• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

ASUS LLC on 8-series AM3 issues: Making the best of worst

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Dolk

I once overclocked an Intel
Joined
Mar 3, 2008
Currently a bunch of the ASUS 8-Series boards have a fun issue with LLC where it creates a huge VDroop.

As of now these motherboards are listed as to have the problem:

M4A89GTD-Pro (USB3 included)
M4A89TD-Pro
CH IV

Whats going on here? Here is the plain and simple fact of whats happening. When LLC is enabled, the correct voltage is applied to the CPUv but when the CPU is in need of more voltage, it is usually limited by the amount given. LLC allows the CPU to use more voltage, creating a much more stable environment. When LLC is turned off, a VDroop is applied to the CPUv. This drop in voltage can be anywere from .05v to .1v.

The VDroop is something that you have to determine yourself.

How to see this in action:
If you have a CH IV board, you can watch this interesting BIOS bug work in real time. Attach a Voltmeter to the CPUv and the GND pin-outs on the board. This will give you a reading of the actual voltage being applied to the CPU.

What to do in the mean time:
Currently ASUS is trying their hardest to create a BIOS that gets rid of this bug. They have been working around the clock to make sure their customers are satisfied. As for now, the consumer has to bear with this little big. I say little because its really more of a headache rather than a major instability causing issue.

There are a couple work arounds that I believe would be appropriet for this situation. First you can easily leave the LLC on and figure out a voltage that will work with the OC. You should look for the bear min voltage in this case. For LLC will take care of the rest and will apply more voltage when needed. I would not suggest to keep the voltage at default 1.35, because LLC doesn't completely make up for everything. Its only designed to help a little. For example a 4.0ghz CPU requires about 1.41v to work 24/7. If you turn LLC on, you can get away with 1.38v in the BIOS.

The draw back to keeping LLC on is that you will not be able to fully control the temperature of your CPU anymore. This can cause instability with the higher OCs that some people are trying to achieve. If you are looking for a high OC, but you are already pushing that temperature to its breaking point, you can turn off LLC. Now the issue you are going to have is determining the exact VDroop you have. My suggestion is to start off with what you would normally use. For example a 4.0ghz CPU will use about 1.41v. With my board (VDroop may very please do not take my example 100%) I have a .05v VDroop. That means that I have to add .05v to the CPUv to achieve an actual 1.41v. So in my case a CPUv setting reads 1.46v, but the BIOS reads at 1.41v.

I hope this helps this helps.


As a side note I will be testing further to figure out how to make the best of this situation. I could use some help though since I don't have to much free time :D
 
Thanks Dolk, I suspected this was a bug and not a feature. It just wasn't "on" with anything I've read.

My CPU-NB also seems slightly off on voltage, are these tied together? (Obviously knowing this I will test extensively and post any and all finding and data) Just curious if anyone knows.
 
Thanks DOLK! i will be trying this out tonight and see if i can hit a higher frequency and keep it cool or just cool off my setup. thanks again!
 
welp... bad news for me so i cant get a stable OC with LLC disabled at all when doing prime under 1.5 v in bios @ 4.0 ghz it will crash instantly for me with the Voltage at 1.5 it will run prime but 1 core will error out interesting enought i noticed that i drop from1.476 at idle down to a low of 1.404 when under load so it was a big drop and to achieve a bootup i had to enable LLC on the CPU-NB or i would crash before i could login. I would like to hear more people trying this to see if its diffrent based off of BIOS ver and just maybe board to board or CPU im running the 0905 bios on my IV.
 
I'm running 905 on my CHIV, what is your CPU-NB speed at? I set it to 2800 and 2.875v to get it stable at that speed with LLC off. I would also raise your cpu voltage a little over 1.5, try 1.51 and if that doesn't work try 1.52, that should make you more stable. I have about a .1v drop on mine at full load. I can boot up in to windows with no problem even at high voltage and no load. Sounds like you have an unstable cpu-nb and not quite enough voltage on the cpu for a 4.0ghz clock. My cpu loves 1.416-1.424V to be stable at 4.0ghz under load. Let me know how it goes.
 
Ill have to try it. im not real comfortable with setting my Voltage to over 1.5 but i guess i can try it. so i suppose though if i have this right because of the vdroop im not really running it as high as the bios says so in theory im safe correct me if im wrong though.
 
I was messing around with the voltages and LLC for a bit when under cold this weekend.

I was very limited to my high end OCing because of the flux in the voltage. When trying to disable LLC, the Voltage would come out correctly, but would drop .1v after load :/
 
I'm now prime stable 12 hours at 4.0 ghz with 2640 nbcpu and htlink. My voltage idles at 1.38 and bumps up to 1.44v with LLC enabled. Have not had any problems so far.
 
I have a 1055T and a Crosshair IV with 0905 BIOS. I experience the same vdroop with LLC disabled, but only on the CPU...if I disable LLC on CPU/NB it just stays @ the 1.51-53v "auto" gives it. I also get the same up to .05v spike on vcore when stressing or just using the system with LLC enabled.

Currently I am running 4GHz(320x12.5)/1.45v("auto")-3.2GHz NB....with all voltages set to "auto", and LLC set to "auto" also.
 
bah!? 1.51-1.53v on your CPU-NB???

Please tell me that was a mistype or I'm going to have to ask you to probe that board.
 
Crosshair IV Formula beta is out! From XS:

bingo13 said:
09/08/2010 Update

I will warn everyone now that updating to the following BIOS code means you cannot flash back to a previous BIOS and we will not have a tool to do it. Unless you are willing to use a BETA BIOS, wait for the official release please.

Crosshair IV Formula - BIOS 055
1. Improved Memory Performance and Clocking
2. C1E/CnQ operation update
3. Voltage enhancements

http://www.mediafire.com/file/hb5uxmg59nm7bed/CrosshairIV-Formula-ASUS-0055.rar

Crosshair III Formula - BIOS 1805
1. Improved Memory Performance and Clocking
2. C1E/CnQ operation update
3. Voltage enhancements

http://www.mediafire.com/file/kdvvx3dapwimu1i/Crosshair-III-Formula-ASUS-1805.rar

Based on the CIIIF results, should be rolling out M3/M4 updates in the near future.
 
IMO this is just another example of Asus rushing stuff out the door before it is properly qualified.
 
This BIOS has been a loooooooong time coming (look at the rest of the thread at XS I linked). They didn't rush anything for this BIOS. They're releasing a strongly vetted beta basically on demand from the community.
 
In my experience and that of many other PC builders Asus has a history of rushing half-baked products to market -- as any search will confirm.
 
Heh, I'm beginning to agree with you.

CrimInalA said:
flashed CHIV 055 bios , previous oc settings are not stable at all . not even when upping voltages and lowering speeds .
(yes i did clear cmos , reset to defaults and did settings again)

So now running linx at F5 default bios settings to see if that's stable .

After all that time, you'd think they would do better than that. :bang head

Glad I haven't been home to flash it yet! I think I'll deal with the Vcore adjustments necessary and stick with 1005.
 
Yes this is exactly the problems ive faced.
Wont be buying Asus again for the frustration factor.
Not going higher than 4ghz coz of the heat issue.
 
Last edited:
ASUS isn't the only manufacturer that's had power issues over the past year. I don't know what the problem is but they've all got some to one extent or another. ;)
 
Back