rec.autos.simulators

Eric....

In the thread below on the Sears track graphics....you mention turning off
"specular highlights" in OGL....when one wants to do a fair and balanced FPS
comparison between D3D and OGL in N2K2.  I don't see where that feature is
in the rend_ogl.ini.....or on the OGL "advance tab" for the 29.80
WHQL's....or in  RivaTuner.  Can you turn off "specular highlights" and if
so, where?

Thanks....

Tom

Tom, it's in-game on the graphics options page...
--
Biz

"Don't touch that please, your primitive intellect wouldn't understand
alloys and compositions and,......things with molecular structures,....and
the....." - Ash

Don't think so....not in my version of N2K2, at least.  Unless it goes by
different name?  All of the FX checkboxes impact FPS in both OGL and
D3D....so I don't think the setting for specular highlights is there.

TP

I don't think its in NR2002. It was in N4 IIRC.

MadDAWG

I thought reflections was the same thing, but apperently not, it's in the player.ini file

[Effects]
debris=1                                 ; Debris on or off
dirt=1                                   ; Dirt on or off
dust=1                                   ; Dust on or off
fog=1                                    ; Fog on or off
lightmaps=1                              ; Lightmaps on or off
reflections=0                            ; Reflections on or off
shadows=1                                ; Shadows on or off
skid_marks=1                             ; Skid marks on or off
smoke=1                                  ; Smoke on or off
sparks=1                                 ; Sparks on or off
specular=1                               ; Specular highlights on or off

--
Biz

"Don't touch that please, your primitive intellect wouldn't understand
alloys and compositions and,......things with molecular structures,....and
the....." - Ash

It is under graphic options under your "options" screen.

Kerry

Doh!!

Kerry

Believe this is from N4, not N2K2

Sorry, I had forgotten that this was combined with environment mapping
in NR2002 as a single "reflections" option.  So for D3D the
"reflections" graphic option controls just environment mapping, while in
OpenGL it controls both environment mapping and specular highlights.

- Eric

I can't beleive I actually pulled up the wrong ini file, when I knew I needed the one from N2002.
SO it is in the reflections setting now... as Eric verifed.
--
Biz

"Don't touch that please, your primitive intellect wouldn't understand
alloys and compositions and,......things with molecular structures,....and
the....." - Ash

Okay.....thank you Eric.  That would explain why turning "reflections"
on/off impacted the FPS in both OGL and D3D.....only it was a greater impact
in OGL (by about 20%).  So.....to properly compare OGL FPS to D3D
FPS......."reflections" should be turned off......if I understand the issue?
That......(before I've even tested for this) should give the "edge" to OGL
for FPS.....if I understand the issues properly.  Also......if....and this
is a big "if" that I want to confirm with my own track graphic
files......but if OGL in 16 bit has nearly as good of a subjective graphic
quality as D3D in 32 bit (which I did think was the case in N4....but not in
N2K2).....then the graphic quality/FPS comparison should give the clear
"winner"....to OGL.

I'm going to test this tonight on a new "race box" I just built with a
GF4-ti4600.....but I want to install the 29.80 WHQL's on that system first.

I'll report back under this thread....probably by tomorrow
morning....Pacific.

Thanks Eric......(and all the others)....for the info.

Regards,

Tom

PS:  Note to anyone else who wants to do some of this comparison
yourself......here's a word of caution:  When you run the "3D Config"
program (to change your res/color depths and renderer mode)....you need to
go back into the "rend_D3D.ini" file and reset the "anistropic=0" entry each
time.  The 3D Config.exe will reset this value to what it thinks/calculates
will work for your card.  You don't want to have any number in there but
"zero"....for these FPS comparisons.  If you forget to do this....it will
distort your numbers.  Also....those of you running WindowsXP....I think you
are going to run into hitting a low-headroom FPS because of the inability
for WinXP to properly set the refresh rate "in game"......to anything above
75 or 85....not sure which.  I don't run WinXP (still using
Win98SE....partly for this reason)....so I'm not sure of this still being a
problem with the 29.80 WHQL's....but you should check it out because I think
it still is.

Another quick question for Eric:  Is it possible that I will hit a "max FPS"
in N2K2 (assuming there is one....at least theoretically there should
be.....at something near the physics engine max FPS....or no?)......say at
1024x16 in OGL with this GF4-ti4600 (on a 1.8gig...oc'd over 2.0)?  If there
is a calculated max FPS....can you tell me about what it is?  Its okay if
you can reveal that....I understand that could be kind of proprietary
info....no problem.  It would be good to know if there is a max FPS though.

Yes, turning off reflections will "level the playing field" such that
D3D and OpenGL use the same feature set, and yes turning off reflections
will likely give a larger percentage increase in OpenGL since it was
doing more work than D3D with them enabled.  However even before you
turn reflections off, OpenGL should still have a performance edge as the
rendered scene increases in complexity.  For example, D3D and OpenGL
might give identical framerates sitting in the pits during a testing
session, but if you make the same comparison while starting at the back
of a 43 car field you might find that D3D has lost 40% of that testing
score while OpenGL only has lost 20% (I'm making up numbers, but you get
the idea).

Yes there is framerate cap, limited by max_grabs_per_update_w_track_view
in the [Task] section of your core.ini.  Basically this limits the
number of frames that will be drawn per tick (the engine's base unit of
time which corresponds to 1/36th of a second).  I believe the default is
4, which would cap the framerate at 144.  Assuming your machine is
pegging the144fps limit, I don't think there would be any harm to
increase that number (I'm in no way responsible however if the game
fdisks your system or something as a result of that change).

You'll obviously also want to ensure that you're not limited by v-sync
when doing these tests.  The 1.1.0.2 patch disables it automatically in
D3D, but you'll still need to do it at the driver level for OpenGL.

- Eric

though.

Eric...See my comments below ("Pabst")...........

Great info....thanks, Tom.

(Pabst):  I figured the "Atlanta Pits" bench would not be totally
accurate.....scene complexity is fixed and there's little cpu load without
AI cars.  But...you have to start a "bench" and stick to it.....and I've
been using the "Atlanta Pits" bench since early N4 (and the historical data
is useful sometimes when I'm troubleshooting one of our older "race boxes"
in our rental fleet).  I do however, now want to add another "benchmark"
setup......soemthing that accounts for the scene complexity and CPU load
from AI (as you suggested).....any suggestions on something to use......in
addition....to the "Atlanta Pits" bench?  I was thinking of taking a
high-graphic load track (maybe Sears...or Watkins), setting a full AI field,
starting at the back....and taking the average FPS for the pace lap.  Do you
think that would be "max" stress......for a second "benchmark" to use?  Your
suggestions would be welcome.  After I establish this new benchmark
setup.....I'll test the OGL vs. D3D FPS hit.....and express it as a
percentage of the "Atlanta Pits" bench FPS number too.  That should put some
imperical data to your suggestion that OGL's "load hit" is less than D3D's.

(Pabst):  That's great info, thanks Eric.  "Yes"...its pegging the 144 (see
my other post this morning under this string).  LOL....and "auto-Fdisk"
feature would have been discovered.  My systems could use a good an old
Fdisk...now and then! ....lol.....  I don't see any value in changing the
default "4" max grabs....unless its for pure benching work and the max FPS
is being hit.  I doubt under "playing conditions" with "reflections" turned
back on......and full AI fields....that too many machines even today...would
be exceeding (or capable of exceeding) 144 FPS.  But...we are probably not
far from the day that will be a reality.  I assume Papy will be doing some
testing of some high-end boxes to see what happens with higher values for
the max-grab, though?  Don't they let you guys into the CIA computers on
Tuesdays.....so you can play around with high-end stuff.....and have a good
"office race" together? ....lol....  Just kidding!

(Pabst):  Yes...got that one covered and since I'm using Win98SE....there
shouldn't be a problem with insuring vsync=0

Here's the results of the benchtesting I did last night to get a frame
comparison (balanced and equitable) between the Papy OGL and D3D renders for
N2K2:

***System #1 Specs/Settings***
Specs:  K7-Master, 1.0 AMD (at 1466), 512 Ram (PC2100) - 416 "free ram" at
test run, GF3 (ASUS 8600 at default clks with 29.80 WHQL's), SBL X-Gamer,

Video Card Settings (System #1):  D3D = all 29.80 defaults with LOD at 0.0,
vsync=off, aa=off.  OGL = all 29.80 defaults with LOD at 0.0, vsync=off,
aa=off.

N2K2 Settings:  All sliders max, all "detail" max, Reflections=off,
anistropic=off, *core.ini* at defaults except
[FileLRUCache]/CacheSize=14194304, *player.ini* at defaults except
[GraphicOptions]/Texture_quality=100, *rend_d3d.ini* at defaults except
AnisotropicLevel=0, MipMapLODBias=-2.00, *rend_ogl.ini* at defaults except
AnisotropicLevel=0.

Test Results ("Atlanta Pits" Benchtest) FPS:

                                            **OGL**            **D3D**

1024x16                                144*                      133
1280x16                                119                        103
1600x16                                  84                          70
*******
1024x32                                144                        130
1280x32                                108                          96
1600x32                                  72                          64
 * = max FPS

***System #2 Specs/Settings***
Specs:  K7-Master, 1.0 AMD (at 1466), 512 Ram (PC2100) - 409 "free ram" at
test run, GF4 (PNY ti4600 at default clks with 29.80 WHQL's), SBL X-Gamer,

Video Card Settings (System #2):  D3D = all 29.80 defaults with LOD at 0.0,
vsync=off, aa=off.  OGL = all 29.80 defaults with LOD at 0.0, vsync=off,
aa=off.

N2K2 Settings:  All sliders max, all "detail" max, Reflections=off,
anistropic=off, *core.ini* at defaults except
[FileLRUCache]/CacheSize=14194304, *player.ini* at defaults except
[GraphicOptions]/Texture_quality=100, *rend_d3d.ini* at defaults except
AnisotropicLevel=0, MipMapLODBias=-2.00, *rend_ogl.ini* at defaults except
AnisotropicLevel=0.

Test Results ("Atlanta Pits" Benchtest) FPS:

                                            **OGL**            **D3D**

1024x16                                144*                      144*
1280x16                                140                        122
1600x16                                105                          89
*******
1024x32                                144*                      144*
1280x32                                130                        118
1600x32                                  94                          87
 * = max FPS

Remember....the "Atlanta Pits" benchmark does not take into account CPU load
(form AI cars) or scene complexity.

Hope this is helpful to some of you.....in figuring out what  N2K2 NV
renderer you want to use.

Regards,

Tom