NV35 Reviews

Hadn't seen those screens. Ati's 4x looks better than Nvidia's 8x in my opinion. There are far more jaggies on Nvidia's 8x than on Ati's 4x.

 

Hmmm. This is rather interesting. 3 images (nv 8x, nv 6sx, ati6x). All three pixel dimensions 1024x378. All three are very similar pictures with apparant very similar view areas. But three very different JPG file sizes. nv 8x=49,982 bytes, nv 6sx=70,068 bytes, ati 6x=98,092. A factor of 2 from nv 8x to ati 6x. Wow, that ati 6x really makes JPG compression worse.

And the double 4x screenshoot with a pixel resolution of 1024x738 has a file size of 102,254 bytes.

Why aren't PNG files available for all IQ comparisons?

 

Talk to Hardocp, you don't need .png format to see jaggies though...LOD bias, or aliasing sure.

 

In terms of quality, it looks like GeforceFX FSAA maxes out at 4xS. After that it actually seems to look worse. However, even 4xS seems to look a bit worse than ATI's 4x, to me.

 

Demalion,

Seems i wasn't too far from the truth

http://www.beyond3d.com/forum/viewtopic.php?t=5805

 

Do you just ignore what I say as a general practice?

Evildeus, two things:
1) I remain amazed at your persistence in a faith based belief system for video hardware.
2) It is already part of my philosophy to recognize that the reason for a belief does not preclude the belief being right.

I'm not going to deny you the joy of your faith confirmation, but I ask you, for the sake of future time usage, to carefully consider why I believed MDolenc so readily and not your commentary or the screenshots you presented. The reasoning is provided in our prior discussion, and I hope you take the opportunity to achieve some understanding of that reasoning (something still not displayed as of yet).

 

*edit* Doesn't matter

Oh btw, i like this:

 

Yes, this statement speaks for itself. Congrats.
A contrast: "Proven? I can't prove anything without screenshots..."
Please note the rest of that sentence, if you recall where it is located.

Thanks for sharing those questions and answers, by the way.

Hmm...my opinion differs on that.

Future discussion of this belongs in PMs, and perhaps there you can explain this last statement more clearly for me.

 

I don't see the point continuing discussing with you Demalion that's why i edited my post before but seems you saw it anyway. When i provide you with some links you rejects them and now you ask for. Ok take one over here:

That's why i said that the NV35 was only tested on 44.* drivers and not 43.**
http://forum.hardware.fr/forum2.php...1&sondage=&owntopic=&theme=&trash=&subcat=109

When you say that someone can't think, is just belief, because he thinks differently from you and that this person has made some research on his side, i think it's purposely attacking the person.

Good, the discussion is closed.

PS: You said

Seems to me that the abstract is sometimes better than supposed facts.

 

I wasn't asking for a relevant link from you now, I was complaining about the lack of one before MDolenc provided me with useful information elsewhere. Not that the quote seems very helpful, but I'm presuming the further info in the link might be if I read French?

However, these quotes need addressing in public, since that seems to be the priority for the conclusion of this discussion for you :

You don't do my thinking for me. I don't expect you to let me do your thinking for you either...your problem is that your recognition of this idea is one way, and you think you provided info when you provided the conclusions and not the evidence behind it.
I did not say you can't think, I said you presented no information to me that indicated your belief was based on reasoning from evaluating evidence, but instead your trust in the author's presentation despite the flaws I pointed out in it. I think you should consider the text until the difference is apparent to you, this repetition of that is where my effort in that regard ends in this discussion (outside of PMs if you really want me to clarify at length).

Is it?

Guess not.

First, I was referring to your own representation of my argument as the statement that was "completely in the abstract", the only thing I knew about your argument is that you didn't provide the evidence to me, not that the evidence didn't exist. Please regard it as other than an insult when I ask if I am simply overestimating your English reading ability? Given my (lack of) understanding of French, I hope you can regard that as an honest question, as you appear to have thought that statement was an attack on you.

Here is the statement I was referring to, please consider the comment you quoted carefully again in light of it: "It's you that aren't looking at what i said. You are saying that because it's broken on NV30 then Nvidia must be tweaking/cheating on NV35."

That statement is you representing my viewpoint as a statement based only on the abstract, whereas my actual reasoning was based on the captions used, and known factors about the NV30's capabilities and 43.51 GT 4 screenshots...if you go and actually read my reasoning, this is clearly established.
The problem is you seem to be unable to recognize the difference, and consider what you provided in the discussion as equivalent because you think that your representation accurately portrays my statements.

Anyways, that didn't seem to prevent some of my conclusions about NV35 from being wrong, but it seems you won't rest until I also stop thinking your provided discussion was lacking. I do believe that's fruitless since you've consistently failed to respond to an accurate representation of what I state, but if you wish to continue trying I really do recommend PMs, as my side of the conversation already exists in the thread.

 

See PM

PS: Can Nvidia cheat in shadermark? And then how?
http://www.hardocp.com/article.html?art=NDcyLDEy

 

About 100+ pages of reading later...

That is alot of material from some very good and some spotty reviews. What comes across is that both these cards are real powerhouses.

NVidia obivously did well with their architecture and finding a balanced design. I'm sure their drivers (the Detonators 50.xx are meant to have both 32 and 64 bit drivers btw) will only get increasingly better.

The only major criticism I have seen leveled is when will NVidia go to a true 8 pipeline design. The major compliment to them is raw bandwidth plus greatly improved shader op's has really paid off - and so to has the heat management solution. It seems neck and neck - one has better AF and the other has better FSAA - so overall image quality in most cases goes to ATi but NVidia have really caught up in many cases and now their drivers and Control Panel settings are much more honest.

I pay little credit to the Doom 3 benchmarks other than say NVidia do quite well so far - but it should be an even comparision and ATi obivously haven't written a driver to tweak this beta coded game yet. When they have and when both cards can be run through the same code path then it will be very interesting to say where performance will lie.

ATi are hinting the R360 in July will be stronger than NV35 - which is possible if they up memory bandwidth and GPU speed - which is a strong possibility - almost a probability I'd say.

The only factor that stings is the cost given the performance ramp up from here. In just a years time these AUD $1,000 cards will be 30% - 60% behind the performance leaders, probably cost between AUD $450 - $600 and few Directx 9 games will be available - so should we buy them until say at least 5-6 great games are out that would benefit from this level of hardware?

 

Marc, do you mind editing your post to include a link to that huge picture? It's screwing up the formatting of the rest of the page. If you want to include that picture in your post, I believe hardware.fr also has a smaller one you can inline, and you can still include a link to the bigger one. Thanks.

 

My initial thoughts is no they wouldn't bother.

Looking at the digitlife/ixbt review shows a different perspective though

http://www.digit-life.com/articles2/gffx/5900u.html

Look at the pixel shader 2.0 test (about 1/4 of the way down) in their rightmark 3d bench. nv35s performance is as bad as nv30s while r350 is over twice the speed.

 

Intesresting. Thx.

 

What seems to be the problem with Digit-Life's Rightmark pixel shader 2.0 benchmark result on the NV35? They seem to be really low for having 12 shader instructions/clock ability.

 

NV had no access to Rightmark sources ... so expect in some Detonator soon, that RM's scores for NV cards will "jump", as NV will add Rm-specific-optimisations ....

 

Yea, like the did for 3DM03: http://www.beyond3d.com/forum/viewtopic.php?t=5852

and it's called cheating!