Overclock.net › Forums › AMD › AMD CPUs › AMD FX-8350 vcore without LLC question (M5A97 R2.0 mobo)
New Posts  All Forums:Forum Nav:

AMD FX-8350 vcore without LLC question (M5A97 R2.0 mobo)

post #1 of 7
Thread Starter 
System is:
AMD FX-8350 - stock voltage 1.3375
ASUS M5A97 R2.0 (latest BIOS)
Corsair Vengeance 2x4 (8GB) 1600mhz @ 1.5
Seidon 240M closed loop water cooler
Optimal case ventilation.


The M5A97 R2.0 clearly has some issues with LLC. It is not variable, and can only be enabled or disabled. In the bulldozer/piledriver overclocking guide, they also mention that LLC on this board is incredibly dodgy.

I tested this out a bit by turning LLC off and jacking my voltage up to something like 3.6GHz @ 1.47, which at load came down to something like 1.38 and my system is stable. I then set my voltage to a fixed 1.38 (also tried 1.4, 1.41, etc) and turn LLC on and everything is screwed; I assume voltage isn't compensated fast enough when it goes from 0 to 100 load immediately on stress test or that the surge in voltage is way too high and overcompensating. I tested this a bit more, but have certainly come to the conclusion that LLC is to blame here.

My issue is that I'm having trouble pushing my OC higher with fluctuating voltages. Knowing that under load I am going to drop significant voltage (something like 0.1v), is it safe to run at 1.5v or higher - this is knowing that at 1.52v (fluctuates between 1.48 and 1.52) at idle/low load I have SIGNIFICANT room in my thermal margin, we're talking like 55-60c. Then at load my temps barely rise because my voltage is dropping. Is it temperature exclusively I have to worry about here, or is there an actual danger zone with voltage that despite low temps will cause immediate damage if I have some sort of voltage spike or sustained high (1.55) voltage at idle?

And on that thought, what about using one of the cpu power saving functions to throttle clock and voltage at idle so I'm staying away from 1.55v at idle, and then when under load it lets off the throttling and my voltage naturally decreases from vdroop under load? Is that actually possible, or am I really stretching it here?


I am going to get a M5A99X EVO R2.0, but for the time being and the sake of experimentation I want to push this a little harder. Any thoughts?
Edited by thehamburglar - 11/8/15 at 2:11pm
post #2 of 7
You will probably kill the board. I did mine and I barely touched the FX8320 it was running. The board has no LLC controls. I think the "on" setting uses something equivalent to "extreme" on the Sabertooth, which is nuts. No, if you set it to on chances are you will be overvolting more than necessary and running super hot even when on idle. You need to turn it off, but then as you have already noticed, you will be subjected to massive vdrops. To mitigate this, you will need to set your vcore to an unreasonably high value. Unfortunately this will also cause your chip to run a bit warm under idle and you may end up blowing up the phases on the board. If I where you I would just overclock as much as possible on stock volts until you get your new board.
Magicbox
(17 items)
 
crapbox
(13 items)
 
 
CPUMotherboardGraphicsRAM
FX 8320 Sabertooth 990FX Nitro+ RX480 Kingston HyperX Fury 
Hard DriveHard DriveHard DriveCooling
Samsung 850 EVO  Kingston HyperX 3K Seagate Barracuda 7200.14 Noctua NH-D15 
OSOSMonitorKeyboard
Kubuntu  Windows 10 Pro Dell U2515H CM Quickfire TK (Cherry Blue) 
PowerCaseMouseMouse Pad
Cooler Master Silent Pro M 850W Enthoo Pro Logitech G502 Logitech G440 
Audio
Xonar DX 
CPUMotherboardGraphicsRAM
Sempron 3300+ HP stock mobo (laptop) 200M (IGP) 2x1GB PC3200 
Hard DriveOptical DriveOSMonitor
100GB ATA133 DVD/CDRW Kubuntu 32 bit 14.1" (1280x768) 
Power
6 cell 
  hide details  
Reply
Magicbox
(17 items)
 
crapbox
(13 items)
 
 
CPUMotherboardGraphicsRAM
FX 8320 Sabertooth 990FX Nitro+ RX480 Kingston HyperX Fury 
Hard DriveHard DriveHard DriveCooling
Samsung 850 EVO  Kingston HyperX 3K Seagate Barracuda 7200.14 Noctua NH-D15 
OSOSMonitorKeyboard
Kubuntu  Windows 10 Pro Dell U2515H CM Quickfire TK (Cherry Blue) 
PowerCaseMouseMouse Pad
Cooler Master Silent Pro M 850W Enthoo Pro Logitech G502 Logitech G440 
Audio
Xonar DX 
CPUMotherboardGraphicsRAM
Sempron 3300+ HP stock mobo (laptop) 200M (IGP) 2x1GB PC3200 
Hard DriveOptical DriveOSMonitor
100GB ATA133 DVD/CDRW Kubuntu 32 bit 14.1" (1280x768) 
Power
6 cell 
  hide details  
Reply
post #3 of 7
Thread Starter 
Yeah, I figured that it was probably using the extreme equivalent of other boards with the proper LLC power management tools. The thing is that I am not running warm at idle, at all. Far from it. I'm looking at my thermal margin on AMD Overdrive right now on my second screen while I write this. Low load / close to idle conditions, and I have 55-60c of thermal margin remaining. If I'm concerned about blowing the VRM on my board, is it possible to try to reduce, say, CPU/NB & CPU VDDA or similar voltages in order to reduce strain on my VRMs, letting me push my core voltage higher?
Edited by thehamburglar - 11/8/15 at 5:19pm
post #4 of 7
CPU VRM is the limit and only VDD_CR (vcore) is using it´s output.
Lowering the other voltage won´t make a difference.

The board is fine for FX-8K series at stock, but insufficient when overclocked.
Get a proper board if you wish to OC.
post #5 of 7
Thread Starter 
So NB and HT have no bearing on the load on VRMs? I have a pretty rudimentary understanding of power management on mobos, but am quite curious now. I am a low voltage technician by trade, and deal with a lot of sensitive DC voltages (3v-12v) in the field. It's in my nature to try and break this board smile.gif Is there a standard on how power from the different phase types are used as well as power delivery from VRM? If there is a good general purpose article on the subject, send it my way, if you don't mind. I had a quick look on Google, but I couldn't find anything that actually broke down what pins are used for what and why; just the general ideas behind it.

I already ordered a new board, so I think I'm done screwing with this one. Just curiosity driving me at this point.
post #6 of 7
There are several voltage groups and planes.

The two main voltages are VDD_CR & VDD_NB (Group A).
Both of these voltages are controller by the same controller, however the actual regulator circuit is separate due different voltage levels.

The rest of the voltages (VDDA, VDDIO, VDD HT, etc) are regulated by their own switching controllers or linear regulators.

So lowering VDD_NB, VDDA or any other voltage but VDD_CR won´t reduce the strain put on the CPU VRM.
post #7 of 7
Thread Starter 
I see! Thanks:)
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
Overclock.net › Forums › AMD › AMD CPUs › AMD FX-8350 vcore without LLC question (M5A97 R2.0 mobo)