A temporary fix for the excess PCI-E slot power draw for the reference RX 480 cards - Page 12 - Overclock.net - An Overclocking Community

Forum Jump: 

A temporary fix for the excess PCI-E slot power draw for the reference RX 480 cards

Reply
 
Thread Tools
post #111 of 178 (permalink) Old 07-06-2016, 10:30 AM - Thread Starter
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,266
Rep: 749 (Unique: 286)
Quote:
Originally Posted by comagnum View Post

I can test this evening after work if you'd like @The Stilt

Gorod is on it already, but thanks.
If the software works as expected, I'll release it immediately.
The Stilt is offline  
Sponsored Links
Advertisement
 
post #112 of 178 (permalink) Old 07-06-2016, 10:35 AM
... not new ...
 
Join Date: Sep 2011
Posts: 702
Rep: 104 (Unique: 69)
Quote:
Originally Posted by OneB1t View Post


also what piss me really is that they fixed PCI-E power draw issue in 3 days but fixing 2D power usage with multimonitors or in light load like playing youtube takes them exactly 3 years now...


What's worse is that the fix for that is insanely simple: underclock the RAM. The GPU clocks are already very tame in those scenarios, the high power draw is 100% due to the RAM ramping to full clocks.

If AMD would include a few clock tiers for the RAM, the problem would 100% vanish. With full RAM clocks, a 1080p Youtube video uses 186W on my system. With 625Mhz RAM clocks, it uses just 150W. Further, the RAM is usually stable as low as 500 to 550Mhz.
looncraz is offline  
post #113 of 178 (permalink) Old 07-06-2016, 11:18 AM
New to Overclock.net
 
Join Date: Jun 2011
Posts: 710
Rep: 49 (Unique: 30)
thats just simply ******** i know few people who not buying RX 480 because of that multimonitor issue
its not really that much in terms of money but this is about not wasting when there is no need to...

its simply amazing that AMD will fix PCI-E power draw issue in few days which is not really that hot as it looks (if your MB cant handle 110W then you prolly dont want to buy RX 480 anyway as your whole PC is piece of crap) but they overlook their 2D power consumption for few years (also worth to mention that NVIDIA had same issue and they fixed it..)

OneB1t is offline  
Sponsored Links
Advertisement
 
post #114 of 178 (permalink) Old 07-06-2016, 12:44 PM
New to Overclock.net
 
Gorod's Avatar
 
Join Date: May 2016
Posts: 3
Rep: 1 (Unique: 1)
The Stilt program works perfectly ! thumb.gif Many many thanks ! cheers.gif

Card no longer dropping clocks in 3DMark GT1 as it did before the use of The Stilt program , frequencies will remain locked to set values !
The memory over the 2250 limits also working ( wattman now shows extended range for memory clocks) , so far the card was able to pass 3DMark with 2300 . The result is with card on stock cooling and weak old CPU , just to test it quickly , tessellation was disabled in driver (for easy of compare with previous results) and power limit set to +50% in wattman (with 0% it behaves like with +50% before - slight drops in clocks in some tests) , with +50% however stays locked to set values 100% of the time when under load thumb.gif

Gorod is offline  
post #115 of 178 (permalink) Old 07-06-2016, 12:51 PM
 
Join Date: Dec 2012
Location: On a high horse
Posts: 5,181
That's a pretty good score.

Is Polaris memory bandwidth bottlenecked?
TheLAWNOOB is offline  
post #116 of 178 (permalink) Old 07-06-2016, 12:56 PM
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 6,736
Rep: 730 (Unique: 342)
@gorod

Does RX 480 like 390 series / Fury & X have "Power Efficiency" setting in "Gaming" > "Global Settings" of Crimson driver?

I have noted your past results on RX 480 with i7 6700K in HWBot, my Fury X results:-

My 24/7 OC vs RX 480 1460/2250

My HWBot sub of 1175/545 vs RX 480 1460/2250

Tess.tweak done in my results as well.

Looking at your Fury X 1155/575 sub, mine is a bit off on GS score elements redface.gif (compare).
gupsterg is offline  
post #117 of 178 (permalink) Old 07-06-2016, 01:00 PM
New to Overclock.net
 
NomNomNom's Avatar
 
Join Date: Apr 2010
Location: Canada
Posts: 1,557
Rep: 65 (Unique: 59)
Quote:
Originally Posted by Gorod View Post

The Stilt program works perfectly ! thumb.gif Many many thanks ! cheers.gif

Card no longer dropping clocks in 3DMark GT1 as it did before the use of The Stilt program , frequencies will remain locked to set values !
The memory over the 2250 limits also working ( wattman now shows extended range for memory clocks) , so far the card was able to pass 3DMark with 2300 . The result is with card on stock cooling and weak old CPU , just to test it quickly , tessellation was disabled in driver (for easy of compare with previous results) and power limit set to +50% in wattman (with 0% it behaves like with +50% before - slight drops in clocks in some tests) , with +50% however stays locked to set values 100% of the time when under load thumb.gif



The I2C commands he posted in OP also unlocked the memory clocks for you?

NomNomNom is offline  
post #118 of 178 (permalink) Old 07-06-2016, 01:21 PM
New to Overclock.net
 
Gorod's Avatar
 
Join Date: May 2016
Posts: 3
Rep: 1 (Unique: 1)
TheLAWNOOB the card does not seem too bottlenecked by memory but of course scores improving slightly with memory overclock as well smile.gif My card wont go past 2300 with stock cooling , will have to see tomorrow if better cooling helps (maybe memory controller does not like those high temps ??? )

gupsterg yeah RX 480 has "power efficiency" setting in driver , turned it off of course smile.gif Did you up the voltage of your Fury X card ? For some reason they drop efficiency/performance when voltage is increased , but you probably already know that hehe biggrin.gif

NomNomNom that i did not try yet redface.gif

In any case now at least the owners of RX 480 will have a peace of mind that card is working full speed all the time no matter what bench or game , and can play with memory overclocks as well even if they might not go much further over 2250 without voltage mods . So big thanks to The Stilt for that ! thumbsupsmiley.pngsmile.gif
Gorod is offline  
post #119 of 178 (permalink) Old 07-06-2016, 01:21 PM
New to Overclock.net
 
paromi's Avatar
 
Join Date: Jul 2016
Posts: 13
Rep: 3 (Unique: 3)
welli´m reading and reading and still don´t get it..... LOL rolleyes.gif

What are you guys doing and what software did you use

i understand that you got rid of the Thermal Throttle and got higher memory Speeds

Thanks in advance

Keep up the good work thumb.gif
paromi is offline  
post #120 of 178 (permalink) Old 07-06-2016, 01:29 PM - Thread Starter
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,266
Rep: 749 (Unique: 286)
Ok, so here is the tool Gorod just tested above.

It is simply called EMBD (Ellesmere's Backdoor) and it increases the power limits from the stock figures. It also raises the maximum allowed MEMCLK limit from 2250MHz to 2500MHz. It re-configures the display driver and makes no changes to the hardware itself. The functionality of the program is purely "fire & forget" and it is executed only once (once to apply and once to remove, if necessary). Once you have deployed the tool you can delete it if you feel the need to do so.

It changes the configuration so that the new -50% PowerTune setting (in WattMan) is the same exact thing, as the stock configuration (at ±0% PowerTune) was.

After this tool has been deployed, the new power limit figures are:

PowerTune -50% = 110W, ±0% = 165W TDP, +50% = 247.5W TDP. The current limitations also reflect the same changes, so it is likely that you'll actually hit the current limitations of the VRM controller itself before you hit the driver limits.

The voltages or the frequencies have not been changed from the default settings.

https://1drv.ms/u/s!Ag6oE4SOsCmDhDIoWNpvaMKq8OtE

NOTE: Both of the changes (Apply & Clear) require a reboot. Also the program does not support multiple adapters, regardless if they are in standalone or in CrossFire mode.

EMBD V1.2 https://1drv.ms/u/s!Ag6oE4SOsCmDhERPkAmTcANOYbnk
The Stilt 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