Hawaii Bios Editing ( 290 / 290X / 295X2 / 390 / 390X ) - Page 14 - Overclock.net - An Overclocking Community

Forum Jump: 

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

Reply
 
Thread Tools
post #131 of 4256 (permalink) Old 07-18-2015, 09:05 PM
New to Overclock.net
 
Osbios's Avatar
 
Join Date: Jul 2015
Posts: 30
Rep: 2 (Unique: 2)
Holy ****. I need a better vram testing tool or samsung memory is really awesome.

At 950MHz my core will run stable with a -125mv undervolting offset.
So now I was testing where the vram limit would be with that voltage. After all I got rid of low dpm voltage problems of the ram via bios mod.

So... on my short time tests, to get a first hard limit, I only got a single error in occt at 1750MHz.
Osbios is offline  
Sponsored Links
Advertisement
 
post #132 of 4256 (permalink) Old 07-18-2015, 11:04 PM
Iconoclast
 
Blameless's Avatar
 
Join Date: Feb 2008
Posts: 30,080
Rep: 3134 (Unique: 1869)
Quote:
Originally Posted by Osbios View Post

After all I got rid of low dpm voltage problems of the ram via bios mod.

Did you make the DPM voltages the same, or did you force the card to only use one DPM level?
Quote:
Originally Posted by The Stilt View Post

Some AMD tools.

This vague response led me to the "ADL2_Adapter_EDC_ErrorRecords_Get" function listed in AMD Display Library SDK for atiadlxx.dll. However, since I know virtually nothing about programming, I can't really make use of this. I tried to use rundll on this entry point, but if it produced any output, I don't know how to view it.

I don't imagine the tools you have are public, or if you've written your own, you'd be willing to share them?

...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  
post #133 of 4256 (permalink) Old 07-19-2015, 12:20 AM
New to Overclock.net
 
Lard's Avatar
 
Join Date: Feb 2012
Location: Germany
Posts: 110
Rep: 39 (Unique: 29)
Something about to test and compare tight memory timings.
AIDA GPGPU Benchmark - Memory Copy:
It shows in a few seconds if a memory strap is too tight and starts to artifact.
It's not very accurate to compare memory timings, I have to run it few times and pick the best result.
It's inappropriate to test the stability of VRAM OC.

CompuBench:
Ocean Simulation and Particle Simulation profits a lot of tight memory timings, like +17% and +19,7% on my HD 7970.
So it's good to compare memory timings.
It is demanding enough to show instability of VRAM OC.

Stability of VRAM OC and memory timings is different.
But I don't have to care about the memory controller with my HD 7970 like you with the 290.

gupsterg showed that The Stilt's 1250 MHz strap starts to artifact at 1475MHz.
To squeeze the performance out of the straps, you can edit the MHz presets.
Instead of using The Stilt's 1250 MHz memory timings with 1250,1375 or 1500 MHz, you can make like a 1425 strap (1475 -50 MHz for safety purposes).

If gupsterg find the time and can confirm that The Stilt's 1375 MHz strap is stable at 1500 MHz and someone when it starts to artifact, you can do the same with it.
Provided that your VRAM OC is stable first.

The Stilt's modified 290X memory timings:
48 E8 01 = 1250 MHz
1C 19 02 = 1375 MHz
The two timings are equal, only the MHz presets are different.
Hynix memory straps: Warning: Spoiler! (Click to show)
Code:
48 E8 01 01 77 71 33 20 00 00 00 00 08 39 47 2A 50 55 0C 0B 24 20 45 04 00 46 C4 00 22 BB 1C 00 5C 0B 14 20 4A 89 00 A0 00 00 01 20 12 0C 21 1E 51 19 26 13 
1C 19 02 01 77 71 33 20 00 00 00 00 08 39 47 2A 50 55 0C 0B 24 20 45 04 00 46 C4 00 22 BB 1C 00 5C 0B 14 20 4A 89 00 A0 00 00 01 20 12 0C 21 1E 51 19 26 13
Elpida memory straps: Warning: Spoiler! (Click to show)
Code:
48 E8 01 02 77 71 33 20 00 00 00 00 4A BD 47 2A 60 55 0F 0F 23 1D 87 03 00 46 C4 00 22 AA 1C 08 5C 0B 14 20 4A 89 00 A0 00 00 01 20 11 0D 20 23 4A 1D 24 11 
1C 19 02 02 77 71 33 20 00 00 00 00 4A BD 47 2A 60 55 0F 0F 23 1D 87 03 00 46 C4 00 22 AA 1C 08 5C 0B 14 20 4A 89 00 A0 00 00 01 20 11 0D 20 23 4A 1D 24 11
https://www.overclock.net/t/1561904/mlu-bios-builds-for-290x

Example to change MHz presets:
I use The Stilt's 1250 MHz 280X memory timings with 1700 MHz on my HD 7970.
https://www.overclock.net/t/1554360/tahiti-memory-timings-patch-for-hynix-vram
Lard is offline  
Sponsored Links
Advertisement
 
post #134 of 4256 (permalink) Old 07-19-2015, 12:38 AM
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,270
Rep: 750 (Unique: 286)
Quote:
Originally Posted by Blameless View Post

Did you make the DPM voltages the same, or did you force the card to only use one DPM level?
This vague response led me to the "ADL2_Adapter_EDC_ErrorRecords_Get" function listed in AMD Display Library SDK for atiadlxx.dll. However, since I know virtually nothing about programming, I can't really make use of this. I tried to use rundll on this entry point, but if it produced any output, I don't know how to view it.

I don't imagine the tools you have are public, or if you've written your own, you'd be willing to share them?

The tools I use are neither public or my own, so I cannot help you on that.
The Stilt is offline  
post #135 of 4256 (permalink) Old 07-19-2015, 01:01 AM - Thread Starter
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 6,791
Rep: 730 (Unique: 342)
Quote:
Originally Posted by The Stilt View Post

Based on LeakageID it has average leakage characteristics.
However the voltage reported by my software simply doesn´t match the reported LeakageID.
I would expect around 1.225V default voltage with that leakage fuse.

Did the card come clocked to 1000MHz from the factory?

No it did not come as 1000 /1250 .

Same test but the rom it came out of box with 1030 / 1325 , this rom has no voltage offset but show +31mv in MSI AB. So like you said must have programmed IR3567B.
Warning: Spoiler! (Click to show)

Same test but the with Sapphire OC edition rom, this has clock of 1080 / 1410 , this rom has a voltage offset in it.
Warning: Spoiler! (Click to show)
Quote:
Originally Posted by The Stilt View Post

ps. Based on the dump the VRM efficiency on that card is just horrible.
It´s even higher than on reference cards and 166% higher than my efficiency optimizations set it.
Huge switching losses there.

doh.gif ....

And to think 8+2+2 is supposed be better , temps are lowest I've seen comparing with TriX 290 (ref PCB) & Asus DCUII 290X (Custom PCB).

It has mosfets same as ref , Link:- Translated review with VRM component info

Is it poor set up or the fact to run the phases from IR3567B doublers / interleaving used?
gupsterg is offline  
post #136 of 4256 (permalink) Old 07-19-2015, 01:35 AM
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,270
Rep: 750 (Unique: 286)
Didn´t know Sapphire card used doubled VRM configuration.
In that case the efficiency won´t be nearly as bad as the switching cycle get´s halved in the doubling process thumb.gif
The Stilt is offline  
post #137 of 4256 (permalink) Old 07-19-2015, 01:37 AM - Thread Starter
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 6,791
Rep: 730 (Unique: 342)
Phew ....

Looks like the card won't end up on ebay! LOL

Many thanks for taking the time to view & comment thumb.gif .

*** edit ***
Quote:
Originally Posted by Lard View Post

gupsterg showed that The Stilt's 1250 MHz strap starts to artifact at 1475MHz. ....

Hi Lard smile.gif ,
Code:
v1

1126-1250Mhz = The Stilt 1250MHz Timings
1251-1375MHz = The Stilt 1250MHz Timings
1376-1500MHz = The Stilt 1250MHz Timings

Above rom would do 1410MHz, stable IMO. Will try testing with the further test's you've highlighted Lard. I only used 3dmark , Heavensward plus Unigine Heaven & Valley. 1475MHz I got blocks at desktop within seconds of applying freq. Will try to test under 1450MHz , when get time. I also found when running say 3dmark graphics test 1 looped for 10mins+ it would show slight instability, upping GPU voltage by 6.25mv from what I'd normally use for the same GPU / RAM frequency fixed that.
Code:
v2

1126-1250Mhz = The Stilt 1250MHz Timings
1251-1375MHz = The Stilt 1250MHz Timings
1376-1500MHz = The Stilt 1375MHz Timings

With this setup I was able to get 1500MHz thumb.gif , usually I don't give +63mv to GPU but due to lack of time I just applied it make sure it would be stable OC with timings.
3dmark (Preset FS) result (Click to show)
Heavensward (maxed out 1080p) result (Click to show)
Here is a compare of GPU / RAM set to 1080 / 1410 of stock vs v1 vs v2 (Click to show)

I'm tempted to stick with v1 with +38mv 1080 / 1410 for 24/7 use, will do further testing to see max clocks with v1 then clock stock & v2 to same and do another compare.
gupsterg is offline  
post #138 of 4256 (permalink) Old 07-19-2015, 05:13 AM
New to Overclock.net
 
Lard's Avatar
 
Join Date: Feb 2012
Location: Germany
Posts: 110
Rep: 39 (Unique: 29)
Thank you gupsterg for the benches. thumb.gif
It's nice to see some comparisons with The Stilt's memory timings, even if we squeezed them a little bit out. biggrin.gif
Looking forward to your further testing! smile.gif
Lard is offline  
post #139 of 4256 (permalink) Old 07-19-2015, 06:06 AM
New to Overclock.net
 
Osbios's Avatar
 
Join Date: Jul 2015
Posts: 30
Rep: 2 (Unique: 2)
Quote:
Originally Posted by Blameless View Post

Did you make the DPM voltages the same, or did you force the card to only use one DPM level?

I changed the dependency on dpm level/voltage for the 3 memory clocks:


So I currently run at 950/1700/-125mv
Osbios is offline  
post #140 of 4256 (permalink) Old 07-19-2015, 01:13 PM
Iconoclast
 
Blameless's Avatar
 
Join Date: Feb 2008
Posts: 30,080
Rep: 3134 (Unique: 1869)
Interesting. I'll see if I can do the same thing. Main thing holding back my memory OC are black screen crashes when transitioning back to idle voltage.

Thanks.

...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  
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: 2 (1 members and 1 guests)
kithylin
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