• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Hack 9500 into 9700...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
CizMeyeAzz said:
I was following these threads for awhile and decided to buy a Radeon 9500np 128mb from NCIX last week friday, its a revision 3 board with Infineon 3.3ns memory. Have done the w1zzard software hack and there are no artifacts at all.

Please post your 3dmark fill rate scores. Even if there aren't any artifacts that doesn't mean it was successful... I'll explain shortly.
 
CizMeyeAzz said:
I was following these threads for awhile and decided to buy a Radeon 9500np 128mb from NCIX last week friday, its a revision 3
board with Infineon 3.3ns memory. Have done the w1zzard software hack and there are no artifacts at all. I tried q3, nfs3, ut2003, 3dmark2001 and a bunch of ATI demos including pipe dream and car demo. Memory overclocks to 610 and core maxes out on the Rage3d slider to 360mhz without artifacts. The core produces a ton of heat when overclocked to this level. I find my 3dmark scores to be inconsistent, jumping back and forth between 11500 and 12500 at default speeds. Before the hack I got 10100 points so this is a definate improvement.

CizMeyeAzz

ME -
yes - Radeon 9500np 128mb from overclockers.co.uk
yes- Have done the w1zzard software hack
does it work -yes - but to be real honest I see only one flash
of checks in the dragostic hi-res - I say 99%
no probs in games.

3dmark2001 se
from / to
8730 / 9797
fill single 918/1537
fill multi 1103/2171

no overclocks yet ---

THanks to W1zzard and all his mates

using a tyan 2460 dual xp1600+ MB
 
i did the mod but i didnt do a before and after ; does this show i have 8 pipelines now ?

rad.txt
 
Remember guys (and possibly some gals:p ) that the DX9 demos from ATI are really testing the limits of the card out by using *ALL* the functions originally designed into the card. I have absolutely no artifacts with the 3DMark2001SE, but 3 out of the 4 demos show artifacts using DX9!!

Of course, we're not playing any DX9 games yet, but this mod is definitely not perfect! There probably is a reason they are 9500 cards and not 9700!

Here's another point to add. Maybe the 9500 are missing some other capacitors or resistors on the PCB itself. Does anyone have a 9700 and a 9500 so they can compare? (Although this doesn't really explain why people are also having sucess either...)
 
Well I got my 3d Mark upto 12151.

My fill rates are

Single texture: 1090.7
Multi texture: 1466.2

My RAM on my board maxes out at 303MHz and the GPU can go as high as 378MHz with stock cooling.
 
MeyeAzz....you dont have to remove the transistor. Just bridge 1-2 and since the resistance is less the electrical current will flow through 1-2 and not 2-3. Xbit explained this. And if you read the articles it will also explain why if the soft mod doesnt work the hard mod wont either. Please read xbit.

My god fedex!!!!!!!!!!!!!!!! Get me my new mbob faster!!!!! Once the nforce2 is up its 9500 modding time!!!!!!!
 
AC... that's odd. A successful mod should double the mult-texturing fill rate. (as i oulined in http://forum.oc-forums.com/vb/showthread.php?s=&threadid=157567 )

Shaotai.. You may be on to something. I'm wondering if the default voltage for the 9500 differs from the 9700s. That would take yeild into account in seeing the checkerboard patters. Hmmm.... Hopefully I'll be able to complete the voltage mod on both cards tonight to see if that makes any differance. I'm not willing to believe that this many chips just have flat out broken pipelines. It seems to me that fab processes shouldn't work that way, but then again what the hell do I know. I think it stands to reason tho that one or two pipelines being fudged seems awfully odd... I mean if the failure rate is that bad, then what other parts of the chip could possibly be malfunctioning on any of the cards? How do they test each pipeline to it's full extent? And they'd have to do some programming magic to get the driver or board to figure out *which* of the 8 pipelines are working.. and if they just relied on a certian set of pipelines, that would make for a relatively large ammount of wasted chips with 4 working pipelines in the wrong place. If there's just a hair seperating a 8 pipeline chip from a 4 pipeline chip, what's to gaurentee that all 9700s are working properly? Has anyone heard of a 9700 displaying this behavior?

The other odd thing I noticed is how some cards halfway mod. Sounds like AC's has done that... what with the fillrate of 1500. For a successfully modded card the fillrate should be in excess of 2000. I've got one of these cards ( again, as I outline http://forum.oc-forums.com/vb/showthread.php?s=&threadid=157567 here) These cards don't display any artifacts, but their performance boost is tiny compared to the cards that are fully successful. How can that make sense? Say 6 of the 8 pipelines were working... is the chip somehow dynamically disabling 2 of the pipelines and operating on 6? It's not operating on 4, because some things score far in excess of the standard 9500 w/ 4 pipelines, so what the hell is hapening here?
 
Im sure all the stuff you described can ahppen. But if the 9700's are faulty you simply RMA it. Thats the way it have always been done.
 
I'm curious to know what this hardware difference is, keep us posted.

I just got my card today from BananaPC.com it's a sapphire with infineon ram. Here are some scores:
Defaults/wizzard Driver.
3dmark 10,951/13,188
game1-lo 141/158
game1-hi 61/62
game2-lo 185/258
game2-hi 105/135
game3-lo 163/177
game3-hi 77/79
nature 60/89

fill rate:
single 927/1532
multi 1103/2171

system:
Epox 8k5a2+, XP1600+ @1805mhz(190x9.5) 2.1v watercooled
XMS PC3200 256mb (2-2-2-5) +.10v C:=2x60G Raid 0

I'm not getting any aftifacts yet, but all above was done with DX8.1 on XP Pro w/SP1. I'm downloading DX9 right now and some of the ati demos. I'll post a followup after I goto DX9.

peace.
unloaded

ps: no oc on the vid card yet, after done testing I'll flash bios or try unwinder's tools.
pss: i think only having 256 ram is hurting my fill rates and hi res scores.
 
TruckChase!
Here are my 3dmark scores,

http://service.futuremark.com/compare?2k1=5647130


My fillrate has always been good but my scores are really inconsistent. My 3Dmark scores were much more consistent with my Radeon 8500.


NO LIFE,

Thanks, I checked out the Xbit article, though my concern now is that my Rear Defogger grid repair kit my not be as electrically conductive as the conductive glue they used. If my Rear Defogger repair kit provides more resistance than the resistor, all my 8 pipelines wont open up because electricity flows with the least restrictive path. I guess the only way to find out is to try it.


Cheers,


CizMeyeAzz
 
TruckChase! said:
Shaotai.. You may be on to something. I'm wondering if the default voltage for the 9500 differs from the 9700s. That would take yeild into account in seeing the checkerboard patters.

heh... quoting myself. I found out that at least the core voltage on the 9500 series is the same as the 9700. So much for that theory. Gonna try the volt mod now.
 
Did anyone try only the latest Rivatuner to activate 256bit/8pipes on the R9500np?
If not could you please try out?
 
Grisu4, it is solely dependant on the card if it will work. I have done the Rivatuner and w1zzards way and my card will not get the correct fill rates. The Rivatuner way is super easy, and I wish my card would comply, but it won't on mine.
 
AC Slater said:
Grisu4, it is solely dependant on the card if it will work. I have done the Rivatuner and w1zzards way and my card will not get the correct fill rates. The Rivatuner way is super easy, and I wish my card would comply, but it won't on mine.

Did you try conecting the pins with conductive glue already - or do you still think about RMA?

btw I see that's you're running a Seagate RAID (0 or 1?). Do you use the onboard RAID (SIlicon Image or HPT372?) or a PCI card? How are your HDTach scores?
 
Last edited:
AC Slater said:
Grisu4, it is solely dependant on the card if it will work. I have done the Rivatuner and w1zzards way and my card will not get the correct fill rates. The Rivatuner way is super easy, and I wish my card would comply, but it won't on mine.
did you flah the card with a 9700 bios yet?
 
Back