review at chip.de

[string]

actually after thinking over the chip review is as awfull as the tomshardware. just in the other side.

1. it doenst mention at all the "smart"(quilsomething or what its called) implementation of full scene antialiasing but test only 4xFsAA with geforce, claiming they would be fans of "clever" solutions at end. (letting the clever solution of the opponent out.

2. they say that for the case the 16xFAA fails there is still like with the competitors a 2x2 FSAA there, but dont lose a word that (no reason not beliving the tomshardware results) about the unusuability of this mode for actually doing something more than writing it on boxes

3. they also dont test what actually would be intresting. : playability in surround. playability with upcoming games (ut2) with different setups.

4. where is more than a quote of press releases information about the 2d quality. especially on crt monitors which have to pass 1 adapter (+connection) and than another connection.

i hope ms from lostcircuits gets the card soon.

[TdB]

Originally posted by Rags
I still stand by my statement that benchmarks don't show the whole picture. They tell nothing about how well a card will play in a game. They don't show the smoothness, the minimum, etc. That is something that can only be done when the hardware is in your hands.

Rags

thats why i will wait and see all the comments from murcers with parhelias, before i decide.
I feared that there would come a day where benchmarks no longer, told everything about performance, and how smooth games ran, It is about bloody time we stop using AVERAGE fps when benchmarking, perhaps even framerates at all.
Unfortunately, no reviewers has a benchmarkings suite that tells what this card can do.
thats the sad part of it, the focus on worstcase scenarios could backfire on matrox, because they won´t win any standard benchmarks with it, and I think that is sad, because they have the same focus as me,(and alot of other gamers), yet the won´t get any credit for it.

[Indiana]

Uhm, have you ever seen those "intelligent" NVidia FSAA techniques?!?
Quincunx looks crap.
NVidias 4x FSAA looks good, a bit worse than ATIs 4x quality smoothvision, but in contrary to the ATI it's just fast enough to be usable.

But you can't really compare Quincunx to Matrox' AA cause the quality difference is HUGE. What IS bad about the chip review is that they're not saying any word about anisotropic filtering and about the quite low setting the Parhelia seems to use.

[the maddman]

Serious Sam detects the type of card you have and changes how it renders. I bet that they don't detect the Parhelia, and fall back to 2 texture multipass rendering, instead of using all 4 TMUs. Give Croteam time to release a patch.

Originally posted by TDB

thats because there are no polygon edges in that test.
It is the only test that use 4 layer textures though, so parhelias 4 tmus/pipe, allows it to pull ahead of the competition.

EDIT: correction serious sam:se, also uses up to 4 layers of textures in some levels, most often only 3 layers, so parhelia should beat everything in this test, since it has twice the tmus(Texture Mapping Units), it should render SS:SE in a single pass(the gf4ti only has 2tmu/pipe, and must do multipass, AFAIK), with 4 pixels per clock.
i wonder what is holding it back?
drivers perhaps?

[Gear]

Originally posted by Rags
I still stand by my statement that benchmarks don't show the whole picture. They tell nothing about how well a card will play in a game. They don't show the smoothness, the minimum, etc. That is something that can only be done when the hardware is in your hands.

Rags

It would have been a nicer review if on the FPS
benchmarks, every bar would start at the minimum FPS, just to see how worse it can get in a game. A maximum FPS score would not help much, as it wouldn't be noticed in gameplay anyway...

Or , maybe in the future mSPF (milliseconds per frame) a.k.a. smoothness factor

[xortam]

Originally posted by TDB
... because they won´t win any standard benchmarks with it, and I think that is sad, because they have the same focus as me,(and alot of other gamers), yet the won´t get any credit for it.

I don't care what others think. I buy HW that suits my needs regardless of what's considered popular. My only concern is the quality of the product and the continued support of the HW I buy.

[TdB]

that could be an explanation for the SS:SE performance, but why didn´t they just enable it, in the advanced options?

[TdB]

Originally posted by xortam
I don't care what others think. I buy HW that suits my needs regardless of what's considered popular. My only concern is the quality of the product and the continued support of the HW I buy.

but it still, makes it difficult to make a decision based on reviews.

[string]

dont get me wrong. i even have my old geforce3 64 on ebay for sale already knowing the parhelia would come, and atm running on my beloved g400

the quinux nvidia thing may have some flaws, but so does the 16xfsaa of matrox still has. to find out which one is hardware related and which one a driver issue would be interesting in a review.
i see no point tho why just not talking of that feature at all.
as someone that has matrox cards for like 12 years, i dont need biased reviews or ppl thinking for me. i want hard facts and a lot of them.
fsaa+a the anisotropic gets sadly a very blurry picture -> here would be also more interesting. driverside? hardwareside?

[P5ycho]

You can custom configure ss2 for quad texturing.

[isochar]

From the pics I saw, nVidia's 4xFSAA doesn't hold a candle to Matrox's 16xFAA. (IQ-wise and performance-wise) However, the anistropic on the Matrox card looked awfully blurry.

[xortam]

Originally posted by TDB

but it still, makes it difficult to make a decision based on reviews.

You can really get turned around by reading reviews. Reviews are probably good for throwing out the worst from consideration and calling attention to functionality that may interest you but ultimately you've got to pays your money and takes your chances. Many stores allow free exchanges which should help you find the right product for you. It normally only comes down to a couple choices.

[TdB]

Originally posted by P5ycho
You can custom configure ss2 for quad texturing.

yes i know, my question is: does the reviewer know?
if he does, then why isn´t it faster?

[Jutai]

i love how promising it looks with the features turned on the loss is so small, obviously the scores are a little on the low side (really low for some) but we all know continued driver tweaking can really help this big M might be able to reach the goal of playability with all the features turned on. I'd also really love to see a higher clocked "MAX" version with the 256meg offering

[nggalai]

Originally posted by Rags
I still stand by my statement that benchmarks don't show the whole picture. They tell nothing about how well a card will play in a game. They don't show the smoothness, the minimum, etc. That is something that can only be done when the hardware is in your hands.

Rags

*nods*

but be patient--in some 20 minutes, 3DCenter.de 's NDA expires and the review should be online. Mark that 3DCenter benches nothing but games, no tech demos, no 3DMarks or anything . . . should be an interesting read.

ta,
.rb