[micamica1217]

some quotes from B3D....
full read here...LOOK.


from pocketmoon66,
""quick brain dump from the Nvidia dev conference in London on Monday/Tuesday (nothing under NDA...)

Both VS3 and PS3 paths have been added to Fry Cry.

In general, gains from using Geometry Instancing usually between 0 and 10%, though up to 40% in worst (best!) case.

Devs wanted to use PS3 to loop through dynamic number of lights in a single pass but PS3 has limitations (in specs) which prevent this from being implementable. So use static branching for up to 4 lights per pass. This created more instructions than ps2 could handle so need ps3.

Gains are good - in game fps went from 20 to 30 in demo.

HDR - Did not require PS3 but requires 16bit blending and filtering. Devs replaced 'hacks' ( glare/flare etc) with HDR. No AA when using HDR. on 6800U playable (40-50fps) @1024x768.

Very Nice effect

Other Stuff

Nice demo of soft shadows (penumbra's) with PS3.0. Basically do a a small number of shadow sampler tests. If all 0 then you'r out of shadow, if all 1 your in shadow otherwise you are in the penumbra so do 64 samples to get your shadow level.

Works well because the expensive sampling needed is only done if youre initial small sample set is not wholly in or wholly out of shadow. 2xfaster than equivalent PS2 shader (not shown)

Unreal Engine 3 look stunning. Tools where very clever (graphical shader creation tool for artists). Tim S excited about 6800 SLI Target dx9 as minimum spec with release date 2006, blah, blah.

Same flame eating, fire breathing, power tool sparking entertaiment as for the NV30 launch ?!""


from pocketmoon66,
""
"Cali3350 wrote:
Any 1.2 release date revealed? "


Didn't Ask But they had a bug where the screen would flash black for a few moments. So don't hold your breath.

All in all the far cry devs where very open and honest. Where they though PS3 was not usefull they said so

The 0-20% gain they got using Geometry Instansing is taken up by pushing the sprite/mesh object boundary further out - so objects get rendered as sprites when at a certain distance but on NV40 they can push this distance out which they said looks nicer.

The images looked so good with HDR that even without AA and at low res it was very cool ""

--
ATI fans will know this guy....
--

from Humus,
""
"pocketmoon66 wrote:
Nice demo of soft shadows (penumbra's) with PS3.0. Basically do a a small number of shadow sampler tests. If all 0 then you'r out of shadow, if all 1 your in shadow otherwise you are in the penumbra so do 64 samples to get your shadow level. "

Works well because the expensive sampling needed is only done if youre initial small sample set is not wholly in or wholly out of shadow. 2xfaster than equivalent PS2 shader (not shown)


Unfortunately for nVidia, I have developed a technique at work that will work as a drop-in replacement of dynamic branching that works in the most important situations where you'd otherwise need or prefer ps3.0. You'll probably have to resort to very esoteric effects to really require ps3.0. The effect you mention sounds like a perfect example where this technique would work equally well (or maybe even better) than using dynamic branching, depending on how much the cost of dynamic branches are.
Another typical example is the "early-out" kind of optimization where you for instance detect that the pixel is backfacing the light or is outside the light radius and just return zero instead of going through the all the lighting computations. I have implemented this for a typical "Humus-demo" scene with four lights. With early-out enabled I get 45fps. Without it's 14fps. That's more than 3x performance improvement. If you prefer to dwell in the darkness, the difference is 136fps vs. 25fps. I'm not even sure nVidia's will be able to match that performance increase with ps3.0. ""

***added bonus that is not related to this***
(cool 3Dc demo located here....LOOK. )




I just thought you boys that are looking into the new cards would like to see this...

mica

[Dragonprince]

This just verifies what Ive felt about PS.3, its not much of an advance and really nothing to get excited about.

[rsjrv99]

I never really thought PS 3.0 was such a big deal. And as DragonPrince said, this verifies it.

[JigPu]

At least the speed gains seem to be somewhat tangable. That's pretty much the only thing I was looking forward to in PS3, and would have been dissapointed if it could only shave off something like 3FPS. 10 isn't bad, though could be better.

I hope that 10 isn't mosty from drivers (as some posters were thinking) though...
JigPu

[micamica1217]

Originally Posted by JigPu

At least the speed gains seem to be somewhat tangable. That's pretty much the only thing I was looking forward to in PS3, and would have been dissapointed if it could only shave off something like 3FPS. 10 isn't bad, though could be better.

I hope that 10 isn't mosty from drivers (as some posters were thinking) though...
JigPu

from that summary ps 3.0 isnt even being used. the only part of shader model 3 being used is instancing, and thats a 0 to 10% gain.

mica

[Phoenix87]

Originally Posted by micamica1217

from that summary ps 3.0 isnt even being used. the only part of shader model 3 being used is instancing, and thats a 0 to 10% gain.

mica

So how can any judgements about PS3.0 be made?

[micamica1217]

Originally Posted by Phoenix87

So how can any judgements about PS3.0 be made?

and that's the point....
if nVidia wanted to show SM3.0 they could do so quite easy with some homemade demos or beta game tests.
yet what did they realy show???

""Devs wanted to use PS3 to loop through dynamic number of lights in a single pass but PS3 has limitations (in specs) which prevent this from being implementable. So use static branching for up to 4 lights per pass. This created more instructions than ps2 could handle so need ps3.""

what the above quote tells me is that even with SM3.0, they felt like skiping dynamic branching for some (performance?) reason, and went for static instead. why? I have no idea.
yet they still needed SM3.0 to do the static branching????
I'm not sure that's true or false...we need a guru in here.

my take on all of this is that nVidia (and developers) are starting to understand that SM3.0 might not be the "be all, end all" performance boost some might think.
my thoughts are that nVidia needs to start working on getting some WHQL drivers out for the 6800 cards...
newer updated drivers seem like to be ware nVidia is going to get the most from the 6800 cards...performance wise.

I hope that after the first WHQL driver comes out for the 6800 cards, that it doesn't take another 3 months before nVidia puts out another WHQL driver.

mica

[JCLW]

AnandTech has a review of the 1.2 patch: http://www.anandtech.com/video/showdoc.html?i=2102

Originally Posted by AnandTech

We are here today to test out the new patch on six different levels in FarCry and see if the new methods, which CryTek were able to include in their new path, offer any kind of advantage. As the game play experience is meant to be the same no matter what card we're using, we'll clear the air before we start, and say that there will be no new eye candy available through the SM3.0 path. The game should be rendered exactly the same way it was under SM2.0...

Originally Posted by AnandTech

Image quality of both SM2.0 paths are on par with eachother, and the SM3.0 path on NVIDIA hardware shows negligable differences. The very slight variations are most likely just small fluctuations between the mathematical output of a single pass and a multipass lighting shader. The difference is honestly so tiny that you can't call either rendering lower quality from a visual standpoint. We will still try to learn what exactly causes the differences we noticed from CryTek.

The main point that the performance numbers make is not that SM3.0 has a speed advantage over SM2.0 (as even the opposite may be true), but that single pass per-pixel lighting models can significantly reduce the impact of adding an ever increasing number of lights to a scene.

[JCLW]

BTW: you guys might find this interesting as well: Humus dynamic branching with PS2.0 demo

[OBLIVIONLORD]

Wow on Anandtech's demo with AA and AF on it gave 10fps more then the x800Xt-Pe but, was 7fps slower when AA and AF were disabled.

I would really like to see how these cards do in both HL2 and Doom3

[Slimmy]

Originally Posted by OBLIVIONLORD

I would really like to see how these cards do in both HL2 and Doom3

That all i want to know!

[speed bump]

August 3 We will hopefully all know(crosses fingers) Hopefully I also have my 6800Gt to go with it.

[micamica1217]

Originally Posted by OBLIVIONLORD

Wow on Anandtech's demo with AA and AF on it gave 10fps more then the x800Xt-Pe but, was 7fps slower when AA and AF were disabled.

I would really like to see how these cards do in both HL2 and Doom3

paying about $600.00 and haveing to use water cooling to have a 6800UE...
I think most peeps will pass on this real quick.

the 6800u/gt models are what most will buy.
at $400.00 and $500.00 retail pricing, that's already quite alot of green.

but, untill I get an email back from Derek Wilson, I'm quite sure all tests were done with only brilinear filtering running on the 6800 cards.
sad.


still, 1%-15% performance increase is nothing to joke about.

Originally Posted by Anandtech review

Even some of the benchmarks with which NVIDIA supplied us showed that the new rendering path in FarCry isn't a magic bullet that increases performance across the board through the entire game.

even Derek see's clearly, that SM3.0 doesn't help all the time, and is more a slight help at some rare moments.

edit: update, the 6800 cards had briliner filtering running on all tests.

Hello,



Sorry about that, we used 1GB of memory.



We also used the default control panel options for both ati and NVIDIA with one exception – vsync was changed to always off for both cards.



Thanks,

Derek Wilson

CPU & Graphics Editor

AnandTech.com




--------------------------------------------------------------------------------

From: mica1217 [mailto:mica1217@XXXXXXXXX.net]
Sent: Friday, July 02, 2004 6:55 AM
To: derek.wilson@anandtech.com
Subject: FarCry v1.2 SM3.0 Path Analysis



hello,



I would like to thank you for a great review of what is going on in the new patch of FarCry.



I do have two questions:



1) you state the speed of you system memory (even the timmings used)...yet could you please tell me the amount of memory used in the test?

512mb? 1gig?



2) the numbers look like for the nVidia cards that the default options in the CP are used....

is trilinear filtering optimazations turned on? (brilinear)

or is it turned off? (full trilinear)



any other CP settings that were used might help me as well.



again, I'd like to thank you for your overview on the new farcry patch, and would like to thank you for your time.

mica

mica

[micamica1217]

some more info....here.

ooops, and it looks like a few sites might have to change thier graphs too, seems that they showed us wrong results....LOOK.

from Brandon

ellingsen: Yes, you and Chris Ray are correct, AA is handled the opposite on ATI cards, you have to use the control panel to set AA in Far Cry. IIRC it was the 61.11 driver where NVIDIA forced users to use the app to set AA in Far Cry instead of the control panel, I remember because I complained about it.

The reason our overall results vary so considerably is because we're using AA & AF on both NVIDIA cards, whereas Anand is using just AF on his NV boards. If you look at both of our ATI numbers with 4xAA/8xAF you'll see they results are pretty similar, with mine a little higher (most likely because my CPU is faster).

I'm sure he'll update his graphs accordingly. Right now the ATI boards are doing more work than the NVIDIA cards (4xAA & 8xAF versus just 8xAF on the NV cards) which is why the consistently come out ahead by such a large margin.

mica

[micamica1217]

more of what the patch 1.2 does:






seems like you have to cut the performance of the x800 cards in order to beat them with SM3.0.....

mica

[Cowboy X]

Farcry PS 3.0 at www.firingsquad.com

http://firingsquad.com/hardware/far_cry_sm30/

[Dragonprince]

mica

nice find and a darn good reason not to update to 3.0....

[hkh]

What it comes down to is smooth game play I dont give a crap if a card renders a game at 250fps 40-50 is more then good enough for me I get 52fps on my overclock 9800pro with every thing on high but no AA or AF.

[OBLIVIONLORD]

10fps over the competition is a buy in my book and as far as the price goes.. we will see when its announced and has gone mainstream.

[micamica1217]

Originally Posted by OBLIVIONLORD

10fps over the competition is a buy in my book and as far as the price goes.. we will see when its announced and has gone mainstream.

there are a few things you might be missing:

1) the anandtech review is borked because they never ran any AA on any test.
that's right, they have 4xAA listed, yet they must have ran the tests with 4xAA in the CP and not the game.
I gave a link to the B3D thread that is talking about this...it goes on for at least a few pages.

edit: there seems to be an added edit to the Anandtech review now:

UPDATE: It has recently come to our attention that our 4xAA/8xAF benchmark numbers for NVIDIA 6800 series cards were incorrect when this article was first published. The control panel was used to set the antialiasing level, which doesn't work with FarCry unless set specifically in the FarCry profile (which was not done here). We appologize for the error, and have updated our graphs and analysis accordingly.

2) who is going to run any of the high end 6800 cards with brilinear filtering turned on?
yet in the anandtech review, I show my email reply from Derek that states that brilinear filtering is running.
turning this "optimisation" off will decrease your framerate more.
(by as much as 10fps less)

3)while the new patch1.2 adds SM3.0 suport, it also increases the framerate for NV cards w/o PS3.0, yet decreases framerate across all ATI cards....by as much as 20%
this is how when you then test with PS3.0 that the nVidia 6800 cards are winning over the x800 cards.
(no thanks, I'll stick with patch1.1 and still beat the 6800u with PS3.0 running...look at the graphs I've posted above)

4)looking more at the SM3.0 tests from many sites, I'm thinking that this patch is ether beta at best, or a disgrace.
I can never remember when a patch would not only add a bug, but give you a performance decreace by as much as 20% for one perticular IHV while giving you more frames for the other (and I'm not talking about SM3.0 at this time).

5) seems I'm not the only one who is seeing that SM3.0 is not what is giving the nV cards all the added boost in framerate:

from kmolazz,
""X800 cards performance drops a little with the 1.2 patch and the boost nvidia cards get isn't only from sm3 but also some other improvements in the patch. ""
HERE.

mica