So tonight , I will have extended time with a Radeon 9800

ben6

Regular
I'm going to be spending a lot of time with a Radeon 9800 tonight for an article tomorrow. Any requests?

Edited, because I haven't had the extended time yet lol .
 
ben6 said:
Um, demalion, no.

:( ...

... :idea: :!: ....

[mind trick]
You WILL send demalion your 9800
[/mind trick]


...

:p

Enough sci-fi geekiness, back to hunting up some n-patch benchmark tool download links as well as some other things it might be interesting to test...
 
Using the same drivers, benchmark the 9800pro in comparison to the 9700pro (if available) at equal clockspeeds (325/310). Also, use Ilfrin's pixelbench benchmark if at all possible; it measures the mips rate of the processor. Get it here.
 
Do you have Auqanox 2 or even 1 - run it at High Resolution with AF and FSAA and let us know how a game with heavy use of DX 8.1 shaders performs please.
 
The primary interest I have k is for comparison to the 9700 Pro at the same clock speeds.

There is an n-patch tool at http://www.tommti-systems.de/start.html

Things don't look too promising in terms of improvements from the one benchmark I seem to recall, but maybe something new could be exposed with variety.

There are some other tools there that could be very interesting, including Himself's suggestion with the 9800 at 9700 P clock speeds (which I realize now is possibly what he meant). AA and aniso performance comparisons would enhance this. If you could, a suite of tests for the full 3dmark03 benchmark suite at 9700 P clock speeds would be also be nice...and there seem to be a bunch of data-that-could-be-graphed-in-interesting-ways output options for 3dmark03, and I don't think frame rate over time graphs are widespread for that benchmark yet.

What I'm also interested in is some XOR comparisons with the same exact frame rendered for some pixel shader and maximum aa/aniso, to confirm exactly equivalent output (or point out any slight differences), but I'm not sure of a convenient tool for such comparisons.

For some relaxation, FRAPS + the DX 9 demos (do any have benchmarking functions built in?) for comparison to the 9700 Pro could be interesting too.

Finally, I'll repeat my suggestion to see if these registry keys make any difference in performance (more interesting with the 9800 to test now):

[HKEY_LOCAL_MACHINE\SOFTWARE\ATI Technologies\Driver\0000\atidxhal] (the path under Windows 9x...should make it easier to find under XP), the entries "PVSCode" and "TexStages" set to both "0" and "1" for contrast (assuming the entries do anything at all currently), maybe with the Fillrate Tester application.


Hmm...well, as much as the above as you are willing to do, if you're not going to send it to me... :)
 
FRAPS + C&C: Generals? Maybe just play around with some popular demos for IQ + AF speed purposes, like Freelancer and Pro Race Driver?
 
demalion said:
There are some other tools there that could be very interesting, including Himself's suggestion with the 9800 at 9700 P clock speeds (which I realize now is possibly what he meant).

Well comparisons in general with multiple cards, sure, clock normalized would be a bonus. :)

demalion said:
Finally, I'll repeat my suggestion to see if these registry keys make any difference in performance (more interesting with the 9800 to test now):

[HKEY_LOCAL_MACHINE\SOFTWARE\ATI Technologies\Driver\0000\atidxhal] (the path under Windows 9x...should make it easier to find under XP), the entries "PVSCode" and "TexStages" set to both "0" and "1" for contrast (assuming the entries do anything at all currently), maybe with the Fillrate Tester application.

The entries under XP will have a path under CurrentControlSet/Class/Video/SomeGUID if I recall correctly.

Do you know what these entries are supposed to be about? I have seen them before in regmon logs, no idea what they are for.

If you wanted to [re]install R3DTweak, you could edit tweaker.txt to add them in using the following. Append to the end, doesn't really matter.

HKR,"%twkreg%\DX PVSCode","KeyID",,"%D3D_KEY%"
HKR,"%twkreg%\DX PVSCode","ValueName",,"PVSCode"
HKR,"%twkreg%\DX PVSCode","Category",,"%CAT_DIRECTX%"
HKR,"%twkreg%\DX PVSCode","DisplayName",,"PVSCode"
HKR,"%twkreg%\DX PVSCode","IsStringNumber",,"1"
HKR,"%twkreg%\DX PVSCode","Default",,"0"

HKR,"%twkreg%\DX TexStages","KeyID",,"%D3D_KEY%"
HKR,"%twkreg%\DX TexStages","ValueName",,"TexStages"
HKR,"%twkreg%\DX TexStages","Category",,"%CAT_DIRECTX%"
HKR,"%twkreg%\DX TexStages","DisplayName",,"TexStages"
HKR,"%twkreg%\DX TexStages","IsStringNumber",,"1"
HKR,"%twkreg%\DX TexStages","Default",,"1"

That would automatically sort out the key for you, regardless of OS.
 
nascar 2003, il2 sturmovik (or forgotten battles). basically, what i want to see is if the 9800 can improve my exerience with the applications i use.
 
dude you scared me with the tittle ... but then i noticed the tittle said radeon. I was afraid u were going to use the flow fx for something a little hot and steamy ... yawn got a final tommorow ... yey.
 
Randell said:
4xAA Performance in Neverwinter Nights. Absolutely stinks on the 9700Pro.
if its just 4xAA and greater, its probably (maybe) using AGP texturing. Maybe the game uses more texture space than is left over.
 
Althornin said:
Randell said:
4xAA Performance in Neverwinter Nights. Absolutely stinks on the 9700Pro.
if its just 4xAA and greater, its probably (maybe) using AGP texturing. Maybe the game uses more texture space than is left over.

well thast with the 64meg texture option, I'll can always choose the 32meg textures and see what happens.

How does it perform with 4xAA on a Gf4? Does anyone know?

Dont use Fraps, Fraps misreads NWN, and will gave me a result of 40+fps, whereas the ingame framerate counter was showing 15-20fps, which judging by ots smoothness was correct.
 
Back
Top