Discussion in 'AMD / ATI' started by Dragonheart, May 1, 2013.
Yes I can, I did voltage to 1.300V with editor, now it works with 1.3
And the problem is; IF I set 1.300V; this will be my default voltage. It's possible to set lower voltage, but lowest one will be 1.238V.
I also tried Matrix bios, but couldn't open the bios file. If there will be lowest and highest voltage setting, would be great.
My latest settings for my 7970 Top (I'll use this one)
I did set to 1260 V with editor. I know my limit is 1180Mhz - 1190Mhz with games. So my lowest voltage is 1190 now (It's higher than default bios voltage) With 1.256 - 1265 V is ok for 1180Mhz.
Pretty cool though I have an error (see below) when reading any reference bios... tried running as admin too.
So, there was an accident that someone tested on his locked boost bios that got things working.
With that russian site, check method 1, that has the asic table to voltage.
If it doesn't have zeros in between yes they say it is unusable. That is the native bios for the v2 cards (locked voltage in boost). I edited the remaining info that I could, the ASIC values and wow, it worked!
A response to this modded bios:
Ok, I've thoroughly tested all tree of robnitro's BIOS-es, actually 2 of them, F43 BIOS mod doesn't work on rev2 cards, it won't bootup after flashing (4 people tested it on at least 7 cards as I know of). But that doesn't matter as the other two, especially FY1 works great. FX1 works as promised but it is the FY1 that put my jaw to the floor.
Basically with FY1 mod you get unlocked voltage on rev2 boards. You can crank up voltage all the way from 1.056 to 1.25, it is all just based on your card's ASIC quality and choosen speed. GPU-z this time shows correct values and with some tweaking you can get all the sweet spots you couldn't get with F43 or FX-FY1. The trick is though to use Trixx as there was no way to force AB to even read card's voltage. One of the highlights of this mod is state 2 where you set GPU to 900MHz, memmory to whatever pleases you and check this out - voltage to 1.000 volts. I didn't want to go any lower because of high risk for the GPU itself, but even though my card has low ASIC it managed to work in that mode for more than a day before I shut it down. Temperatures with this settings oscillates between 58-61 with fans set to 33%. smile.gif
BIOSes automatic fan curve stays horrible though (F43 doeas a much better job here), but it is not that important to me anyways.
One more great feature is that HDMI finally works! smile.gif
This is one superb mod and trust me, this is quite a replacement for the f43 on rev2 boards, as with this one you have complete control over your card.
robnitro, you should send FY1 mod with complete description to Gigabyte, maybe they will hire you for doing such a fine job. thumb.gif
So, with the boost cards that are non-reference you can change voltage to be lower, by choosing a different asic score to limit at. I end up setting the rest to be 99 03 88 03 etc.. so no ASIC ever uses those low ones, and the ones I don't want , I set a really LOW asic, like 01 01! The two sets correspond to boost (02 ff) and 2nd state (01 ff)
3d Clocks are easy to find. For 7950, search for 48 e8 01 02 ff for boost and 48 e8 01 01 ff for reg 3d (in boost they use this as state 2- explained below). It's reverse hex... 01 e8 48= 125000 ram. The 3 bytes before that (in my case 90 5f 01 - 900mhz for 01 ff, and a0 86 01- 1000mhz for 02 ff boost) are the clocks.
For powertune, the boost bios has some advantages. When the TDP is too high, instead of just clocking down, once the clock goes below the 2nd state (signified by the 01 ff clock) it uses that voltage. So, in my bios I found that 1.09v is OCCT stable till 1020. I set state 2 to be 998 mhz at 1.09v. I have boost at 1.18v, stock 1000 in bios but 1100 in CCC. Once pt drops the clock below 998, it uses much less power (at 1.09v- 16 a VRM in, instead of the non boost running similar fps in occt at 1.18v), keeping the card cooler!
example: same fps with boost and non boost, 640x480 occt 4.4.0- shaders 5 (best for errors) error check on... 600 fps:
1.18v- way high, too much I trigger OCP- about 25 amps VRM in (12v)
1.09v- doable, 17 amps VRM in (12v)
I've done the 50% tdp mod and it makes my vrm's too hot, causing them to throttle back to 500mhz UVD state. I did mod that state to be 750 mhz as it was 0.95v and stable until 800 or so. It makes the impact of throttling less. Uses same power at low load, like playing youtube... voltage is directly related to the square of it for power. (Power=voltage(squared)/resistance)- resistance is same at same 3d work done. More work, less resistance, higher power use.
BTW, my asic is 76%
If you need any info, like a hex comparison with my test locked boost bioses, just email me at robnitro ( AT YAHOO DOT COM)
I'm not good at programming, but great in haxoring hex, lol. I've also modded my car ECU
I get an error when clicking confirm, despite choosing 7900 and it saying reference, no matter what bios I check
See the end of this message for details on invoking
just-in-time (JIT) debugging instead of this dialog box.
************** Exception Text **************
System.IndexOutOfRangeException: Index was outside the bounds of the array.
at Radeon_HD_7000_BIOS_Editor.MainForm.btnConfirmAsic_Click(Object sender, EventArgs e)
at System.Windows.Forms.Control.OnClick(EventArgs e)
at System.Windows.Forms.Button.OnClick(EventArgs e)
at System.Windows.Forms.Button.OnMouseUp(MouseEventArgs mevent)
at System.Windows.Forms.Control.WmMouseUp(Message& m, MouseButtons button, Int32 clicks)
at System.Windows.Forms.Control.WndProc(Message& m)
at System.Windows.Forms.ButtonBase.WndProc(Message& m)
at System.Windows.Forms.Button.WndProc(Message& m)
at System.Windows.Forms.Control.ControlNativeWindow.OnMessage(Message& m)
at System.Windows.Forms.Control.ControlNativeWindow.WndProc(Message& m)
at System.Windows.Forms.NativeWindow.Callback(IntPtr hWnd, Int32 msg, IntPtr wparam, IntPtr lparam)
************** Loaded Assemblies **************
Assembly Version: 220.127.116.11
Win32 Version: 4.0.30319.18047 built by: FX45RTMGDR
Radeon HD 7000 BIOS Editor
Assembly Version: 18.104.22.168
Win32 Version: 22.214.171.124
Assembly Version: 126.96.36.199
Win32 Version: 4.0.30319.18047 built by: FX45RTMGDR
Assembly Version: 188.8.131.52
Win32 Version: 4.0.30319.18021 built by: FX45RTMGDR
Assembly Version: 184.108.40.206
Win32 Version: 4.0.30319.18044 built by: FX45RTMGDR
************** JIT Debugging **************
To enable just-in-time (JIT) debugging, the .config file for this
application or computer (machine.config) must have the
jitDebugging value set in the system.windows.forms section.
The application must also be compiled with debugging
<system.windows.forms jitDebugging="true" />
When JIT debugging is enabled, any unhandled exception
will be sent to the JIT debugger registered on the computer
rather than be handled by this dialog box.
Sorry for the progress moving slow, classic problem - I have no spare time
Anyway, I've made some fixes and another good thing happened, so i've decided to share it with you.
First - v0.42 is attached to the first post.
1) Hopefully the problem with asic's textbox on machines with different locales is solved. Tell me if it's not
2) Bug with saving path when application and bios are in the same directory (or even drive) is fixed also with adding standart File-Save As dialog.
3) I had many reports that utility picks up wrong string from voltage table depending on users asics. I've doublechecked everything and the picking algorithm works exactly as it's described in the original article. Probably it needs some fixes so as a temporal solution i've added an optional feature of simply patching every string of voltage table with the signal/VID u've selected. It has no downsides, imho, but for current voltage to being picked correctly I still have to find another solution.
And some good news. test2013 - author of all these infos i've used about what and how to edit published another manualmanual about frequencies editing (somehow I can't translate it with google translate, try at your own if needed).
I've tried changing the GPU and MEM frequencies limits and stock 3D frequencies on my PowerColor 7850 and it works!! No BSODs or other stuff, no need to reinstall driver - clocks and new limits are there right after the reboot.
It will be pretty easy to implement, so I want to implement it before going back to all these voltage stuffs. The only problem - I'll need user to enter his current clocks to find the right places in BIOS which we will patch then. Forcing user to enter so much info is not nice, but..
Can't promise any date, but I'll try to release an update ASAP. Anyway, this method is as simple as PL editing, so you can try it manually by yourself.
P.S. Sad RBE is not going to rebirth.. so much things it could have for HD7xxx...
It lives on in spirit right here. Awesome work so far.
No patch buton in 0.42 ?
we'll use "Save As" !??
I tried in every way but does not change the default voltage is always on my card gigabyte 7870 1.219vm controlled with Open Hardware Monitor.
I do not know if I'm wrong or is it a bug of bioseditor..wtf:
but you may not be able to unlock the bios of 7870 with chip ... Pitcaim at a voltage greater than 1219mV .. I'll buy 'never a card with bios and voltages blocked I feel just the Dissatisfied of this board ... next time I'll find better first purchase.:-(((((((((((
Ok now it works without crash.
One annoying thing though:
Open file, it only allows choice of .rom.
Some bios files, like gigabyte use the extension to designate version (ex: r759wgd3.f43).
I put in * and hit enter to see all files, no biggie..
But with save as, same issue, rom only, and it adds that extension, which I change after back to .f43 because of short DOS 8.3 names in atiflash.
cant change the voltage, wont let me type in the box
edit: nm, you cant change the voltage, you increase the signal.
monitor programs return what ADL says for target voltage, not necessarily whats in there.
check the Vdcc levels while loaded to check for voltage increases.
for example, my stock Vdcc generally runs around 1.01 for my stock target VID of 1.170.
if i increase my target to 2.50, and my Vdcc exceeds my stock target of 1.170, then i know its patched.
This latest build has worked for me, as the previous ones did not. Thank you very much! I was able to take the unlocked reference ASUS 7970 bios from TPU and raise the 3D voltage up to 1.263 shown in AfterBurner. The flash was a success for my reference Diamond 7970. Card is running stable at 1250 / 1750. Selecting new voltage of 1.254 in this utility gave me 1.263 show in AfterBurner after the flash.
Okay just tried this editor with my Asus DirectCU II Top and it didn't make any changes, max volts are still 1.170v even though I changed it in the editor to 1.266v
Edit Update; right seems like my bios isn't flashing!!!!! Using atiwinflash and whenever I try to flash I keep getting the message that the rom file isn't found????? (even though the little window pops up and looks like it's flashing but quits halfway through with the message "xxx.rom not found) even though the bios file is in the same folder as the atiwinflash program????
Any help would be appreciated......is there some way Asus would have locked this card so the bios can't be flashed????
flashed heaps of cards bios before but never had this issue....card is version 1.01 on PCB
and hit enter to list the files, looking for your rom file.
If you don't see the file there or the name cut off, you have to type it that way. example:
thislongname.rom would be thislo~1.rom in dos... stupid 8.3 filename limit!
Otherwise, try this to see if atiflash finds the file itself and shows info using your rom file name.
atiflash -biosfileinfo filename.rom
Also, one time I had a usb flash drive go intermittent bad... some files were "zombied". They were there, worked fine in windows, but in dos I had issues.
You can try copy filename.rom test.rom
If that fails, that may be the issue.
Ummmm I'm talking about atiwinflash wich can be run through the command prompt in windows, not atiflash which is run in dos.
Update....looks like it did actually flash when I tried atiflash under dos. Power Control Settings have been raised to 50% in CCC and Asus GPU Tweak is showing Max Voltage for GPU as 1.276v (set 1.266v in bios editor).........but no other software is showing the voltage increase.
GPU-Z is till showing max volts under load as 1.170v as is MSI Kombuster. Only ASUS GPU Tweak is showing max volt range to 1.276 so I wonder if it is actually being applied under load or not???????????
always verify this stuff by measuring with a multimeter
How about non reference card?
Confirmed with multimeter voltage at load is actual 1.29v
Atiwinflash= picky garbage. It even causes me issues on stock roms lol.
It may be something that I noticed:
This voltage editor doesn't update the 16 bit checksum.
Example: I have a stock bios with checksum 37632 (16 bit in hex workshop).
If I change voltages using this program and hit save as, and then do a compare in hex workshop, only the voltage registers change, there is no checksum correction.
So now, the checksum becomes 37469.
Atiwinflash doesn't like this. But atiflash dos doesn't seem to care.
Normally, when I manually edit bioses, after I am done, I read the 16 bit checksum, then scroll down to where there are 00's then FF's and adjust this to get the same checksum. If your checksum is too low and there is no space in the 00's, then you can change many FF's to 00's which also works to change the checksum as 16 bit checksum can only go to zero and then resets back at the max value. I have no idea where the checksum value is stored, as it is different for different bioses but adjusting the FF/00's works for me to correct it.
Will you be able to implement adjusting/raising the overclock limits shown in CCC for core and mem, don't necessarily need to have raised clocks hard fixed in bios (current bios clocks are fine) but would be nice to get raised CCC limits like 1400MHz core and 1800MHz mem.
not unless he finds a way to get around the driver hash checking...
Have faith......his Russian ......those brain-e-acks always find a way!
however W1zzard spent a lot of time trying to find a way around it on the 6xxx series with RBE.
If he couldnt find a way, i hold little hope someone else will.
im not really concerned about clocks to be honest like i was on the 6xxx, as my stock max is 1500/1950. i couldnt hit those numbers with nitro and all the volts in the world.
Separate names with a comma.