AM3+ Voltage Questions

robertpartridge

New member
I've been working on the OC for my FX-6300 now that I finally replaced the stock cooler with a Cryorig H7. I've got it right now at 4.4GHz / 1.44375V and it's running fantastic. I had it running at 4.5GHz / same voltage and it worked great for about a day (even during some gaming) but then while watching Youtube the system froze up on me. I tried bumping the voltage to 1.45V and that didn't help any.

If I want to try for 4.5GHz again, should I try voltage at 1.45625V on the chip? Should I try bumping the chipset voltage? What are safe voltages to consider?
 
So this one yes?

W0kacLf.jpg
 
OK not good news. I would not push more than 1.4v on that board.

If you look to the left of the socket (or above in the pic) you will see a row of 1R2s. Next to them (on the left or above in thepic) there are sets of 4 little black things. They are your voltage regulators and they have no cooling at all. I *think* the board is 4 phase though it may be 6 but that is all irrelevant because there is no cooling. On later boards they looked like this.

o6n4EC9.jpg


Note the huge cooler and 8 pin power connector.

I would not push that CPU more than you can get on 1.4v. It's likely the voltage is being throttled any way, which is probably why your CPU froze. I don't recommend it, but if you ran Prime 95 on that board and watched the voltage in CPUZ you would probably see it fluctuating. Don't do that, it's a recipe for disaster but yeah, I think you have found your limits.

What exactly are you using to overclock it? the FSB or the multi? I would say just OC the FSB but again you are going to be heat/voltage limited.

Having said that you've done OK on that CPU.
 
OK not good news. I would not push more than 1.4v on that board.

If you look to the left of the socket (or above in the pic) you will see a row of 1R2s. Next to them (on the left or above in thepic) there are sets of 4 little black things. They are your voltage regulators and they have no cooling at all. I *think* the board is 4 phase though it may be 6 but that is all irrelevant because there is no cooling. On later boards they looked like this.

o6n4EC9.jpg


Note the huge cooler and 8 pin power connector.

I would not push that CPU more than you can get on 1.4v. It's likely the voltage is being throttled any way, which is probably why your CPU froze. I don't recommend it, but if you ran Prime 95 on that board and watched the voltage in CPUZ you would probably see it fluctuating. Don't do that, it's a recipe for disaster but yeah, I think you have found your limits.

What exactly are you using to overclock it? the FSB or the multi? I would say just OC the FSB but again you are going to be heat/voltage limited.

Having said that you've done OK on that CPU.


I may boot to windows and have a look at the voltage to see how it's looking - if it's fluctuating much under load. I'm more of a Linux guy, but I checked and my normal monitoring software doesn't have the voltage and if I do it from the command line I get a load of sensors and no good description if I'm looking at the right ones or not.

I ran all my stress testing using the "stress" command - which you can use to max the cpu at 100% for a set period of time (which I find is a really nice feature).

To answer your question though, the OC is through the multiplier.
 
OK well I found when overclocking both of the 8320s I had (both on Asus boards but I had the one with the cooling) that using the FSB yielded higher overclocks with less voltage. I managed 4.6ghz out of the lesser board (the other was a Crosshair V FZ).

I don't know how good you are with tools etc but you could always make a cooler for it or get one from a dead board and mod it on.

With FX chips it really is down to how much current the board can provide and how stable it is. And it's not going to be very stable with no cooling at all.

The situation is eased somewhat because you have a hex core, not a 8 core but yeah, it's still going to be a huge factor.
 
My only past experience with OCing a FSB in the past was over a decade ago on my Athlon XP-M 2600+ that I pushed to 212FSB (limited by the RAM). How does this work with these chips. I didn't bother with FSB because I didn't want to screw with RAM.

Also - I just found the Vcore sensor on Linux so I'm gonna monitor that to see what's up. I'm already seeing it fluctuate between 1.44 and 1.43V just in normal use.

I'm all for cooling it but I've got to figure out how I want to go about that. I was looking at some small heatsinks that could be applied with thermal tape. I dunno.
 
OK cool man. I forget what temps the VRMs throttle at but I have 50c and 70c floating around in my head.

My first forray into AM3+ was an Asrock 990FX Extreme (myass) 3. It had 4+1 VRM and I could only get 4.2ghz stable with a fan blowing over the pathetic little VRM sink it had on it.

In theory you should not do any damage. It's not like the Gigabytes where you can basically disable the VRM throttle and pee in the wind, the Asus will do whatever it takes to protect itself.
 
Just re-watched Tom's watercooling roundup video and wondering ... do you think the Cryorig A80 would possibly address my issues by pointing the "block fan" toward the VRMs?
 
Back
Top