kids" who spend thousands of dollars increasing the horsepower of their cars but those machines have never seen a race track.

Granted, the newer, faster cards help out in the games like Serious Sam: The Second Encounter. I'm very curious to see if some of the new features of the Parhelia, like the quad vertex shaders and quad texturing will make a significant difference.

The hardware displacement engine sounds very intriguing. I'm proud of Matrox for playing the same game as NVIDIA by making suggestions to Microsoft as to what should be included in the next iteration of DirectX.

[The_King]

Trying to be as unbiased as possible, whats depth adaptive tesselation and how useful will it be?

[impact]

It's the objects_far_away_will_be_less_detailed_than_object s_near_us thing.

[Dr Mordrid]

Right. In one of Matrox's demo scenes it reduced the polygon count from 150,000+ down to just 17,000. That gives you a huge advantage in rendering speed.

Dr. Mordrid

[The_King]

Originally posted by Dr Mordrid
Right. In one of Matrox's demo scenes it reduced the polygon count from 150,000+ down to just 17,000. That gives you a huge advantage in rendering speed.

Dr. Mordrid

God I hope that game developers use the card so that these features are actually implemented...

[The PIT]

Whats this about driver support for win2k and xp striaght away with 9x/me following. I can't see that being right.