My interesting 6800nu softmodding day

Vallybally

Senior member
Oct 5, 2004
259
0
0
Woah, what a long day. I just received my new BFG 6800 OC (128mb) yesterday and I spent about 12 hours (up till 4am last night) trying to tweak the thing for extra performance using the excellent RivaTuner 15.2. Here's my adventure, struggles and story:

First off, I'm not an OCer. Not on CPU/FSB, not on video card. Maybe down the road, but right now I tend to play things safe. Anyway, I wanted to see the performance increase from my old PNY GeForce FX 5600 Ultra to my new BFG GeForce 6800 OC. Percentage wise, the initial results showed a very strong upgrade, but performance wise (score and FPS in 3DMark05) it was a tad lower than I had expected. I realize the new '05 version is very tough and strenous, designed to last a couple of years, but I still read most ppl were getting higher than I got.

My stress tests involved an average from a set of 3 consecutive (reboot each time) 3DMark05 and 3DMark03 runs. My sequence of improvements were: 5600u --> 6800nu stock --> 6800nu stock with updated mobo BIOS --> 6800nu softmodded.

3DMark2005 ? PNY GeForce FX 5600 Ultra = 436
3DMark2003 ? PNY GeForce FX 5600 Ultra = 3473

3DMark2005 ? BFG GeForce 6800 OC = 2560 (was hoping for slightly more)
% increase (from 5600u): 487%!
3DMark2003 ? BFG GeForce 6800 OC = 9196
% increase (from 5600u): 165%

3DMark2005 ? BFG GeForce 6800 OC ? updated Abit BIOS = 2589
% increase (from 6800 stock): 1%
3DMark2003 ? BFG GeForce 6800 OC ? updated Abit BIOS = 9218
% increase (from 6800 stock): 0%

3DMark2005 ? BFG GeForce 6800 OC ? updated Abit BIOS ? RT 15.2 softmod (16/6) = 2840
Failed (artifacts)
3DMark2003 ? BFG GeForce 6800 OC ? updated Abit BIOS ? RT 15.2 softmod (16/6) = 10030
Failed (artifacts)

Uh oh, it looks like I was one of the unlucky 50% where I couldn't successfully unlock the 4 extra pipes and 1 extra vertex shader. Oh well, np let's just roll it back to its default 12/5 and be happy. Problem, doing this didn't really clear the artifacts. Were they there initially before softmod? I don't remember, but highly doubt it. Well, let's try to completely uninstall the 61.77 drivers and also use Driver Cleaner and reinstall a new set of drivers the 66.81. Success! Not only are the artifacts cleared up, but I seemingly can now successfully unlock the full 16 pipes and 6 vertex shaders with no ill consequences (artifacts)! The increased results for 16/6 seem normal, but the amazing thing is that the new 66.81 drivers themselves seemed to increase 3DMark05 results on my stock (12/5) card by a whopping 751 points from 2589 to 3340! Here are the results:

3DMark2005 ? BFG GeForce 6800 OC ? updated Abit BIOS/66.81 ? RT 15.2 softmod (12/5) = 3340!
3DMark2005 ? BFG GeForce 6800 OC ? updated Abit BIOS/66.81 ? RT 15.2 softmod (16/5) = 3608

3DMark2005 ? BFG GeForce 6800 OC ? updated Abit BIOS/66.81 ? RT 15.2 softmod (16/6) = 3719
% increase (from 6800 stock): 45%!
3DMark2003 ? BFG GeForce 6800 OC ? updated Abit BIOS/66.81 ? RT 15.2 softmod (16/6) = 10123
% increase (from 6800 stock): 10%

My personal conclusions:
-If you start this somewhat intimidating journey of softmodding (or OCing for that matter), don't give up too quickly.
-At least for me, the 66.81 drivers seem much better and more stable than the 61.77 drivers.
-If I ever plan to OC down the road, I can likely squeeze an easy extra 10% overall performance with some light/safe OCing.
-I should be ready for EQ2 and anything else that comes up for the next couple of years.
 

imported_Bleh

Senior member
Sep 30, 2004
433
0
0
Ya I got the same card with everything unlocked (16/6) and the 66.81 increased my mark05 score by about 900 to like 3900 or something like that.
 

usedink

Junior Member
Oct 11, 2004
9
0
0
congrats to u :)

My leadtek one didn't manage to work... and my 3dmark scores were not boosted with forceware 66.81 :p

Btw, can i run at 12 / 6 ?? wat does the vertex thingy add ?? how do i test it ?
 

Travis6586

Member
Aug 12, 2004
106
0
0
I've got a 6800 GT, not a 6800 NU, but the 66.81 Forceware drivers boosted my score ~850 more points. Really happy with the added performance in Doom 3 too.
 

Vallybally

Senior member
Oct 5, 2004
259
0
0
usedink, you should see a small performance if you do 12pipes/6vertex and yes, you can select that option.

Travis, I hope that these are indeed real world(game) results and not just optimized drivers intended just to boost 3DMark scores (ie, sleezy marketting).
 
Oct 19, 2000
17,860
4
81
Originally posted by: Vallybally
Travis, I hope that these are indeed real world(game) results and not just optimized drivers intended just to boost 3DMark scores (ie, sleezy marketting).
It's been universally known for a while that the 65.** series and up give a lot more performance to 3DMark05. I've yet to hear anything about this being an "optimized" driver code for 3DMark, but I also gained somewhere between 500-800 marks switching from 61.77 to 66.81.

I also hope this is a marked improvement over all DirectX 9 games, but I'm too lazy to test :).
 

Vallybally

Senior member
Oct 5, 2004
259
0
0
I guess game benchmark programs would be a good start. Isn't there like a FarCry and Doom3 benchmark program that measures FPS?

If these 66.81 drivers are so good (any known issues?), why doesn't nvidia make them official on their website?
 

ScrewFace

Banned
Sep 21, 2002
3,812
0
0
It seems both ATI and nVidia card owners got a fair boost from new drivers. I have a Radeon X800 XT PE oc'ed to 552/1194 and with the Catalyst 4.9 drivers I got 4,600. With the beta 4.10's I got 6,100!!! A 1,500 point increase. Unbelievable, isn't it?:)
 

Travis6586

Member
Aug 12, 2004
106
0
0
Originally posted by: Vallybally
I guess game benchmark programs would be a good start. Isn't there like a FarCry and Doom3 benchmark program that measures FPS?

If these 66.81 drivers are so good (any known issues?), why doesn't nvidia make them official on their website?

Yes, and in Doom 3 I gained 3 FPS. In Far Cry, I gained 5 FPS. These drivers (66.81) were from NVIDIA's FTP server. I don't know why they're not listed on the website yet. Maybe their awaiting some sort of official certification? Who knows!
 

Slacker

Diamond Member
Oct 9, 1999
8,623
33
91
Your '05 scores seem low, what are the specs for the rest of your system.

With a p4 3ghz 1gb pc3200 dual ch. and pny 6800gt using the 66.81 drivers 3dmark05 scores 5143
 

Tyrant222

Senior member
Nov 25, 2000
802
0
0
Hey I just got my BFG 6800 OC today, softmod seems to work. 3dmark 2k3 and 2k5 have run multiple times with no artifacting. I gained about 1k marks in 2k3 and 400 marks in 2k5.

Although, I cannot seem to overclock using rivatuner. Even moving the slider down to decreese clock speed still causes the test to not pass.

I've heard others with the same problem. Im just looking for anyone whose experimented with this problem and found any ways around it.

Once again I am completely arfifact free.

EDIT: i disabled the test and seem to be ok at 750 mem now, will be increasing over the next few days.
 

Vallybally

Senior member
Oct 5, 2004
259
0
0
Btw, I did not see huge artifacts, but is some extra sparkles or white dots in the Tree man (while bug flies in) considered artifacts? I am not even certain.
 

jiffylube1024

Diamond Member
Feb 17, 2002
7,430
0
71
Originally posted by: ScrewFace
It seems both ATI and nVidia card owners got a fair boost from new drivers. I have a Radeon X800 XT PE oc'ed to 552/1194 and with the Catalyst 4.9 drivers I got 4,600. With the beta 4.10's I got 6,100!!! A 1,500 point increase. Unbelievable, isn't it?:)

Well, that was because of a known issue where ATI cards weren't working with the extra 128MB on 256MB X800 series cards.

Congrats on the kick@ss score though - wow, that's higher than my Mark03 score with my 9800 Pro!
 

302efi

Golden Member
Jul 16, 2004
1,539
1
81
Originally posted by: Tyrant222
Hey I just got my BFG 6800 OC today, softmod seems to work. 3dmark 2k3 and 2k5 have run multiple times with no artifacting. I gained about 1k marks in 2k3 and 400 marks in 2k5.

Although, I cannot seem to overclock using rivatuner. Even moving the slider down to decreese clock speed still causes the test to not pass.

I've heard others with the same problem. Im just looking for anyone whose experimented with this problem and found any ways around it.

Once again I am completely arfifact free.

EDIT: i disabled the test and seem to be ok at 750 mem now, will be increasing over the next few days.

i had that problem with rivatuner for days until I figured this out :

I also thought my 16x1 pipes mod was "arfifact free"...but everytime I tried to over clock it...it would fail ...even when I under colcked it, it would fail.....BUT..when I changed the pipes back to stock it would pass and give me about 380/801...So my findings were, even though it seems that the 16x1 pipeline mod worked with no artifacts, it didnt...:(