Hawaii Bios Editing ( 290 / 290X / 295X2 / 390 / 390X ) - Page 11 - Overclock.net - An Overclocking Community

Forum Jump: 

Hawaii Bios Editing ( 290 / 290X / 295X2 / 390 / 390X )

Reply
 
Thread Tools
post #101 of 4278 (permalink) Old 07-17-2015, 07:57 PM
New to Overclock.net
 
Lard's Avatar
 
Join Date: Feb 2012
Location: Germany
Posts: 110
Rep: 39 (Unique: 29)
Quote:
Originally Posted by gupsterg View Post

No worries smile.gif .
The ROM Lard edited for me is for Vapor-X 290X OC edition ROM (Custom PCB), this latest ROM from Sapphire support has even higher TDC / TDP / MPDL than others I have.

Date in ROM is 06/02/15 , date format in ROM's is MM/DD/YY , so very new biggrin.gif , 015.045.000.015 .

VX290XOC_Tight_Timings.zip 99k .zip file

MD5 Checksum: 9DD107D60B58BA60ED74D6A198FE602C

The tighter timings are in straps

1126-1250Mhz
1251-1375MHz
1376-1500MHz

I have Hynix AFR ICs and like the Stilt stated in this thread tighter timings require lower RAM Freq.

With unmodified rom I have run 1475MHz for lengthy periods (even done 1550MHz for benches) without issue but with modified rom @1475MHz I start see blocky display @ desktop doh.gif , 1410MHz with tight timings gives same 3dmark score as when RAM @ 1500MHz. I only did quick tests last night, may do more soon.

As I have the unmodified and modified rom comparing them I can see what Lard added and as soon as get spare time can pickup how to mod this into roms.
In all three straps was The Stilt's 1250 MHz memory timings.
For the 1500 MHz strap I edit The Stilt's 1375 MHz memory timings.
I hope it runs now without artifacts at 1500 MHz.
VX290XOCmem.zip 99k .zip file
Attached Files
File Type: zip VX290XOCmem.zip (99.1 KB, 49 views)
Lard is offline  
Sponsored Links
Advertisement
 
post #102 of 4278 (permalink) Old 07-18-2015, 01:54 AM - Thread Starter
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 7,160
Rep: 774 (Unique: 354)
Quote:
Originally Posted by Blameless View Post

Nope, as pointed out to me by sugarhell, GCN achives DP by combining SP elements, and there doesn't even seem to be a way to physically remove DP hardware from the architecture without a serious redesign.

So a softmod would work to enable? not that I'd need to do it, just curious wink.gif and many thanks for posting info.
Quote:
Originally Posted by Osbios View Post

Well I was just editing with hex editor so far. I didn't see there are 3 memory clock values.

Post 1 heading GPU / RAM Clock Editing was created just for edits to highest GPU / RAM freq clock state. At the time we were not aware of the other sections in PowerPlay table, that section now needs updating due to the modding that people are doing to lower DPMs in the other areas of PowerPlay. This has also reminded me I must edit credits to include netkas for kicking off the editing on Guru3D.

We were first identifying hex and what it did, then got wrapped up in voltage mod. This was discovered by comparing the Stilt mining roms with stock, then there was also the VoltageObjectInfo table. This was where the GPU core voltage offset was found. We were still hesitant to test voltage mod until the Stilt turned up in the thread and of course after clarification from the man himself it was a green light smile.gif .

I know myself I have been spending all spare time on roms, so far many hours have gone into it.

I'm just an enthusiast, some of this has been a learning curve for myself blushsmiley.gif.

I don't have all the answers and am hoping through the thread we can work on progressing understanding, this is what occurred on Guru3D and all of the information in post1 is derived from work as a collective.
Quote:
Originally Posted by Osbios View Post

I still have some issues.
1:After a restart the minimum voltage needed for a given gpu/memory frequency seems to change.
2:Powerplay seems to be way more "active" because now I'm using way less Wattage. In some cases I lower the offset voltage and under load I actually read out more vddc... stuff like that.

Good to read it worked for you as well.

For your experience 1 are you getting a reading of 1.0xxx? if so some voltages set in rom don't work even if SVI 2 compliant so I'm thinking the negative offset is hitting those ones?

For your experience 2 I've found some benching tests hit higher VDDC than others, are you comparing like for like?
Quote:
Originally Posted by Lard View Post

In all three straps was The Stilt's 1250 MHz memory timings.
For the 1500 MHz strap I edit The Stilt's 1375 MHz memory timings.
I hope it runs now without artifacts at 1500 MHz.
VX290XOCmem.zip 99k .zip file

Many thanks again, I really appreciate you doing this thumb.gif . Still not had time to look at hex edits and looks like I won't as will be testing this new rom biggrin.gif .
gupsterg is offline  
post #103 of 4278 (permalink) Old 07-18-2015, 03:06 AM
Iconoclast
 
Blameless's Avatar
 
Join Date: Feb 2008
Posts: 30,086
Rep: 3135 (Unique: 1869)
So I took a look at the PT1 bios with Hawaii BIOS Reader and I'm not sure what was done to it to make only the maximum DPM state the only functional one.

The only changes I see are increased TDP/TDC/Power limits. All the different clocks and DPM states seem to be intact, so what's the deal?

I'm pretty new to editing these things and I'm sure I've overlooked something.
Quote:
Originally Posted by gupsterg View Post

So a softmod would work to enable?

No idea. Stilt's results seem to imply such, but there could still be some hardware mechanism preventing this that doesn't involve physically blocking the capability at the die level. Have to ask him; I've never tried to unlock FP64.

...rightful liberty is unobstructed action according to our will within limits drawn around us by the equal rights of others. I do not add 'within the limits of the law,' because law is often but the tyrant's will, and always so when it violates the right of an individual. -- Thomas Jefferson
Blameless is offline  
Sponsored Links
Advertisement
 
post #104 of 4278 (permalink) Old 07-18-2015, 04:55 AM
New to Overclock.net
 
Osbios's Avatar
 
Join Date: Jul 2015
Posts: 30
Rep: 2 (Unique: 2)
Quote:
Originally Posted by gupsterg View Post

Good to read it worked for you as well.

For your experience 1 are you getting a reading of 1.0xxx? if so some voltages set in rom don't work even if SVI 2 compliant so I'm thinking the negative offset is hitting those ones?

For your experience 2 I've found some benching tests hit higher VDDC than others, are you comparing like for like?

Maybe my first issue has more to do with the tools that sample the vddc from the gpu. I guess they just show the voltage at a single moment and not the average over the time from the last sampling.

I use mainly OCCT for testing, because it comes with error checking. Its easy to see when the memory does not get enough voltage for its clock, because then my display shows a LOT of artifacts. But if the GPU does net get enough voltage it will crash my system before I see anything except for OCCT errors.

But the different min. voltages needed for clocks after restart is real. Maybe there is some kind of dynamic initialization happening on start of the bios or the driver that tests for leakage? And this may varies a bit at every start and changes some offsets, too? I'm just guessing here.
Osbios is offline  
post #105 of 4278 (permalink) Old 07-18-2015, 05:21 AM
New to Overclock.net
 
sugarhell's Avatar
 
Join Date: Apr 2012
Location: 38.051289,23.709791
Posts: 9,327
Rep: 564 (Unique: 287)
Quote:
Originally Posted by Blameless View Post

Nope, as pointed out to me by sugarhell, GCN achives DP by combining SP elements, and there doesn't even seem to be a way to physically remove DP hardware from the architecture without a serious redesign.

Yeah this is correct. GCN shaders can do either SP or DP. You cant disable on the hardware the DP capabilities.

Amd just lock on the drivers the ratio of DP. Imagine a 290x with the full DP ratio. It would hurt firepro sales.

Quote:
Originally Posted by BiG StroOnZ
Warning: Spoiler! (Click to show)
No I said if Vega is clocked @ 1600MHz out of the box, I will eat my shoe on Twitch.tv.

Sound good? wink.gif
Warning: Spoiler! (Click to show)
Quote:
Originally Posted by Juicin View Post
Not in games, which is what 99.9999 percent of us care about

And unless you own a business, you get paid for your time. So your compute power just makes more work for you not less

Ryzen is a joke to the vast majority of the market who would consider buying their product.
sugarhell is offline  
post #106 of 4278 (permalink) Old 07-18-2015, 06:39 AM - Thread Starter
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 7,160
Rep: 774 (Unique: 354)
Quote:
Originally Posted by Blameless View Post

So I took a look at the PT1 bios with Hawaii BIOS Reader and I'm not sure what was done to it to make only the maximum DPM state the only functional one.

The only changes I see are increased TDP/TDC/Power limits. All the different clocks and DPM states seem to be intact, so what's the deal?

I'm pretty new to editing these things and I'm sure I've overlooked something.

I haven't looked at the PT roms yet smile.gif .

Doing a quick compare
Warning: Spoiler! (Click to show)
1. First produce tables for PT1 rom via atomdis.

2. Find a stock rom (I'd go with an Asus one) with same version number to minimise differences for comparing in HEX editor. When viewing rom in hex editor you will see a version near top end of it in ANSI text area.

3. Now compare command or data tables between roms which you wish to by selecting areas using offsets / length stated in atomdis tables for each rom.

Doing a long / deep compare
Warning: Spoiler! (Click to show)
First produce tables for PT1 rom via atomdis.

Find a stock rom (I'd go with an Asus one) with same version number to minimise differences for comparing in HEX editor. When viewing rom in hex editor you will see a version near top end of it in ANSI text area.

Produce tables for the stock rom, check command tables and data tables have as close a length as possible.

Note: If lengths differ it could be those tables have extra data for changes. If all lengths are same then you could be looking for hex value changes and of course it could be both.

Next order the command tables & data tables, you may also find empty areas between sections in them, I have noted them in some roms. These empty areas you may need to account for when doing multiple tables compare in one hit, especially if lengths differ.

Any differing tables between roms it's easier to cut those out and compare and then do bulk compare of rest of rom, again account for empty areas and their lengths.

Note: When you start cutting / comparing do it from end of rom upwards, as then you can use the offsets (+lengths) in the tables lists.

Here is a bios structure / order I created when comparing two different version number roms, use image for ref not the offset values in it as they will differ.
Warning: Spoiler! (Click to show)

Only offsets you can use from above image are end of padding area after command tables and UEFI data area start plus end of UEFI padding area.

As you can probably tell a Non UEFI rom ends at end of command tables padding area.

As a side note you can add a UEFI to non UEFI rom, Plug2k did it for his 295x2 using HD7XX UEFI Patch Tool Beta. I also tested his method on a Non UEFI rom and that way plus finding GOP in ANSI of that bios section identified UEFI data location in bios.
Quote:
Originally Posted by Osbios View Post

Maybe my first issue has more to do with the tools that sample the vddc from the gpu. I guess they just show the voltage at a single moment and not the average over the time from the last sampling.

GPU-Z , sensors page , click a value (not text label) and you can select instant , min , average , max.

I usually use MSI AB for monitoring and like the way with the graphs comparing multiple items you can note if clock / volts dropped for say a scene change, etc.

I also use GPU-Z as some of the data it gets you can't access in MSI AB and at times both.
Quote:
Originally Posted by Osbios View Post

... Maybe there is some kind of dynamic initialization happening on start of the bios or the driver that tests for leakage? And this may varies a bit at every start and changes some offsets, too? I'm just guessing here.

If I go lower volts than stock for DPM 0 - 6 I'll see what happens with mine.
gupsterg is offline  
post #107 of 4278 (permalink) Old 07-18-2015, 07:17 AM
New to Overclock.net
 
Osbios's Avatar
 
Join Date: Jul 2015
Posts: 30
Rep: 2 (Unique: 2)
Quote:
Originally Posted by gupsterg View Post

GPU-Z , sensors page , click a value (not text label) and you can select instant , min , average , max.

I usually use MSI AB for monitoring and like the way with the graphs comparing multiple items you can note if clock / volts dropped for say a scene change, etc.

I also use GPU-Z as some of the data it gets you can't access in MSI AB and at times both.

That is irrelevant. It just shows the min, ave, max value of all samples.
To get meaningful data you would need to sample the GPU as fast as it switches the voltage.
Quote:
Originally Posted by gupsterg View Post

If I go lower volts than stock for DPM 0 - 6 I'll see what happens with mine.

You can just start a gpu load like furmark or occt and then play around with voltage during GPU load. You just have to set the voltage up again before you stop the GPU load. That way you never go into idle mode where default bios setting would be to low for it to be stable with undervolting offset.
Osbios is offline  
post #108 of 4278 (permalink) Old 07-18-2015, 11:24 AM
New to Overclock.net
 
fyzzz's Avatar
 
Join Date: Nov 2014
Posts: 911
Rep: 60 (Unique: 26)
I love the modded the bios stilt gave me. I only have one gripe with it and that it is that voltage is so low. Changing the dpm7 voltage in hawaii bios reader changes nothing.
fyzzz is offline  
post #109 of 4278 (permalink) Old 07-18-2015, 12:06 PM
New to Overclock.net
 
Ized's Avatar
 
Join Date: Feb 2013
Posts: 420
Rep: 22 (Unique: 18)
Quote:
Originally Posted by fyzzz View Post

I love the modded the bios stilt gave me. I only have one gripe with it and that it is that voltage is so low. Changing the dpm7 voltage in hawaii bios reader changes nothing.

He somehow forced a really low voltage. He stated there was "extra protection" and it was limited to 1.36875V with but after droop (I guess) its no where close for me either.
Ized is offline  
post #110 of 4278 (permalink) Old 07-18-2015, 12:08 PM - Thread Starter
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 7,160
Rep: 774 (Unique: 354)
Quote:
Originally Posted by fyzzz View Post

I love the modded the bios stilt gave me. I only have one gripe with it and that it is that voltage is so low. Changing the dpm7 voltage in hawaii bios reader changes nothing.

What is The Stilt MLU rom name?
What voltage do you require?

Did you change DPM7 voltage in all all six sections in hawaiireader?

1. GPU freq table
2. RAM freq table
3. VCE limit table
4. UVD limit table
5. SAMU limit table
6. ACP limit table

Also note some of his MLU rom have a gpu core voltage offset programmed in rom, account for that voltage offset when modifying DPM7 voltage.
gupsterg is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off