Overclock.net › Forums › Graphics Cards › AMD/ATI › Software for R9 285 bios edit
New Posts  All Forums:Forum Nav:

Software for R9 285 bios edit - Page 53

post #521 of 1079
Quote:
Originally Posted by mynm View Post

**Not a problem.thumb.gif
**Do you have java installed?, perhaps you have an old not compatible verion. Is just click on AtomBiosReader.jar.
**Yes, internal voltage offset will solve VDDCI rising, but I don't know how to do an internal voltage offset, I think that it can be done via BIOS VoltageObjectInfo table http://www.overclock.net/t/1561372/hawaii-bios-editing-290-290x-295x2-390-390x/2530#post_25102436 , but I don't have tested it. So the only way I know to lower VDDCI is via usVddgfxoffset.
**I see that not all 280, 380 and 380x have the same NCP81022 controller, I see one with an IR3567B controller
No java installed haha so I'll do that. So I installed Java runtime 1.6.0 x86 and x64... Says cannot find amd.gui.atombiosreader .. program will now exit...
So how would I proceed with the AIDA64 i2c dump to determine my voltage controller type?
I'm sure it's probably the same as other 380Xs so I'll have to find out.

Try this and send back brother. I'll test it.

TongaEdit.zip 42k .zip file

I see...
ATOM_POWERPLAY_SCLK_DEPENDENCY_OBJECT_1
usVddcOffset value : 65288 (Meaning the dynamic known and monitored max of 1.219v is what 65288 stands for) I think there is further significance to this number 65288.. it doesn't make much sense but 1306 in the table, actual is 1218 so... 1306 - 1218 (hwinfo reports 1219 but +/- 1 in hwinfo is common) = 88 ... meaning the offset of 65288 is -88mv from 1306 so 1306 is the max voltage respectably.

I've tested various different changes but I don't know what to do with 65288 to make gains, I changed to 65289 insta-green screen 3d load.. changing to 0 .. instant black screen under 3d load .. so back to 65288 .. changing 1306 up to 1394 .. 88 higher than 1306 so the offset would default to 1306.. no go .. I'm stuck until I find out more.

Thanks again sir, I very much appreciate it. It seems changing that value in hex would be in location A65C before value of 00, after value of decimal value of 100 would be value 64 in hexadecimal. Save in Hxd. Open in TongaBiosReader, save, reopen to make checksum ok. Good to go huh?
Edited by chris89 - 8/30/16 at 3:11pm
post #522 of 1079
Quote:
Originally Posted by chris89 View Post

Warning: Spoiler! (Click to show)
No java installed haha so I'll do that. So I installed Java runtime 1.6.0 x86 and x64... Says cannot find amd.gui.atombiosreader .. program will now exit...
So how would I proceed with the AIDA64 i2c dump to determine my voltage controller type?
I'm sure it's probably the same as other 380Xs so I'll have to find out.

Try this and send back brother. I'll test it.

TongaEdit.zip 42k .zip file

I see...
ATOM_POWERPLAY_SCLK_DEPENDENCY_OBJECT_1
usVddcOffset value : 65288 (Meaning the dynamic known and monitored max of 1.219v is what 65288 stands for) I think there is further significance to this number 65288.. it doesn't make much sense but 1306 in the table, actual is 1218 so... 1306 - 1218 (hwinfo reports 1219 but +/- 1 in hwinfo is common) = 88 ... meaning the offset of 65288 is -88mv from 1306 so 1306 is the max voltage respectably. So to achieve a gain on the offset, would be possible to change 65288 to 65200 0 offset... for 1306mv biggrin.gif

Thanks again sir, I very much appreciate it. It seems changing that value in hex would be in location A65C before value of 00, after value of decimal value of 100 would be value 64 in hexadecimal. Save in Hxd. Open in TongaBiosReader, save, reopen to make checksum ok. Good to go huh?

You are welcome.thumb.gif

For i2c dupm try this better: https://www.youtube.com/watch?v=_UaRpL0Jjz8

Edited: This how to tump i2c with aida64: https://www.youtube.com/watch?v=lVlV0u2QpNI

Edited: If you have voltage control by Afterburner you can see your voltage control chip like is explained here:
Quote:
Originally Posted by gupsterg View Post


You need to do some investigation into what is the voltage control chip on you card? for example if it's NCP81022 (Sapphire R9 285 Dual X TPU review).

Next get info on the chip.
Quote:
2.1.9. NCP81022 voltage controller

NCP81022 is single-loop controller, which can control only one voltage output. NCP81022 supports voltage control in offset form only. NCP81022 supports output voltage monitoring in VID readback form, which means that it can report target programmed voltage only.
This controller can reside in 20h - 27h device address range only. The controller is identified by 2-byte register 99h (must be equal to 001Ah) and 2-byte register 9Ah (must be equal to 1022h). The controller has no model specific settings.

Quote from, also perhaps view datasheet from manufacturer.

Now this chip can only control 1 voltage, now to know if this is what MSI AB is using click on the "i" icon, example below is my Hawaii card.





Are you able to get AiDA64 SMBus dump on your Tonga card?

There isn't a significance to 65288 it is only a number for identificate DPM voltage for EVV voltages. I have tested to change the number but it give me black screen, so don't do it.

usVddcOffset isn't an Vddc Offset.

usVddcOffset in location A65C It is 00 00 only for DPM0 because it isn't an EVV voltage, if you change it to 64-->64 00 (number have to be reversed), vddci for DPM 0 will go to +100mv from 900mv to 1v, because core voltage for DPM0 is 900mv.

For example, usVddcOffset for DPM7 is at A6A9 is cd ff -->ff cd (reversed)--> -51 in word mode, is -50mv. If you whant to change it for example to -250mv DEC value you have to go windows calculator with programmer mode and word mode (default is qword), put -251 (is always +1) and you will get FF05 HEX value-->05 FF (reversed), and if you have 1.350v for core at DPM7 you will get 1.1v for VDDCI.

Don't change it if you don't understand it.

Edited:
Quote:
Originally Posted by ducegt View Post

so I actually increased VDDCI +25mV over stock and what happened is my VDDC voltage went up from 1.2v maximum to 1.225v.

Are you talking about VDDCI instead of VDDC here? I have done "millions" of tests and I don't see that VDDC increase.
Edited by mynm - 9/1/16 at 3:59am
post #523 of 1079
Quote:
Originally Posted by mynm View Post

Are you talking about VDDCI instead of VDDC here? I have done "millions" of tests and I don't see that VDDC increase.

I am mistaken. I installed the new version of Trix, had upped my voltage, and had restarted my computer thinking Trix wouldn't be activated upon restarting, but it was and without any icon in the system tray. I saw more voltage because trix overvolt was in effect.
Gaming Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
Intel 7700K ASRock Z270 K6 Gaming Fatality RX VEGA 64 TridentZ 3600 15-15-35 
Hard DriveCoolingOSMonitor
Samsung EVO 850 Corsair H115i Windows 10 Nixeus NX-Vue24a 
KeyboardPowerCaseMouse
Corsair K70 Rapidfire  Corsair RM850X Corsair 600Q Logitech G303 
Mouse PadAudio
Steelseries Qck+ Integrated  
  hide details  
Reply
Gaming Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
Intel 7700K ASRock Z270 K6 Gaming Fatality RX VEGA 64 TridentZ 3600 15-15-35 
Hard DriveCoolingOSMonitor
Samsung EVO 850 Corsair H115i Windows 10 Nixeus NX-Vue24a 
KeyboardPowerCaseMouse
Corsair K70 Rapidfire  Corsair RM850X Corsair 600Q Logitech G303 
Mouse PadAudio
Steelseries Qck+ Integrated  
  hide details  
Reply
post #524 of 1079
Quote:
Originally Posted by ducegt View Post

I am mistaken. I installed the new version of Trix, had upped my voltage, and had restarted my computer thinking Trix wouldn't be activated upon restarting, but it was and without any icon in the system tray. I saw more voltage because trix overvolt was in effect.

OK thanks thumb.gif
post #525 of 1079
Quote:
Originally Posted by mynm View Post

You are welcome.thumb.gif

For i2c dump try this better: https://www.youtube.com/watch?v=_UaRpL0Jjz8

Edited: This how to dump i2c with aida64: https://www.youtube.com/watch?v=lVlV0u2QpNI

Edited: If you have voltage control by Afterburner you can see your voltage control chip like is explained here:
There isn't a significance to 65288 it is only a number for identificate DPM voltage for EVV voltages. I have tested to change the number but it give me black screen, so don't do it.

usVddcOffset isn't an Vddc Offset.

usVddcOffset in location A65C It is 00 00 only for DPM0 because it isn't an EVV voltage, if you change it to 64-->64 00 (number have to be reversed), vddci for DPM 0 will go to +100mv from 900mv to 1v, because core voltage for DPM0 is 900mv.

For example, usVddcOffset for DPM7 is at A6A9 is cd ff -->ff cd (reversed)--> -51 in word mode, is -50mv. If you whant to change it for example to -250mv DEC value you have to go windows calculator with programmer mode and word mode (default is qword), put -251 (is always +1) and you will get FF05 HEX value-->05 FF (reversed), and if you have 1.350v for core at DPM7 you will get 1.1v for VDDCI.

Thanks again, so -251, would make 1100 vddci, have you tested so memory stays low?

If you can mod me up an actual 1400mv bios without altering 65288, I'd appreciate it a lot. I'm still learning.

I understand what your saying and I'm just trial and error learning the hex editing.

By the way my card won't burn up, I set +200mv, 20% power increase and the card was cool to the touch(box fan by case and gpu fan 100%) saw 50-60C but got a score of 980 on Unigine Heaven 1920x1080 extreme ultra 8x aa... Between 1188Mhz and 1219Mhz they need +20% with 175-200+mv which is in the 1400mv range totally stable for 1188, 1219 is iffy.. needs +200mv

Thanks

TongaEdit.zip 42k .zip file
Edited by chris89 - 9/1/16 at 5:33am
post #526 of 1079
Quote:
Originally Posted by chris89 View Post

Thanks again, so -251, would make 1100 vddci, have you tested so memory stays low?

I don't have tested more than 1.25V for core. But vddci will stay low.
Quote:
Originally Posted by chris89 View Post

If you can mod me up an actual 1400mv bios without altering 65288, I'd appreciate it a lot. I'm still learning.

I don't know now how to do it, I will try to know how to do it. But I think that 1.4v is to high.
post #527 of 1079
Quote:
Originally Posted by mynm View Post

I don't have tested more than 1.25V for core. But vddci will stay low.
I don't know now how to do it, I will try to know how to do it. But I think that 1.4v is to high.

Yeah you don't want to push a stock card because it just plain can't do it, and is barely able to pull off base clocks and volts.

As far as vddci at 1100 etc, that's what -250mv from actual? You sent me 1175mv vddci which resulted in 1250mv on the core and memory... Which memory stays no more than 1.194v under stock no matter how high the core voltage which I have had it upwards of 1475mv... The memory could use an undervolt to no more than 1150mv or 1100mv even to cool down the vregs so they don't throttle and can maintain below the 80C max incorportated into the thermal engineering guidelines of amd's powertune to prevent them from burning at over 100C. That 80C limit allows the card to last a long time, but we want to avoid the limit and reduce the temp so it doesn't quite hit 80C after 8 hours of load, like max 70C would do quite well.

Thanks again and yeah 1400 millivolts is too much for an untouched card for surely it is far too high, I know this because I briefly tested the card by stock and it ran horrible and smelled vreg burning a lot.

Then I found the heatsink surface contacting the core vregs was not only uneven but the gap was nearly 2mm from the regulator, meaning the pad had to be thick which creates way too much resistance and horrendous thermal performance.

Anyway after you invest into the best thermal material and remove the gaps between the vregs/core/ram so there is no gap at all between the chip and heatsink then the 1mm fujipoly compressed has almost no thermal resistance yielding peak performance for the gpu core, and keeping the memory modules cool. However I find Arctic Silver Ceramique 2 works best on the core diode and is better than any other paste. Although the memory modules cannot be utilized to peak potential because of the complete and utter lack of cooling at all on the memory module voltage regulators. So the memory is constantly throttling because the regulators heat up, slow down and cool down, heat up, then slow down and cool down it's a never ending memory issue. Not clearly apparent at all to anyone but myself who knows about such things.

Anyway yeah man throw me 1400mv bios, my card can handle it easily.
Edited by chris89 - 9/1/16 at 11:07am
post #528 of 1079
Quote:
Originally Posted by chris89 View Post

Yeah you don't want to push a stock card because it just plain can't do it, and is barely able to pull off base clocks and volts.

As far as vddci at 1100 etc, that's what -250mv from actual? You sent me 1175mv vddci which resulted in 1250mv on the core and memory... Which memory stays no more than 1.194v under stock no matter how high the core voltage which I have had it upwards of 1475mv... The memory could use an undervolt to no more than 1150mv or 1100mv even to cool down the vregs so they don't throttle and can maintain below the 80C max incorportated into the thermal engineering guidelines of amd's powertune to prevent them from burning at over 100C. That 80C limit allows the card to last a long time, but we want to avoid the limit and reduce the temp so it doesn't quite hit 80C after 8 hours of load, like max 70C would do quite well.

Thanks again and yeah 1400 millivolts is too much for an untouched card for surely it is far too high, I know this because I briefly tested the card by stock and it ran horrible and smelled vreg burning a lot.

Then I found the heatsink surface contacting the core vregs was not only uneven but the gap was nearly 2mm from the regulator, meaning the pad had to be thick which creates way too much resistance and horrendous thermal performance.

Anyway after you invest into the best thermal material and remove the gaps between the vregs/core/ram so there is no gap at all between the chip and heatsink then the 1mm fujipoly compressed has almost no thermal resistance yielding peak performance for the gpu core, and keeping the memory modules cool. However I find Arctic Silver Ceramique 2 works best on the core diode and is better than any other paste. Although the memory modules cannot be utilized to peak potential because of the complete and utter lack of cooling at all on the memory module voltage regulators. So the memory is constantly throttling because the regulators heat up, slow down and cool down, heat up, then slow down and cool down it's a never ending memory issue. Not clearly apparent at all to anyone but myself who knows about such things.

Anyway yeah man throw me 1400mv bios, my card can handle it easily.

Do you have tested the 1.100v for vddci bios?, you can test it without changing 65288, only changing voltage offset with Trixx or AB to see if it is stable.
post #529 of 1079
Quote:
Originally Posted by mynm View Post

I have not tested more than 1.25V for core. But vddci will stay low.
I don't know how to do it, but I will look into it though. I think that 1.4v is to high.
Have you tested the 1.100v for vddci bios?, you can test it without changing 65288, only changing voltage offset with Trixx or AB to see if it is stable.

Thanks I just now tested 1.100vddci and the ram ran at 1.100v and about 8 1/2 watts max, more stability than higher volts even at 1,563Mhz non-effective, effective is what times 3,4,5 for effective?

Anyway saw 1.225v on the core... I like it bud, great job.

So far on 1100vddci only 1094Mhz is possible. Since once you step up above 1094Mhz on the core, the 1100vddci isn't stable no matter how much volts you crank through the core. Which is 1400mv realistic max, it can do 1475mv but isn't happy at all. At 1400mv is pretty happy and stable.

So a little bit more than 1100vddci is necessary as 1100vddci max is 1094mhz core, 1125vddci would yield 1125Mhz stable, 1150vddci is enough for 1157Mhz, 1175vddci is enough for 1188Mhz, 1200vddci is enough for 1219Mhz, and 1225vddci is enough for 1250Mhz.

However for me to see higher clocks I need 1400mv or 1.400v because big clocks spike to as high as 1.393v when your up in the 1188 & 1219Mhz range.

The 1.100vddci bios did great for the ram but only enough core volts for 1094Mhz running cool & quiet.

**Maybe you can send a 1150vddci/1400 core and a 1225vddci/1400 core.

Thanks

R9_380X.zip 42k .zip file
Edited by chris89 - 9/1/16 at 3:12pm
post #530 of 1079
OK but 1.4V is to high. I will send you the bios by PM, so don't share it at the forum for safety.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › Software for R9 285 bios edit