Open beta test: 4770/4890/5850/5870/5970 owners

[BAGZZlash]

I need more beta testers. I implemented extended voltage control options I need tested. There are several groups of testers I need:

1. 4770/5770 users
   I implemented L6788A voltage controller chip support (thanks to LKJask). Many 4770s and 5770s utilize this chip. So if you have a 4770 or a 5770 you wish to volt mod, give it a try.
   
   
2. 4890 users
   As 4890s use a VT1165 to control memory voltage, I implemented memory voltage modding capability for this card. Please try it out.
   
   
3. 5850/5870/5970 users
   GPU and RAM voltage modding should be available now. Please test it in detail.

Please note: This software is beta. You use it more than ever at your own risk!
Again, this software is beta. I need your test results. Please do not just use, be pleased if it works and remain quiet. Communicate! Thanks.
If you have overdrive enabled, please be sure to disable it before flashing any RBE modded BIOS.

Download the beta version here.

 

[DirectorC]

That does it, I'm moving (back) to the red team.

 

[BAGZZlash]

Could you please be a little more specific?

 

[DirectorC]

Could you please be a little more specific?

LOL, sorry:

For the last 11+ years I have used nothing but nVidia cards, except a 9500 Pro in 2003. In a few days I will be going back to ATI

 

[human_error]

ooo nice - gonna try and test this now on my 5970 be back with results soon.

**edit**
when im trying to change the gpu voltage to 1.1625v (max voltage enabled in over volt tool) in the clock settings i am unable to save the bios - this seems to be linked to me only able to enter 4 numbers in the voltage value (not a big deal, but some users may hit it if trying to permenantly overvolt it to ATi recommended voltages) - i get an error screen saying "Unable to write clock settings. This may be due to illegal values set. - Debug information: 'WriteClockSettings' reported error".

**edit 2**

i can modify the memory registers on my gpu0 bios but the option is greyed out when trying to modify the gpu1 bios (not sure if that's intended or not - i assume not?)

here are the 2 stock bios i am trying to modify (the one ending B0 works with the memory registers, the one ending B1 does not).

 

[BAGZZlash]

**edit**
when im trying to change the gpu voltage to 1.1625v (max voltage enabled in over volt tool) in the clock settings i am unable to save the bios - this seems to be linked to me only able to enter 4 numbers in the voltage value (not a big deal, but some users may hit it if trying to permenantly overvolt it to ATi recommended voltages) - i get an error screen saying "Unable to write clock settings. This may be due to illegal values set. - Debug information: 'WriteClockSettings' reported error".

Thanks. Should be a bug, I'll look into it.

**edit 2**
i can modify the memory registers on my gpu0 bios but the option is greyed out when trying to modify the gpu1 bios (not sure if that's intended or not - i assume not?)

here are the 2 stock bios i am trying to modify (the one ending B0 works with the memory registers, the one ending B1 does not).

Yeah, I noticed this before. Should be no bug. Apparently the second BIOS does not program the memory voltage controller as the first BIOS does the job. Are there three or four VT1165s on a 5970? I think all the memory chips are volt-controlled by just one VT1165.

But most important: Does the volt-mod work?

 

[Reefer86]

would over volting the memory on the 4890 mean higher memory clocks? or just higher stable clocks overall? im going to try this now and report back, but i just wanted to know before i go ahead.

 

[BAGZZlash]

would over volting the memory on the 4890 mean higher memory clocks? or just higher stable clocks overall? im going to try this now and report back, but i just wanted to know before i go ahead.

Higher memory voltage should grant higher memory clocks, yes.

 

[human_error]

Thanks. Should be a bug, I'll look into it.



Yeah, I noticed this before. Should be no bug. Apparently the second BIOS does not program the memory voltage controller as the first BIOS does the job. Are there three or four VT1165s on a 5970? I think all the memory chips are volt-controlled by just one VT1165.

There are 4 values on the 1st BIOS, so i assume there are 4.

But most important: Does the volt-mod work?

Flashing the modified BIOSes now - i've changed the gpu and memory registers and the gpu voltage for 3D mode - i assume that's all i need to do?

**edit**

just tried the card with modified volts and clocks, blue screened on windows start so i'll try it with just modified volts.

 

[BAGZZlash]

There are 4 values on the 1st BIOS, so i assume there are 4.

No. These are the registers of one controller chip. Counting both BIOSes, there are three times four values: The master BIOS and its GPU and memory registers and the slave BIOS with its GPU registers only.

Flashing the modified BIOSes now - i've changed the gpu and memory registers and the gpu voltage for 3D mode - i assume that's all i need to do?

You should also match the power play information. The voltages there are kind of a "guideline value". The power play script requests these voltages from the controller and the controller delivers the closest approximation to that. (At least as far as I know.)

 

[human_error]

No. These are the registers of one controller chip. Counting both BIOSes, there are three times four values: The master BIOS and its GPU and memory registers and the slave BIOS with its GPU registers only.


You should also match the power play information. The voltages there are kind of a "guideline value". The power play script requests these voltages from the controller and the controller delivers the closest approximation to that. (At least as far as I know.)

Right - im going to test now with only voltage changes - bios0 has all memory registers set to 1.15v and both bioses have the highest gpu register value modified to 1.162 as well as the voltage in the clock/voltage section (i assume this is the powerplay section) changed to match the new registry value. Testing now, brb.

 

[BAGZZlash]

Right - im going to test now with only voltage changes - bios0 has all memory registers set to 1.15v and both bioses have the highest gpu register value modified to 1.162 as well as the voltage in the clock/voltage section (i assume this is the powerplay section) changed to match the new registry value. Testing now, brb.

Yeah, that's the stuff!

 

[human_error]

Yeah, that's the stuff!

blue screened again on windows boot - i noticed significant display corruption as it started to load my desktop - i'll give her a go with diff volts again but i'll only change the volts a little bit as opposed to max overvolt spec, to see if there's an instability there.

**edit**

just tried to change the gpu register 0x18 (4th one available) to 1.162 (to match the value in the powerplay - whenever i enter 1.162 or 1.1620, click ok then return to the register values it has changed to 1.1625 (which is the max value in overvolt, but can't be matched to the powerplay voltage settings as it is to 4 sig figs after the decimal which cant be saved in powerplay options at the moment) - are the values you can enter there limited to certain voltage multipliers? - nevermind, i tested and can see it is linked to certain voltage multipliers

 

[BAGZZlash]

blue screened again on windows boot - i noticed significant display corruption as it started to load my desktop - i'll give her a go with diff volts again but i'll only change the volts a little bit as opposed to max overvolt spec, to see if there's an instability there.

Wait. You flashed using WinFlash, after success you rebooted, noticed display corruption during boot and then windows crashed into a blue screen. Right?

just tried to change the gpu register 0x18 (4th one available) to 1.162 (to match the value in the powerplay - whenever i enter 1.162 or 1.1620, click ok then return to the register values it has changed to 1.1625 (which is the max value in overvolt, but can't be matched to the powerplay voltage settings as it is to 4 sig figs after the decimal which cant be saved in powerplay options at the moment) - are the values you can enter there limited to certain voltage multipliers?

Yeah. Both values (i.e. those for the registers and those in the power play table) are saved using different formula. The register value is saved in one byte, so it's accuracy/granularity is limited. RBE automatically fits the user's input to this "grid" and matches the input to the closest legal value.

 

[human_error]

Wait. You flashed using WinFlash, after success you rebooted, noticed display corruption during boot and then windows crashed into a blue screen. Right?

No - im using atiflash in a win98 boot partition on my usb stick (cmd only). the only voltages i am modifying are the full speed 3d voltages (which the card does briefly run on windows startup), so that i don't cause instability when in the BIOS so i dont have to get my spare card out. my overdrive is unlocked but disabled at the moment and i'm on the cat 10.1 release drivers.

Yeah. Both values (i.e. those for the registers and those in the power play table) are saved using different formula. The register value is saved in one byte, so it's accuracy/granularity is limited. RBE automatically fits the user's input to this "grid" and matches the input to the closest legal value.

ahh fair enough - i'll now test with the values in both at 1.15 which is one of the few values which matches nicely with the limitation of how many figures can be saved in powerplay, hopefully this will work.

**edit**

still bluescreening on windows startup Only thing i can test now is with only changing the gpu volts or only the memory registries - if you want i can do that? otherwise i've tested at stock clocks and settings other than the voltages, and i'm using voltages i know the card is happy with so im not sure what more i can do.

 

[BAGZZlash]

when im trying to change the gpu voltage to 1.1625v (max voltage enabled in over volt tool) in the clock settings i am unable to save the bios - this seems to be linked to me only able to enter 4 numbers in the voltage value (not a big deal, but some users may hit it if trying to permenantly overvolt it to ATi recommended voltages) - i get an error screen saying "Unable to write clock settings. This may be due to illegal values set. - Debug information: 'WriteClockSettings' reported error".

Fixed that issue. Uploaded the new file so the download link in the first post is up-to-date. Download this new version.

Only thing i can test now is with only changing the gpu volts or only the memory registries - if you want i can do that?

Would be lovely. Maybe you can test both seperately in order to use the method of elimination to find out what causes the problem. Many thanks.

 

[human_error]

Fixed that issue. Uploaded the new file so the download link in the first post is up-to-date. Download this new version.

New issue: it accepts my values and saves however when re-loading a modified file it has rounded the value so instead of 1.1625 it is 1.163 - not sure if that's due to limitations in the bios for the number of values you can store there or if it is another bug.

Would be lovely. Maybe you can test both seperately in order to use the method of elimination to find out what causes the problem. Many thanks.

Okies - testing with GPU voltage changes only (powerplay and registry changes) - memory remains the same.

 

[pantherx12]

'll give this a go : ]

 

[BAGZZlash]

New issue: it accepts my values and saves however when re-loading a modified file it has rounded the value so instead of 1.1625 it is 1.163 - not sure if that's due to limitations in the bios for the number of values you can store there or if it is another bug.

No, that's about the fix. Three decimal places is the maximum precision, so the entered value becomes "rounded" to this.

Okies - testing with GPU voltage changes only (powerplay and registry changes) - memory remains the same.

Looking forward to the results.

 

[human_error]

No, that's about the fix. Three decimal places is the maximum precision, so the entered value becomes "rounded" to this.


Looking forward to the results.

woo! the memory works great - changed the values to 1.15, rebooted and it's working great - my overvolt tool is reporting the correct voltage has been set.

I had a blue screen with the gpu registry and powerplay values both changed - will now test with just the powerplay voltages changed to see if it is the registry for gpu voltages causing the problem.

**edit**

now this is very interesting - with the regestries for both memory and gpu changed to new values it is working fine - the problem seems to be with me changing the gpu voltage in the powerplay options. Just booted with 1.15v on memory, 1.1625v on gpus and the overvolt tool is reporting that the correct values have been set. Ive just run some benchmarks with only the regestry power values for the memory and gpu changed (ie nothing in the powerplay section changed) and it is using the correct voltage values - since i only changed the highest voltage setting for the gpu it only uses the higher voltages when in 3d mode, otherwise it runs normally. Since the bluescreens seem to be originating from the changes to the voltage value in powerplay i would suggest disabling the voltage options there for 5970 users at least, as the registry values do the same job without bluescreening.

 

[BAGZZlash]

Huh. That's interesting, pleasant and curious. Could you test what happens if the powerplay voltage is changed but not exactly to the register value? For example, use 1.1625V for the register and, say, 1.17V for the powerplay setting. Still bluescreen?

 

[pantherx12]

Seems to of locked up after I got it to look for winflash,it found it but now its been doing nothing for 5 minutes, when I mouse over I get the loading wheel ( win 7)

Has it crashed or is this normal and it just takes a bit?

 

[BAGZZlash]

Seems to of locked up after I got it to look for winflash,it found it but now its been doing nothing for 5 minutes, when I mouse over I get the loading wheel ( win 7)

Has it crashed or is this normal and it just takes a bit?

Takes a while. Just cancel and browse to the file.

 

[pantherx12]

Seems I'm not having much luck, telling me there is no ATI card to be found now : [

 

[human_error]

Huh. That's interesting, pleasant and curious. Could you test what happens if the powerplay voltage is changed but not exactly to the register value? For example, use 1.1625V for the register and, say, 1.17V for the powerplay setting. Still bluescreen?

took the working bios with the modified regestries, changed the 3d voltage in powerplay to 1.17v (not the same as regestry value) and it BSOD - exactly the same as before, i notice some graphics corruption on the welcome screen (the blue background one for windows 7) and then it BSODs.