1.25v+ on Vega64? - Overclock.net - An Overclocking Community

Forum Jump: 

1.25v+ on Vega64?

Reply
 
Thread Tools
post #1 of 20 (permalink) Old 02-13-2018, 02:03 PM - Thread Starter
New to Overclock.net
 
whiteskymage's Avatar
 
Join Date: Oct 2013
Posts: 104
Rep: 0
Talking 1.25v+ on Vega64?

Hey there,
I got a power table and I want to ask, has anybody managed to get anything higher than 1.25v? I have flashed my Vega64 with an LC bios, because I've put an EK waterblock on it. I manage to get 1690-1720MHz on the core while gaming. I just wanna see if going higher than 1.25v would allow it to clock higher. Like I think 1.275 would be fine to stop there, not gonna push for 1.3v cuz I am not sure if temps will allow it. Currently with the max overvolt, my GPU stays at about 59-63C so I do have a little headroom to try out the OC lottery.
whiteskymage is offline  
Sponsored Links
Advertisement
 
post #2 of 20 (permalink) Old 02-13-2018, 02:25 PM
New to Overclock.net
 
rancor's Avatar
 
Join Date: Oct 2007
Location: Iowa
Posts: 1,227
Rep: 102 (Unique: 93)
Buildzoid would be the one to talk to but from my understanding going higher than 1.25V requires hard mods. You can't raise the voltage by changing power play tables.

i7-4770K Overclock
Blue
(18 items)
CPU
Intel Core i7-4770K (4.4GHz 1.35V)
Motherboard
Gigabyte Z87X-UD4H-CF
GPU
XFX Radeon RX VEGA 64
RAM
16GB G.Skill Trident X 2400MHz 10-12-12-31
Hard Drive
Samsung EVO 840 500GB
Power Supply
Corsair HX1050W
Cooling
EK Supreme HF + mpc355 + 2 x Xspc RX360 + EK-FC Radeon Vega
Case
Corsair 700D with Side Window
Operating System
Windows 10 Pro
Monitor
Nixeus EDG 27
Monitor
ASUS MG279Q
Monitor
ASUS MG279Q
Keyboard
Xarmor U9BL-S
Mouse
Steelseries QcK+
Mouse
Roccat Pure Owl-Eye
Audio
Swan M50W
Audio
Odac+O2 & Sennheiser GSX 1000
Audio
Sennheiser 6XX
CPU
AMD Ryzen 2600X
Motherboard
GIGABYTE X470 AORUS GAMING 7 WIFI
GPU
Sapphire R9-290 Vapour-X
RAM
G.Skill Flare X
Hard Drive
Samsung 960 EVO
Power Supply
Seasonic Prime Ultra Titanium 650W
Cooling
Noctua NH-U12S SE-AM4
Case
Fractal Design Meshify C - Dark
Monitor
Pixio PX277
Audio
Swan M10
▲ hide details ▲


rancor is offline  
post #3 of 20 (permalink) Old 02-18-2018, 08:35 AM
New to Overclock.net
 
Spacebug's Avatar
 
Join Date: Dec 2013
Posts: 66
Rep: 2 (Unique: 2)
It is possible to get higher than 1,25v vcore if you mod the softpowerplay table.
However it is a bit of a pain as both core clock and core voltage has to be set in the powerplay table.
If you try adjusting those values in wattman after setting up the pplay table wattman tries to save your over 1,25v vcore, cant do that and reverts back to max 1,25v.

So every time you want to change voltage or core clock settings you have to do that through the powerplay table and afterwards just open up wattman and enable the settings from the pplay table, not changing any coreclock och core voltage values.
Spacebug is offline  
Sponsored Links
Advertisement
 
post #4 of 20 (permalink) Old 02-22-2018, 04:37 PM - Thread Starter
New to Overclock.net
 
whiteskymage's Avatar
 
Join Date: Oct 2013
Posts: 104
Rep: 0
Quote: Originally Posted by Spacebug View Post
It is possible to get higher than 1,25v vcore if you mod the softpowerplay table.
However it is a bit of a pain as both core clock and core voltage has to be set in the powerplay table.
If you try adjusting those values in wattman after setting up the pplay table wattman tries to save your over 1,25v vcore, cant do that and reverts back to max 1,25v.

So every time you want to change voltage or core clock settings you have to do that through the powerplay table and afterwards just open up wattman and enable the settings from the pplay table, not changing any coreclock och core voltage values.
That's fair enough. All I want is to push this card to the absolute maximum while gaming on Kingdom Come Deliverance just to get a little higher in settings while maintaining 40-50 FPS on 3840x1600 resolution. If I end up using 450W of power for the smallest increase from High to Very High in a couple of the graphical settings while maintaining 1700MHz and stable FPS, so be it, Power efficiency be damned, TDP be damned - I can cool this card on water with fans and pump going at 100% while I play with headphones and not hear them.

I am not really sure which of the values in the power play table changes voltage. Which of the values do I need to change to reach 1.275 or 1.3v?
whiteskymage is offline  
post #5 of 20 (permalink) Old 02-23-2018, 08:29 AM
New to Overclock.net
 
Spacebug's Avatar
 
Join Date: Dec 2013
Posts: 66
Rep: 2 (Unique: 2)
I'm using a handy editor for softpowerplay table created by a fellow OCN member Halsafar i think it was.
Links should be in the vega preliminary bios... Thread i think.

But be shure to check performance afterwards, i got negative performance scaling past 1.2V vcore on my vega64, probably due to the boost algorithm sensing too high current consumption and decided to decrease maintained coreclock.
Tried fixing that through software but soon gave up in favor of a hardmod to the cards current sense circuit.
Spacebug is offline  
post #6 of 20 (permalink) Old 02-23-2018, 04:25 PM - Thread Starter
New to Overclock.net
 
whiteskymage's Avatar
 
Join Date: Oct 2013
Posts: 104
Rep: 0
Quote: Originally Posted by Spacebug View Post
I'm using a handy editor for softpowerplay table created by a fellow OCN member Halsafar i think it was.
Links should be in the vega preliminary bios... Thread i think.

But be shure to check performance afterwards, i got negative performance scaling past 1.2V vcore on my vega64, probably due to the boost algorithm sensing too high current consumption and decided to decrease maintained coreclock.
Tried fixing that through software but soon gave up in favor of a hardmod to the cards current sense circuit.
Is there a way to completely un-restrain Vega64 from any power limits like Steve did with his Vega56 in this video:
by only editing the power play tables? I just wanna get rid of any power limits FIRST, before looking into cranking up the voltage. As I can see they can push Vega56 to a power consumption of 447W, which if that's possible, I can probably go higher with my Vega64. I have enough headroom to go far. Currently, my power play table looks like this:

"PP_PhmSoftPowerPlayTable"=hex:82,02,08,01,00,5C,0 0,22,07,00,00,03,2B,00,00,1B,\
00,48,00,00,00,80,A9,03,00,F0,49,02,00,96,00,08,00,00,00,00,00,00,00,00,00,\
00,00,00,00,00,02,01,5C,00,1B,02,12,02,94,00,6A,01 ,B4,00,FE,00,7A,00,8C,00,\
88,01,00,00,00,00,3E,02,00,00,90,00,74,02,39,01,0F ,01,63,01,00,71,02,00,71,\
02,02,02,00,00,00,00,00,00,08,00,00,00,00,00,00,00 ,05,00,07,00,03,00,05,00,\
00,00,00,00,00,00,01,08,20,03,84,03,B6,03,E8,03,1A ,04,4C,04,7E,04,B0,04,01,\
01,46,05,01,01,84,03,00,06,60,EA,00,00,00,40,19,01 ,00,01,DC,4A,01,00,02,00,\
77,01,00,03,90,91,01,00,04,6C,B0,01,00,05,00,08,D0 ,4C,01,00,00,00,80,00,00,\
1C,83,01,00,01,00,00,00,00,88,BC,01,00,02,00,00,00 ,00,B4,EF,01,00,03,00,00,\
00,00,90,0E,02,00,04,00,00,00,00,80,32,02,00,05,00 ,00,00,00,E0,54,02,00,06,\
00,00,00,00,00,71,02,00,07,00,00,00,00,00,03,60,EA ,00,00,00,40,19,01,00,00,\
80,38,01,00,00,00,08,28,6E,00,00,00,2C,C9,00,00,01 ,F8,0B,01,00,02,80,38,01,\
00,03,90,5F,01,00,04,F4,91,01,00,05,D0,B0,01,00,06 ,C0,D4,01,00,07,00,08,6C,\
39,00,00,00,24,5E,00,00,01,FC,85,00,00,02,AC,BC,00 ,00,03,34,D0,00,00,04,68,\
6E,01,00,05,08,97,01,00,06,EC,A3,01,00,07,00,01,68 ,3C,01,00,00,01,04,3C,41,\
00,00,00,00,00,50,C3,00,00,01,00,00,80,38,01,00,02 ,00,00,24,71,01,00,03,00,\
00,01,08,00,98,85,00,00,78,B4,00,00,60,EA,00,00,50 ,C3,00,00,01,80,BB,00,00,\
60,EA,00,00,94,0B,01,00,50,C3,00,00,02,78,FF,00,00 ,40,19,01,00,B4,27,01,00,\
50,C3,00,00,03,B4,27,01,00,DC,4A,01,00,DC,4A,01,00 ,50,C3,00,00,04,DC,4A,01,\
00,90,5F,01,00,90,5F,01,00,50,C3,00,00,05,00,77,01 ,00,90,91,01,00,00,77,01,\
00,50,C3,00,00,06,90,91,01,00,6C,B0,01,00,00,77,01 ,00,50,C3,00,00,07,6C,B0,\
01,00,6C,B0,01,00,90,91,01,00,50,C3,00,00,01,18,00 ,00,00,00,00,00,00,0B,E4,\
12,D0,07,D0,07,50,00,0A,00,54,03,90,01,90,01,90,01 ,90,01,90,01,90,01,90,01,\
00,00,00,00,00,02,04,31,07,DC,00,DC,00,DC,00,90,01,00,00,59,00,69,00,49,00,\
49,00,5F,00,73,00,73,00,64,00,40,00,90,92,97,60,96 ,00,90,55,00,00,00,00,00,\
00,00,00,00,00,00,00,00,00,00,00,00,02,02,D4,30,00 ,00,02,10,60,EA,00,00,02,\
10

The bold numbers are what I changed. It's what Buildzoid (rip spelling) had in his power play table when he was flashing his 2 Vega cards in Xfire. In my power offset, I can go to upto 150% (in Wattman). I do however find that while benchmarking, the GPU-only power, measured by MSI afterburner, can go up to 375W.
The question is - is there any other limit on this power play table that I am hold down on?
whiteskymage is offline  
post #7 of 20 (permalink) Old 02-24-2018, 08:55 AM
New to Overclock.net
 
Spacebug's Avatar
 
Join Date: Dec 2013
Posts: 66
Rep: 2 (Unique: 2)
I think, from my limited experience with these cards (my one vega64) that there are two problems we're facing when trying to un-restrain the vega cards.

One is powerbased throttling, which seems to be affected by the powerlimit or current limit in the softpowerplay table.
When I first messed with my card i got some throttling that was remedied by higher power and or current limit.

Perhaps a foolish approach but I assumed that the card wouldn't pull enough power anyways to blow itself up, so I set all the current and power limits up to 800 A and W respectively, as well as about 200% powerlimit for good measure.
As to sort of bypass the limiter by setting the limit high enough.
Probably excessive but it did the trick in prevent throttling.


The other problem is the boost algorithm that the cards use on the higher p-states, called AVFS or is it ACG?
Anyhow, it basically decides by itself what frequency the core will run at based on i guess a number of things, including current/power consumption.
What i noticed was that past 1.2V vcore the maintained clock in benchmarks got lower and lower as i increased vcore, hence lower performance and benchmark score.
I tried poking around in the softpowerplay table but nothing I tried regarding power or current had any impact on this behavior.

Tried to compensate by setting P7 clocks higher instead but then i would sometimes get scenarios where the gpu didn't pull quite as much power, and the boost algorithm thought it had power to spare so clocks increased too much resulting in crash.
But do try for yourself if it works on your card, i can only speak from my experience of my one card...

This whole boost thing where it basically decides by itself which clocks to run is really a royal pain when trying to push the card.
The one thing i found worked was to hardmod the current sense circuit so the card senses less power drawn then it actually uses.
That i think brought the sensed power consumption that the gpu thinks it is consuming down below a threshold i guess where the card didn't reduce clocks when more vcore was added.

The best thing would be if someone found a way to kill the boost function all together, so you can set a frequency and it will hold it no matter what.
But i'm not sure that will happen anytime soon :/

Last edited by Spacebug; 02-24-2018 at 09:00 AM.
Spacebug is offline  
post #8 of 20 (permalink) Old 02-24-2018, 10:10 AM - Thread Starter
New to Overclock.net
 
whiteskymage's Avatar
 
Join Date: Oct 2013
Posts: 104
Rep: 0
Quote: Originally Posted by Spacebug View Post
I think, from my limited experience with these cards (my one vega64) that there are two problems we're facing when trying to un-restrain the vega cards.

One is powerbased throttling, which seems to be affected by the powerlimit or current limit in the softpowerplay table.
When I first messed with my card i got some throttling that was remedied by higher power and or current limit.

Perhaps a foolish approach but I assumed that the card wouldn't pull enough power anyways to blow itself up, so I set all the current and power limits up to 800 A and W respectively, as well as about 200% powerlimit for good measure.
As to sort of bypass the limiter by setting the limit high enough.
Probably excessive but it did the trick in prevent throttling.


The other problem is the boost algorithm that the cards use on the higher p-states, called AVFS or is it ACG?
Anyhow, it basically decides by itself what frequency the core will run at based on i guess a number of things, including current/power consumption.
What i noticed was that past 1.2V vcore the maintained clock in benchmarks got lower and lower as i increased vcore, hence lower performance and benchmark score.
I tried poking around in the softpowerplay table but nothing I tried regarding power or current had any impact on this behavior.

Tried to compensate by setting P7 clocks higher instead but then i would sometimes get scenarios where the gpu didn't pull quite as much power, and the boost algorithm thought it had power to spare so clocks increased too much resulting in crash.
But do try for yourself if it works on your card, i can only speak from my experience of my one card...

This whole boost thing where it basically decides by itself which clocks to run is really a royal pain when trying to push the card.
The one thing i found worked was to hardmod the current sense circuit so the card senses less power drawn then it actually uses.
That i think brought the sensed power consumption that the gpu thinks it is consuming down below a threshold i guess where the card didn't reduce clocks when more vcore was added.

The best thing would be if someone found a way to kill the boost function all together, so you can set a frequency and it will hold it no matter what.
But i'm not sure that will happen anytime soon :/
So you've set:
DC 00 -> 220W Socket PowerLimit to 800W
DC 00 -> 220W Battery PowerLimit to 800W
DC 00 -> 220W Small PowerLimit to 800W
2C 01 -> 300A Tdc Limit to 800A
I guess?

Anyway, isn't there a way to do a shunt mod where I would put a little bit of liquid metal on some resistors to decrease their resistance so that gives me an additional power limit?
I saw this threat about the Titan X shunt mod (https://www.overclock.net/forum/69-nv...idia-gpus.html) , and I was thinking, well, won't that work on Vega 64 as well? Both cards have shunts right? Or are you using another hardmod which buildzoid suggested using a variable resistor?

@buildzoid join up, if you have the time

Last edited by whiteskymage; 02-24-2018 at 10:29 AM.
whiteskymage is offline  
post #9 of 20 (permalink) Old 02-26-2018, 03:42 AM
New to Overclock.net
 
Spacebug's Avatar
 
Join Date: Dec 2013
Posts: 66
Rep: 2 (Unique: 2)
Quote: Originally Posted by whiteskymage View Post
So you've set:
DC 00 -> 220W Socket PowerLimit to 800W
DC 00 -> 220W Battery PowerLimit to 800W
DC 00 -> 220W Small PowerLimit to 800W
2C 01 -> 300A Tdc Limit to 800A
I guess?

Anyway, isn't there a way to do a shunt mod where I would put a little bit of liquid metal on some resistors to decrease their resistance so that gives me an additional power limit?
I saw this threat about the Titan X shunt mod (https://www.overclock.net/forum/69-nv...idia-gpus.html) , and I was thinking, well, won't that work on Vega 64 as well? Both cards have shunts right? Or are you using another hardmod which buildzoid suggested using a variable resistor?

@buildzoid join up, if you have the time
Yeah, i guess, the power limit figures are correct, i stopped caring about the hex values/hex editing after i found the softpowerplay table editor.

Now however, when i have hardmodded the current sense circuit i think I have those values at or near stock values since the card don't sense the real power consumption those values are i guess non issue.
For now I have 1.34V Vcore and about 1770MHz maintained coreclock during gaming and benchmarks, gpu-z reports max 242W for gpu only powerdraw, when it probably is closer to around 600W...

Shunt mods with liquid metal won't work as the amd cards don't use shunt resistors to calculate power consumption.
Instead the voltage controller, from what I've understood, calculates current consumption based on voltage drop across the inductors of the phases, or some such...
Buildzoid would probably know more about that.

Anyway, buildzoid showed in one of his initial Vega FE videos on YouTube how he hardmodded the current sense circuit by soldering a resistor between a cap and resistor for each pair of vcore phases.
It was not fun mind you to solder on so small components for a guy with little in experience in soldering as myself...

The hardmod buildzoid showed with variable resistors for vega I think was for Vcore and Vhbm control, not for current/power measurement.
Spacebug is offline  
post #10 of 20 (permalink) Old 02-26-2018, 06:09 PM - Thread Starter
New to Overclock.net
 
whiteskymage's Avatar
 
Join Date: Oct 2013
Posts: 104
Rep: 0
Quote: Originally Posted by Spacebug View Post
Yeah, i guess, the power limit figures are correct, i stopped caring about the hex values/hex editing after i found the softpowerplay table editor.

Now however, when i have hardmodded the current sense circuit i think I have those values at or near stock values since the card don't sense the real power consumption those values are i guess non issue.
For now I have 1.34V Vcore and about 1770MHz maintained coreclock during gaming and benchmarks, gpu-z reports max 242W for gpu only powerdraw, when it probably is closer to around 600W...

Shunt mods with liquid metal won't work as the amd cards don't use shunt resistors to calculate power consumption.
Instead the voltage controller, from what I've understood, calculates current consumption based on voltage drop across the inductors of the phases, or some such...
Buildzoid would probably know more about that.

Anyway, buildzoid showed in one of his initial Vega FE videos on YouTube how he hardmodded the current sense circuit by soldering a resistor between a cap and resistor for each pair of vcore phases.
It was not fun mind you to solder on so small components for a guy with little in experience in soldering as myself...

The hardmod buildzoid showed with variable resistors for vega I think was for Vcore and Vhbm control, not for current/power measurement.
I manage to OC my GPU to like 1690MHz but it more maintains 1660-1680 most of the time. My power consumption for GPU-only shows 370-391W in GPUz. Seems like I will go into hardmodding at some point, but I just don't have the balls to do it now that GPUs are so expensive on the market. Anyway, what did you solder really if I may ask? Could you take a picture?
whiteskymage 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