Rivatuner 12.1 + SoftR9700 released!

Discussion in 'RivaTuner Advanced Discussion forum' started by Hilbert Hagedoorn, Jan 13, 2003.

  1. Agreed. Insulting people is never a constructive approach to ANY situation. Neither does it solve anything nor does it bring any other benefits. All it does is make people angry or sad.

    Having said this, we have such a large amount of people here with so many different systems and configurations, it would simply be IMPOSSIBLE to address every problem every person is having and therefore it is simply impossible for Unwinder to first analyze everybody's problem and then come up with a solution to it.

    Even though we all have different hardware in our computers, IDEALLY, IF we could ALL format our harddrives, make a fresh install of Windows and then go through the exact same steps as described here to do the mod, THEN our reported problems would have a MUCH better reflection on whether the drivers are causing the problem. As it is, people overclock, use Powerstrip and do tons of other things and then blame the mod on the drivers or Unwinder. I'm NOT referring to Pussycat or ectx here, it's just that there are SOME people here who claim things (as opposed to asking) without really understanding the implications of what they are saying. Unwinder gets frustrated about this. If I were in his place, I know I would be pissed myself if someone insisted on things he/she didn't really understand and I KNEW to be false.

    I'm not however, ruling out the possibilty that there could be something in the drivers causing some problems for certain people but then again, it's a proven fact that more than 50% of the problems (maybe even more) on these forums people are having are user related. I wouldn't say that I'm a newbie and yet I too had to spend a couple hours when I bought my Radeon 9500 to get the mod working on Windows XP but I never thought the problem was with Unwinder. In the end, the mod worked albeit I got some artifacts in some OpenGL games.

    The good thing is, Unwinder is still hanging around here which means that he STILL CARES about the people having problems otherwise if he didn't give a crap, why would he even bother replying?

    We should not abuse Unwinder's efforts to help us here by doing our best to analyze or sort out the problems we are having ourselves first (me and Pussycat did hours and hours of testing here to find out that our problems were related to HyperZ) before asking help from him (which will only frustrate him). If we can't figure out our problem, we should at least try to isolate the problem (ie. if you have Powerstip, uninstall it to make sure that it's not the problem).

    To reiterate though, rude behaviour either by us OR Unwinder is unnecessary and senseless.
  2. ectx

    ectx Guest

    Yes, a lot of you guys did many tests that really help me.

    OK. Enough said. Let's get back to business...


    A question for you on your applying the script to your 9700:

    In RivaTuner 2.0 RC12.2, Unwinder made available a "reverse mod" feature:

    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.

    As far as you can tell, do you know what you did is similar to this reverse mod? Also Do you know anyother people that can create artifacts on 9700's?

  3. merullait69

    merullait69 Guest

    did you manage how to deactivate hyperZ functions in OpenGL? the same problem i had in directX i have it now in opengl but this one doesn't seem to react at registry mods.
  4. Yes, you are right. I didn't have any problems in DirectX but in OpenGL I wasn't able to disable HyperZ. The thing is, I returned the card that I got as I was actually seeing some minor artifacts when starting up Soldier of Fortune even when the card was not modded. So on monday, I am ordering a new card and will see how things turn out with it. Sorry I can't tell you personally whether using the registry hack or the program Pussycat posted here to disable HyperZ works for OpenGL or not as at the moment, I don't have a Radeon 9500.

    Still waiting to hear from Unwinder on whether he will release a script to disable HyperZ or not. If he releases the script, then it won't matter if it's DirectX or OpenGL, HyperZ will be disabled with it in all API's.

  5. It's not really similar. I 'soft-modded' a 9700 to a 9700p, and 'unmodded' it just by installing the regular drivers.
    I did not hear of others managing to do the same, however I also never heard of other people trying to do what I did: they just flash the BIOS and are done.
  6. stevestrike

    stevestrike Guest

    Two things:

    It is not nesecary to 'mod' a 9700 card to a 'pro'. The only difference is that the 9700 has a slower clock/memory speed. Again, no need for the mod here.

    Try using Rage3D tweak to disable HyperZ. You can get it here: http://www.rage3d.com/r3dtweak/

    I don't have any artifacts, so I can't test it for you guys.
  7. I tried all the options of Rage3DTweak to discover that virtually NONE of its settings changed anyhing. It seems to be putting registry keys in totally unrelated places and therefore not doing SQUAT (overclocking included). I tested it with Catalyst 3.0, Omega drivers, the Catalyst 2.5 drivers and even the Platinum drivers to check if it could have broken driver support for a particular set of drivers. This was not the case. In the end, I came to this conclusion:

    Rage3D Tweak is designed for Windows 9X and is useless for XP.

    I had on the other hand had some success with the Radeonator software (it asks you what OS you have before you install it and therefore puts the registry keys in the right places depending on the OS you have) BUT that wouldn't disable HyperZ for OpenGL either. It did have the ability to change some setting in OpenGL however so it's far from useless. For instance, it had the ability to force a 16 bit z buffer in OpenGL (but this can be done from the ATI control panel too).
  8. i use rage3d in XP and it works great.
  9. So you are able to confirm that the settings you change are actually having an effect. How do you confirm them? Can you see the changes? Just curious.
  10. juno53

    juno53 Guest

    I have used Unwinder's script and also Wizzards drivers to softmod 2 cards with slightly different results on each card, 1 had the usual checkerboard but the other wasnt as bad and just had small white pixels

    Disabling hierarchicalZ on both cards in direct3d solved the problem 100%
    Still havnt worked out how to disable it in opengl yet though:(

    This makes me strongly beleive Ati bin chips with faulty hierarchicalZ units and use these in 9500 cards, and when we fool the drivers into thinking we have 9700 the driver turns hierZ back on causing the corruption

    So please stop giving Unwinder a hard time about his script causing the problems and hopefully he will help us out with a patch to disable hierachicalZ at a driver level

    PS. Keep up the good work Unwinder Rivatuner has really helped me out on my geforce cards in the past

  11. Well said Juno! My thoughts EXACTLY!!

    Yes, Unwinder might be our only hope for this...
    Let's wait and see. Unwinder looks at these forums so he cares about our problems. I for one believe that he will make a script for us to disable HyperZ/Hierarchical Z (whichever it is! :D )

    I'm still curious as to how that person can get Rage3d tweak to run flawlessly with WinXP. Rage3D tweak has TONS of setting to change and testing whether these setting really work or not would be pretty diffucult IMO...
  12. Molestro

    Molestro Guest

    Viscious Circle

    Well I am one of those unlucky guys with the artifacts with the ATI RADEON 9500 PRO. However I used rTool to turn off HyperZ (used the German version even though I don't speak German Pussycats instructions were adequate). Then I experienced no more artificats, and coincidently no more performance. My chipset still reads 9700 and the ati2mtag.sys is still replaced, yet I only gained about 1% of a performance gain... I hope I'm not wasting everyones time when I now mention that I have a vastly outdated comp: AMD athlon 950 ghz. Is this the problem or should I see atleast somewhat of a significant performance gain from this card? Anyway I tested CS with OpenGL, worked fine. Then I boldly re-enabled HyperZ and everything still worked fine. I'm at a loss, because the card is performing like it did pre-patch, yet it the damned chip is being recognized as a 9700...
    I'm going to attempt to get my friend to lend me his 9700 PRO so I can see if this really is as good as it gets on my PC.
    btw Thx pussycat and Unwinder and all those others who contributed!
  13. You probobly made a typo there but just to make sure, you meant a Radeon 9500 non-pro, right? You know that you can't convert 9500 pros into 9700's because they only have a 128bit bus. Radeon 9500 non-pros on the other hand, made by SOME manufacturers such as Sapphire and 3DConnect have the 256bit bus on the circuit board just like the 9700's and so can be modded into 9700's. This actually has been said at about 10000+ times already probably but I wanted to say it 1 more time neverthless!! :)

    OK, that cleared up, I'm assuming you have a Radeon 9500 non-pro and did the mod and disabling HyperZ disabled your mod.
    This makes no sense really. Disabling HyperZ should have a very marginal performance hit really. You probably wouldn't even be able to notice it. When you tried disabling HyperZ, you must have somehow disabled the pipelines as well...

    Actually, an Athlon 950Mhz WILL be a MAJOR bottleneck for this card. At the moment, even a P4 3.0Ghz with HT doesn't unlock the full potential of a Radeon 9700. I have a P4 1.6A running at 2.8Ghz with my FSB at 175Mhz (175x4=700Mhz effective) and my CPU is STILL a bottleneck for this card. So, then, yes, you might not be seeing much of a change with the mod because your CPU is becoming a bottleneck. Not to say that you have a crappy CPU, it's just that a Radeon 9700 is a MONSTER card is not likely to become a bottleneck to any system anytime soon.

    What I'd recommend is to run the 3Dmark2001 fill rate tests to find out if the mod is working or not. That should tell you.
  14. FinalFantasy

    FinalFantasy Guest

    Molestro, your CPU is 1ghz!!!. That is not bad and you SHOULD notice some performance gain if the soft9700 really worked for you.

    Try to look at the 3dmark2001 results when the videocard is the bottle neck and not the CPU. for example, run 3dmark 2001 at 1280x1024 with 4xAA and 8xAF. With this kind of setting, the videocard will become the bottle neck and not the CPU. Run this test before and after the soft9700 mod. YOU SHOULD NOTICE A HUGE JUMB in your 3dmarks after the soft9700 mod at the above mentioned settings.

    Also look at the multitexture fillrate result in 3dmark2001. THIS IS THE BEST TEST TO USE. Make sure that your videocard and your FSB are NOT overclocked, then run the multitexture fillrate test, you SHOULD get a score around 2200 if you have 9700non-pro.
    REMEMBER: In the multitexture fillrate test, it does not matter if you have P4 3.0ghz or celeron 500mgz or athlon1ghz or 512mbRam or 128mbRam. All of this does not matter, you should always get a score of 2100 to 2200 regardless of your system apecs. You should also get the same score (2100 to 2200) regardless of the resolution or AA or AF levels used in the test.

    Also try to benchmark your card using the codecreatures demo. But you need 512mb of memory ram (you can still run it with 256mb ram, but make sure you have enough virtual memory for windows to use). this demo depends alot on your videocard and not your CPU.
    Make sure that you have the following settings before you run the demo:audio=enabled, resolution=1024x768, AA and AF=disabled. then run the demo. If you have 9700non-pro, you should get a score close to 33fps regardless of the speed of your CPU.
    So for this demo your original 128mb 9500np (without soft9700) will get a score close to 22fps, while any 9700np(with soft9700) or original 9700np will get a score close to 33fps. 9500pro 128mb will get a score close to 29fps. While 9700pro will get a score close to 39fps.
    So again if soft9700 worked for you expect a score aroun 33fps. And if you overclock your card to the level of 9700pro (310mgz memory,325mghz core) you should get the score of 9700pro (which is around 39fps).

    Good luck and I hope this post was not very long.
  15. Re: Two things:

    Yes, there is a need, since my 9700np is overclockprotected.

    It behaved exactly like the 9500p here:

  16. How often do I have to say this here??? I HAD HIER-Z ARTIFACTS ON A 9700 WITH THE SOFT-MOD.

    The 9700 is perfectly OK. So the problem was, at least in my case, with the DRIVERS, NOT with the hardware.

    THAT's what I've been trying to say for the last 10 pages or so.

    PLEASE read them, and THEN try to see if you are still convinced it's a HW problem.
  17. new approach?

    Had anyone with a unmodded 9500 tried to benchmark with HZ enabled and disabled? If disabling HZ drops performance, this would mean that HZ is normally enabled.
  18. FinalFantasy

    FinalFantasy Guest

    Does it really matter??.

    HyperZ is disabled in ALL 9500 non-pro without any exception by default. I hate to think about it, but this means that ATI was decieving their customers when they claim that 9500np features the hyperZ bandwidth- saving technology.

    This helps to explain the horrible performance of 9500np at high resolutions compared to the 9500pro.

    Those hyperZ-related artifacts are hardware problem and not software problem. So everybody, give unwinder a break. I cannot understand why "pussycat" got artifacts on his 9700np. But still, the fact that people experience artifacts with the hardware mod and the software mod means that it is a hardware problem.

    Let us face it: Unwinder may be able to come up with a script that enables the 4 extra-pipelines without enabling the hyperz unit and thus eliminating the artifacts. But does it really matter a lot??????

    Without the HyperZ technology, your modded card will be NO WHERE close to 9700np performance.

    The bottom line is: Unwinder gave us soft9700, it worked for most of us and we are thankfull. If it did not work for you, then try to replace your 9500np with another one that might work or maybe you should better buy 9500pro instead. Unwinder is probably working on a script to fix this, but he may or maynot succeed. And even if he succeeded, do not expect any modest performance gain against your unmodded 9500np. Your card will probably be even weaker than the 9500pro performance.

    We should really stop wasting time in the forums trying to find a solution for the artifacts, because if there was an easy solution it would have been found long time ago.
  19. Molestro

    Molestro Guest

    Must be blind

    I missed the whole deal about non-pros only working :p
    That would explain a whole lot, seeing as how it was indeed not a typo :) But let me get this straight if I may: if I get my hands on a 9500 non-pro 265-bit can I use soft9700 to get it to perform as good as the 9700 non-pro or the 9700 pro (providing I overclock to the pro's appropriate frequency).
  20. Unwinder

    Unwinder Moderator Staff Member

    Likes Received:
    Really? Maybe you should read the thread stevestrike is referring to first? The user is free to ask stupid question, the user is free to ignore my thoughts on it, then the user is free to start crying that I refused to look at his problem. And after all that, the user is free to insult me and call "Arrogant p*$#" when I said that it’s a problem of his knowledge?! I’ve replied in the same way and I don’t think that I’m wrong. The best thing I could do was to close the thread at all after that man appeared here with his claims. And I’m afraid that I’ll do it if at least one more boy with the same manners will appear here again. Do you want to get such reaction from me? You'll get it.

    I’ve explained that guy that the patch cannot work partially, the patch cannot boost fillrate tests only and leave the rest tests in the same application unchanged, it’s stupid to talk about hypothetical problem in the script that can cause such effect. Just stupid. If fillrate is boosted but 3Dmark performance doesn’t increase as expected, it can be caused by the driver’s settings only and it cannot be patch script related issue (vsync or something like that, you’ve mentioned it too BTW). Each man who is able to think must understand it. If he doesn’t understand simple things and continues to talk about the ‘bugs’ I refused to look at, it’s his personal problem.

Share This Page