Overclock.net › Forums › Graphics Cards › AMD/ATI › Polaris Bios Editing ( RX5xx / RX4xx )
New Posts  All Forums:Forum Nav:

Polaris Bios Editing ( RX5xx / RX4xx ) - Page 165

post #1641 of 3297
1) i saw your score on custom run Firestrike, i've asked you to provide me with validated result. I'm suprised by that score, probably the drivers 16.xx do the job or some custom settings. Stock voltage on these cards is around 1.10-1.15V. On your voltage i'm hovering around 1450 mhz. Anyway, please provide some link to validated score. The best would be if you can update the 3dmark which has been recently updated on their page. The best would be if you can update your drivers to the latest ones from AMD - 17.3.3. Such result would be comparable.

2) ok, i can tell you that i'm operating this card on much higher frequencies with higher voltages and my core clock do not fluctuate until VRM reach 115. I do not have any radiator on components which you have highlighted on the picture. Most of them i can even touch by bare hand. Core is rock stable even when VRM temp hits 110 area.

MEM VRM - construction of memory VRM is a bit more complicated, if you would like to cooldown these VRMs you should attach the radiator on the another side of PCB as there is a step voltage regulation, low-side mem vrms are lockated against number 5 on the picture, on another side of the pcb, close to the sames place you will find High-side VRMs. If something may need the radiator that will be High side. I was touching these VRMs on GPU side of the card (no 5) they are cold.

Here you will find answer why low temperature on VRM isn't so important and why mem VRM on gpu side is cold.

https://www.youtube.com/watch?v=qG2e-v94L4M

3) i can understand ppl who use mentioned card in crossfire, that they would like to keep ref cooler on. In my opinion downvolting is a better solution. According to my experience 1266 mhz need around 1.05 V, on stock 1.1 V i can hit 1340 mhz, on 1.2V - 1420 mhz, 1.236V 1450 mhz and finally 1.312V - 1480 mhz

PS

Powerlimit on stock BIOS. If you will move Powerlimit slider to the max 50% you should be able to overvolt +100mv and overclock to 1400 mhz and you should still have enough power. Above mentioned values you need to mod powerlimits in Bios. Of course everything depend on your initial VID.

PS2

there is no need to use any "NASA" thermal conductive materials, watercooled 480 stay below 45 degrees on max load haveliy overclocked. VRM is a different story, heatsink has to be large.
Edited by Na1l - 3/27/17 at 9:28pm
Little facker
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5-6600K @ 4.6 Asrock Fatal1ty Z170 Gaming-ITX/ac Sapphire Ref RX480 8GB @ 1480/2100 DDR4 HyperX Fury 2133 MHz cl14 @ 3100 MHz cl13 
Hard DriveHard DriveCoolingOS
Samsung SM951 Seagate ST2000DM001 Custom loop Win 10 Edu 
MonitorKeyboardPowerCase
Iiyama X2483HSU Ozone Strike Battle Corsair HX520EU Corsair Obsidian 250D 
Mouse
Logitech MX518 
  hide details  
Reply
Little facker
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5-6600K @ 4.6 Asrock Fatal1ty Z170 Gaming-ITX/ac Sapphire Ref RX480 8GB @ 1480/2100 DDR4 HyperX Fury 2133 MHz cl14 @ 3100 MHz cl13 
Hard DriveHard DriveCoolingOS
Samsung SM951 Seagate ST2000DM001 Custom loop Win 10 Edu 
MonitorKeyboardPowerCase
Iiyama X2483HSU Ozone Strike Battle Corsair HX520EU Corsair Obsidian 250D 
Mouse
Logitech MX518 
  hide details  
Reply
post #1642 of 3297
Right on haha yeah that's interesting. I just like to ensure contact on everything. Not that the memory VRM on GPU side needed anything, reporting static 26C.

As far as I saw the 65288 value is on a scale. There is no defined voltage for any given clock, so to speak. The 65288 is range from say 1.05v to 1.231v. Which I ran 1407mhz, without any increase to voltage. Which reported 1.231v in HWInfo max.

I guess it all comes down to how to see more FPS? When your VRM cooks over 90C, so much energy is being lost. Lost energy not able to enter the Core to power more FPS. It's lost by thermal expansion and isn't ideal. By actively cooling all these, we minimize the Loss to the Core. Therefore transmit more power to the core for more FPS.

It's all about isolating the "power" and "properly" directing said "Power" or "Energy" into the core diode. To then push out more FPS, safely and consistently as well. I could run the test over and over again and the score would be very near. Up in the 15,780+ range, give or take 1fps.

Yeah as far as my results, validating and all that I'm not really much to compare. Other than to throw up here and see if it does well...

Later on when I get a PCIe 3.0 Ryzen system I'll probably validate. Since that score was limited by PCIe 2.0 anyway, on a PCIe 3.0 GPU.
post #1643 of 3297
With my 7950 I learned the best use of powertune. And yes VRM temps matter, but up to a point... Your idea of running fans higher is good to reduce power use, like OCCT 450 fps 640x480 shaders 7 1275 1.15v uses 130 w but if I let the chip hotter like 75-80 it will use 140w. Some of it is from GPU chip itself using more power when hot, but anything electrical you have IR losses due to resistance changes especially at the 100+ amps we run.

So now onto how powertune helps, to run a limit.
Let's say what occt uses, crysis engine, it can load up the gpu a lot. Example, a scene in a game that is hard might spike the power use to max out the VRM at the top clock/volts.
To run that scene, you can either throttle due to temp spike/current spike protection
OR
Use a lower state, 100 mhz less lets say, which is 90% of max, but it also uses 90% of the voltage-> 80% less power in watts

Another simple example... let's say your card can handle 180w max on it's vrm without overcurrent protection.
Top step you have 1400 at 1.24 v lets say
Next step below 1270 at 1.14v.

If you hit a scene that loads gpu up to max out the 180 watts at the 1400 clock. Let's say it needed 200 watts ideally.
It's going to throttle some way, by vrm it's more of a hard cut, a stuttery on/off like older cpu's did idle states.

So if instead you had set the limit to 180w or 170w, it would step down to 1270 mhz lower voltage which can do 90% of the calculations at 80% of the power... So instead of needing 200watt max it needs 160 watt max.... and now you have 20 watts room to push more calculations and not hit the VRM/chip current limits.

Sad that these 14nm cpu and gpu still using 1.1xx-1.2xx volts. And some 14nm cpu to overclock past 4.5ghz need like 1,4 volts, same like my old 32nm sandy bridge 2500k lol. We're hitting a limit of size it seems... and thats why these mofos are so power hungry still. The future will probably be where they optimize it to do more with less clock cycles, hopefully vega does it well!
post #1644 of 3297
Quote:
I guess it all comes down to how to see more FPS? When your VRM cooks over 90C, so much energy is being lost. Lost energy not able to enter the Core to power more FPS. It's lost by thermal expansion and isn't ideal. By actively cooling all these, we minimize the Loss to the Core. Therefore transmit more power to the core for more FPS.

This statement is partialy wrong. You are right, there is a loss, but you are wrong telling that if you minimize power loss you will gain some FPS. You can improve power efficiency but it will do not impact the FPS. There is such a huge headroom on VRMs that it can't limit the card within the whole temperature operating range (25 - 125 C degrees).

Just like i said, i would like to compare the Firestrike results because your score is very high comparing to the other results which i found in 3dmark db on described clocks. Score which you have achived i saw only on cards which had ~1500 on core. It is even more misterious if you just told that you are using PCI-E 2.0, 2-3% less performance comparing to PCI-E 3.0.
Quote:
As far as I saw the 65288 value is on a scale. There is no defined voltage for any given clock, so to speak. The 65288 is range from say 1.05v to 1.231v. Which I ran 1407mhz, without any increase to voltage. Which reported 1.231v in HWInfo max.

This is the way how AMD decide to control voltage, it doesn't mean that you do not increse the voltage. Typical voltage on stock card is a 1.15 under load. as you can see your voltage is 0.08 higher. There is no difference if you manually set this voltage in wattman or mod the bios to rise automatically to this level. Anyway - Firestrike smile.gif I would like to compare it, once you will have the results let me know. Thanks,
Little facker
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5-6600K @ 4.6 Asrock Fatal1ty Z170 Gaming-ITX/ac Sapphire Ref RX480 8GB @ 1480/2100 DDR4 HyperX Fury 2133 MHz cl14 @ 3100 MHz cl13 
Hard DriveHard DriveCoolingOS
Samsung SM951 Seagate ST2000DM001 Custom loop Win 10 Edu 
MonitorKeyboardPowerCase
Iiyama X2483HSU Ozone Strike Battle Corsair HX520EU Corsair Obsidian 250D 
Mouse
Logitech MX518 
  hide details  
Reply
Little facker
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5-6600K @ 4.6 Asrock Fatal1ty Z170 Gaming-ITX/ac Sapphire Ref RX480 8GB @ 1480/2100 DDR4 HyperX Fury 2133 MHz cl14 @ 3100 MHz cl13 
Hard DriveHard DriveCoolingOS
Samsung SM951 Seagate ST2000DM001 Custom loop Win 10 Edu 
MonitorKeyboardPowerCase
Iiyama X2483HSU Ozone Strike Battle Corsair HX520EU Corsair Obsidian 250D 
Mouse
Logitech MX518 
  hide details  
Reply
post #1645 of 3297
OK figured I'd start with the 1250 and work my way down if necessary 1250 ran real good for about 5 minutes until clock dropped but the vrm only hit 73 so i'm trying to understand what in the bios told it to downclock at only 73 on the vrm? I figure has to be the temps you set of 84,88,84.

Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
post #1646 of 3297
Nice so is 1250mhz feel fine? Temperature deviation? If it's stable your good. Yeah that clock drop was from 84-88-84C. Another component on the PCB reached 84C and dialed back.

*Side Note*

My system is a Dual Xeon X5650 24 thread rig, with 48 gigabytes of ram. On PCIe 2.0 off all things. AMD Settings set performance and tessellation down. I'm sure that affects a couple fps.

Yeah I hear ya and I can understand why it's not a well known thing. That as a capacitor or a resistor or regulator rise in the temperature, affects the voltage/ wattage/ current requirement to pull the same power that it did at ambient temperature. It does this in ways of resistance. This is a well known principle, that if you heat a wire while measuring resistance, the resistance of said wire will increase in resistance until it melts and breaks the connection.

That's why proper cooling is ideal because as temperature rises, the power consumption from the wall becomes greater. For instance if everything on the PCB used exactly 50C of heat, it would use exactly half as much power than if everything on the PCB used 100C of heat. The resistance would double, requiring more power.

Here's a plain graph showing the voltage resistance as temperature rises, requiring to use more power to do the same thing on a cooler device (overall). This is classic to the Hawaii/ Grenada AMD generation GPU. Though it did and still does perform awesome. It's hotter, so it will use more power overtime until the temperature levels off.


Edited by chris89 - 3/28/17 at 1:20am
post #1647 of 3297
I'd like to show you the results of stock .153 bios These drivers produce the best results in fps and the highest in heat even with fans at 100% the core reaches 89 but the VRM OMG 95 surprisingly it takes awhile to reach 89 about 10 minutes vsync is off cause i only get around 50 fps in shadow of mordor 4k ultra no AA.

Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
post #1648 of 3297
Too bad it runs so hot and not enough fan speed to cool it back down. Do you want faster than 1250mhz on cooler than stock bios? What's the fps comparison between 1250mhz 2067mhz vs 1303mhz 2000mhz?

It's only about 3 billion pixel's and 15 billion texel's faster... could be as much as 5fps... do u need 5fps for 35C hotter?
post #1649 of 3297
i'll go run the shadow of mordor bench with the stock.153 1303 2000 then throw the 1250 2067 at it and report back.
Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
Double Dragon
(16 items)
 
  
CPUMotherboardGraphicsGraphics
AMD FX-8370 970 PRO GAMING/AURA Radeon (TM) RX 480 Graphics Radeon (TM) RX 480 Graphics 
RAMRAMRAMRAM
Crucial Technology  Crucial Technology  Crucial Technology  Crucial Technology  
CoolingOSMonitorKeyboard
Corsair H110i Windows 10 X64 HP 27es Corsair K55 RGB 
PowerCaseMouseMouse Pad
Rosewill Hive 1000 Watt Bronze Certified Phateks P400 Tempered Glass Red/Black Corsair Harpoon RGB Corsair MM800 RGB 
  hide details  
Reply
post #1650 of 3297
Here try this bios too...

By the way to reduce top video card... it looks like you already have a gap, I guess it's normal the board is setting gpu 2 to x8 or something.

I would mod the exhaust plates and remove the case bar between slots to cut 10C at least of gpu 1, and many more off gpu 2. Giving more overhead.

If your careful just remove the plates completely. So no damage. Then remove bar between slots of case for full gpu exhaust width. So air free flows out the back no restriction.

On reference blower it's easier to match all cards temperatures. If you add a circular funnel shaped pad around the circle of the fan inlet on each gpu. Like adhesive vibration padding on a roll. That way intake case airflow is scooped directly into the fan inlet with the extra 0.25-0.5 inches of gap between GPU's.

Jeztur2003_1306mhz_2167Mhz_85C_88C_85C.zip 110k .zip file
Edited by chris89 - 3/28/17 at 1:39am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › Polaris Bios Editing ( RX5xx / RX4xx )