ATI Radeon 9500 modifications can be done by software.

[CougTek]

Saw this on Anandtech's news section. Apparently, the next version of RivaTuner will allow to make a Radeon Pro out of a Radeon 9500 64MB and a Radeon 9700 out of a regular 9500 128MB.

Can't wait.

Every day brings new details about remaking RADEON 9500 chips. However, today’s are jus outstanding: there’s a way to transform RADEON 9500 64MB into RADEON 9500 PRO 64MB by software! That’s what Andrew Vorobyev, our video section editor, said:

• "...Besides the hardware way of remaking RADEON 9500 128MB into RADEON 9700 by transistor re-soldering, discovered by our colleagues from Overclockers.ru, we, together with Alexey Nikolaychuk (aka Unwinder), the creator of RivaTuner , hoped to perform the same by software. Essentially, it was to find 4 missing rendering pipelines "hidden" by ATI, as R9500 has 4 and R9700 already 8 of them.
  
  Alexey succeeded in finding the blockade and got through using RADEON 9500 64MB. As you know, this card has 128-but bus, so you can’t make R9700 from it, only RADEON 9500 PRO 64MB. But we’ve checked Soft9700 on RADEON 9500 128MB and, voila, we got a complete RADEON 9700.
  
  So, we can state the 100% software pare-down of R300. This method of remaking will be available for everyone in the next version of RivaTuner..."

The original should be somewhere on this page.

 

[Clocker]

Saw that. It looks sweet and I'd like to do it. I have $155 in Best Buy gift certificates but BB doesn't have a 128MB 9500 (other than a Pro for $219).

I'd like to get the 128MB 9500 and run it as a 9700. How much better than my 64MB 8500LE at 300mhz core /300mhz memory would that be?

C

 

[Mercutio]

Benchmarks I've seen for the 9500Pro put it to within a hair of the speed of a GF4 4600. The GF4 is generally described as "spanking" the 8500 in fanboy graphics card reviews, so it's safe to say it'll be quite a bit faster than your 8500LE.

If only ATI had done a 9500AIW... sigh.

 

[CougTek]

Just a little follow-up...

I know most of you who are interested in modifying their R300 chips can read X-bit Labs just like me, but just in case some newby steps on this thread in a few months and didn't follow the entire story, I have to add two other links to X-bit Laboratories.

First, apparently and unsurprisingly, about 30% of the Radeon 9500 don't work correctly when all 8 pipelines are enabled :

Our colleagues from the on-line store theFastest.ru told us that the modification of Sapphire RADEON 9500 with 128MB graphics memory into RADEON 9700 sometimes turned out a failure. Having carried out this experiment with quite a few graphics cards, they discovered that in 30% of all cases the cards worked unstably after the hardware modifications. Moreover, almost nothing depended on the BIOS version (Sapphire or Hercules RADEON 9700) and the working frequencies: these 30% of the cards worked incorrectly in any case with 8 pipelines enabled. The “incorrectness” implies some kind of “rippling effect” in 3D applications made by the falling out pixels. Each graphics card showed its own picture, and sometimes even the whole sets of pixels were missing on the screen.

[...]

Some RADEON 9500 chips are none other but the “defective” RADEON 9700 ones. During the chips testing, they check whether all its blocks are functioning properly and if some of them do not work, the chip is marked as defective. However, if the defective blocks are inside the pixel pipelines and there are 4 pipelines at the most that do not work well, then these 4 non-functional pipelines can be simply disabled and the chip can become a RADEON 9500.

How could we then explain the fact, that far not all the RADEON 9500 modification attempts end up in a failure?

Well, we think that there are not too many defective chips, first of all. The chips are either OK, or have such problems, which could not be hidden behind 4 disabled pipelines. In this case to maintain the RADEON 9500 sales on a certain level, they have to turn a number of fully-functional chips into RADEON 9500, especially since the production cost of both is the same. As for our experiments with RADEON 9500 128MB and RADEON 9500 64MB graphics cards, which ended up successfully, we seem to have been just lucky to have got these fully-fledged chips.

In fact, the current situation with RADEON 9500 reminds us of the RADEON and RADEON LE case. If you remember, these chips featured a disabled HyperZ unit and were clocked at lower frequencies. The cards based on RADEON LE cost less than the solutions on the fully-fledged RADEON. Most RADEON LE owners who managed to enable the HyperZ unit got a significant performance increase and didn’t have any problems with the graphics quality. However, after enabling HyperZ by some chips, we received different artifacts, which indicated the defectiveness of the HyperZ unit.

So, it is really sad but not all RADEON 9500 based cards appeared modifications-friendly.

It is quite logical. Sad, but not a surprise.

Also, there's no more need to mess with your soldering iron anymore :

X-bit Labs[/url]]Now you do not have to solder, use conductive glue, to flash a new BIOS and so on. A smart software developer has found a way to enable all eight rendering pipelines of the R300 VPU by modifying the drivers. In fact, the main idea of this “mod” is, again, in changing DeviceID of the graphics processor. In all previous cases we either forced the DeiviceID by soldering the appropriate resistors, or modified it in the BIOS. However, since the driver of the RADEON 9500/9700 graphics card reads the ID from somewhere, there is a lot of sense to simply modify the driver itself, what just was accomplished a few hours ago.

Now all you have to do is to change a number of files of the currently installed ATI drivers on your system and test your graphics card. You can get more instructions and the files themselves from this page.

I'll continue to update the thread if other developments occur.

 

[Clocker]

Thanks. For some reason, every time I got to x-bit it seems teh site is down. I appreciate your updates here.....



C

 

[Groltz]

A new version of RivaTuner has been released to accommodate the ATI fanboys...It includes SoftR9700 support.

Added SoftR9700 patch script for Catalyst 6.200 - 6.255, which allow you to enable software locked 8 pixel pipelines on RADEON 9500 based boards and turn any 128-bit RADEON 9500 into RADEON 9500PRO and any 256-bit RADEON 9500 into RADEON 9700PRO. Warning! You must read and accept RivaTuner's license agreement before using this script.

http://www.guru3d.com/rivatuner/

 

[CougTek]

Thanks Groltz. Nice new avatar BTW.

 

[Groltz]

Thanks Groltz. Nice new avatar BTW.

'Welcome, Coug.

I dug up a few...peculiar...avatars yesterday.

 

[Clocker]

I just ordered my 128MB 9500 from Newegg. I'll let you guys know how it works out for me. Apparently, it doesn't work on some cards.....

C

 

[Groltz]

I just ordered my 128MB 9500 from Newegg. I'll let you guys know how it works out for me. Apparently, it doesn't work on some cards.....

Are you going the RivaTuner route, C?

Good luck with your mods!

 

[Clocker]

I just ordered my 128MB 9500 from Newegg. I'll let you guys know how it works out for me. Apparently, it doesn't work on some cards.....

Are you going the RivaTuner route, C?

Good luck with your mods!

Yeah...I don't see any reason not to just use RivaTuner rather than modding the card, physically.

C

 

[Clocker]

I got my Radeon 9500 128MB today. Ordered it from Newegg on Tuesday and I got it Today (Thursday). Damn that was fast. Shipping was free but I checked the box for 'Rush Order Out the Door Today ' for $3.00 because I was hoping to get the card before the weekend. It was well worth the $3!

Anyway, I think this software patch is working perfectly although I'm not sure how to check frame rates in UT2003. I have all the setting maxed out in UT2003 and it seems super smooth to me. I'll look up how to check FPS during game and check back in. Seems like this is a great value!

C

 

[Clocker]

I'm not sure if this is good...

I was playing at 1152x864 with all graphics maxed out as well as trilinear filtering. I twas haging around 50-60fps. I think I have this card running at the default 325mhz core and 310 (620ddr) mhz memory.if RIvaTuner is working right....

C

 

[Groltz]

I'm not sure if this is good...

I was playing at 1152x864 with all graphics maxed out as well as trilinear filtering. I twas haging around 50-60fps. I think I have this card running at the default 325mhz core and 310 (620ddr) mhz memory.if RIvaTuner is working right....

Can you check your UT2003 results against what the hardware sites are getting in their reviews of the 9500 and 9700?

 

[Clocker]

Maybe later. I'm too lazy/busy to bother.

C

 

[blakerwry]

Dont they have a specific Unreal benchmark program....

 

[Clocker]

I saw something out there but it looked like it was for the DEMO version. I'm not sure if the benchmark results would be 100% correct if I used the retail version that I have. I know the textures in the demo are significantly less complex (according to the Atari people).

C

 

[Mercutio]

Well, did it work or not?

Inquiriing minds want to blow another $200.

 

[Clocker]

Yeah, it seems like it worked great. I just have not had time to benchmark but patching the drivers didn't seem to have any ill-effects and I get what I consider excellent fps in UT2003.


I think there might be a benchmark in the UT2003 folder. I'll try to run that and report back.

C

 

[Clocker]

I ran the canned benchmarks for the fly-bys. All benchmarks were run at 1024/768 with all graphic options maxed out. Here are the results with the patched driver and the card running at the standard 9700 speeds (325Mhz Core/310Mhz (620DDR)). I'm not quite sure how to interpret them. They were in the lowframerate.log file:

dom-suntemple
26.721533 / 120.873169 / 630.810547 fps
Score = 120.875282

dm-phobos2
31.902418 / 118.628716 / 622.980835 fps
Score = 118.645287

dm-inferno
12.724050 / 82.877724 / 391.006927 fps
Score = 82.955521

ctf-face3
38.656586 / 147.084564 / 488.500275 fps
Score = 147.165161

ctf-citadel
38.751419 / 112.740929 / 315.421478 fps
Score = 112.869682

dm-asbestos
33.078899 / 155.019485 / 472.802643 fps
Score = 155.115753

dm-antalus
13.883145 / 98.888214 / 402.599731 fps
Score = 98.889816

C

 

[CougTek]

The easiest way to know if your graphic card performs like a 9500 or like a 9700 is IMO to run 3DMark 2001SE build 330 and to compare with the many other results posted everywhere on the Net.

http://www.futuremark.com/download/?3dmark2001.shtml

 

[Clocker]

About 16% faster than a Geforce4 Ti 4600 for $160 shipped:

Sounds like a good deal to me, Merc. I'd go for it!

BTW: THe numbers I presented earlier may not be 100% valid as I have just updated the drivers to the new Catalyst 3.0's, patched them with the soft-9700 patch, and then ran the 3D Mark

C

 

[Groltz]

Your score is 12.89% higher than my highest attained Ti4600 3DMark. We are running the same CPU at about the same speed, right? (2400+ at 2200Mhz)

If the 9500-9700 conversion patch proves to be a long term "no-problem" mod, that is an excellent performance:cost ratio for a video card.

 

[Groltz]

Another good article on this topic can be read here:

http://www.designtechnica.com/sections.php?op=printpage&artid=58



.

 

[CityK]

People are now trying a 8500 -> 9100 conversion of some sorts (aren't they the same thing?). Info can be found on this Rage3D thread...didn't bother reading it, so no idea whats contained in there.

CK

 

[Clocker]

CK-

i read it. Basically, it is all show and no go. No performance improvement. it just shows up as a 9100 card after a BIOS update.

C

 

[CityK]

Thanks, C. Too bad this info didn't come out a week or so ago, you probably could have squeezed Coug for $5 more on the card you sold him :wink:

CK

 

[Mercutio]

I can confirm that there's no practical difference between an 8500 and a 9000Pro. I think I picked up 3 Quake fps between the 8500DV (slower memory timings than a vanilla 8500) and a 128MB 9000Pro.

 

[Groltz]

Check out this tweaked, freaked and heatpiped 9700:

http://www.ocsystem.com/ocenrad97pro1.html

Pretty intriguing...I wish an independant site had a review of it.

 

[Groltz]

.

A newer version of Rivatuner has been released today. Among other things, that includes:

SoftR9700 has been replaced with new SoftR9x00 script. SoftR9x00 supports both Windows 9x/ME (certified for Catalyst 9069, 9076) and Windows 2000/XP (certified for Catalyst 6200, 6218, 6255 and 6275) and allows to perform RADEON 9500 to RADEON 9500PRO/9700PRO as well as reverse RADEON 9500PRO/9700PRO to RADEON 9500 software modification. Reverse modification was exclusively added special for the testers for simulating RADEON 9500 performance on RADEON 9500PRO/9700(PRO) boards.

http://www.guru3d.com/rivatuner/

.