power phase vs chipset?? which is more important for overclocking

[koven]

currently have a 785G mobo w/ 8 power phases

looking at the biostar ta890fxe as an "upgrade" but noticed it only has 6 power phases

what affect will this have on overclocking? should i look for another 890fx board but with 8 power phases? im trying to take my 955 under water and get ~4.2ghz

 

[Evilsizer]

the more phases the more current can be supplied to the cpu. this becomes wanted when ocing quad core and higher cpus. i have no hands on experience with amd ocing. just following the intel rule of thumb i have come up with. a dual core will need at least 2 phases for mild ocing anything higher look at 4 phases. a quad core will need a min of 4 phases for mild ocing anything higher look at 8phases. this rule of thumb mainly applies to air/water ocing. if your into the extreme side of things then go with more. there is also a point that when reached more phases wont help in any situation.



Are you unable to reach 4.2ghz on the current board? have you checked out XS forums or here or anywhere, to see what oc's some are hitting with your board? as well as with the one your looking at getting? what if hitting a higher clock is a matter of finding the right bios for your setup. it is possible, maybe more so for the intel side of things. i dont know but think it is worth looking into.

 

[beenthere]

Increasing the number of phases in a VRM (power supply module), smoothes out power ripple and noise which is important for clean CPU power. The current may or may not increase with more phases depending on the VRM design. More phases can be better in reducing ripple/noise but don't get hung up on phases as the CPU makers specify what the minimum CPU VRM must be for stable, reliable performance - and they OC their CPUs everyday.

A quality 4 or 8 phase VRM on an AMD mobo even with significant OC'ing should be just fine. A 4+1 and 8+2 VRM design means that the mobo uses a split power plane which is important for independent adjustment of the CPU/IMC BIOS settings. If you can afford the best then it will likely be an 8+2 VRM design but there are no guarantees in overclocking.

 

[Wayward_Son]

This brings up a question I've been wanting to ask. I don't want to threadjack, but it seems like this is an appropriate place to ask:

What is the most likely culprit of vDroop? If I notice that vcore is fluttering or dropping a little bit while stress testing, would the most likely culprit by my power supply (Corsair HX620) or my motherboard's voltage regulation (Gigabye GA-MA785GPMT-UD2H)? I think my motherboard uses a 4+1 phase voltage regulator. It's a mATX board with no heatsink over the mosfets, so it's no an enthusiast-class board... then again, despite Corsair's known quality, it is possible that I happened to get a less-than-stellar one.

Is vDroop more likely to be caused by poor motherboard voltage regulation or poor power supply voltage regulation?

 

[trents]

Increasing the number of phases in a VRM (power supply module), smoothes out power ripple and noise which is important for clean CPU power. The current may or may not increase with more phases depending on the VRM design. More phases can be better in reducing ripple/noise but don't get hung up on phases as the CPU makers specify what the minimum CPU VRM must be for stable, reliable performance - and they OC their CPUs everyday.

A quality 4 or 8 phase VRM on an AMD mobo even with significant OC'ing should be just fine. A 4+1 and 8+2 VRM design means that the mobo uses a split power plane which is important for independent adjustment of the CPU/IMC BIOS settings. If you can afford the best then it will likely be an 8+2 VRM design but there are no guarantees in overclocking.

+1

 

[trents]

This brings up a question I've been wanting to ask. I don't want to threadjack, but it seems like this is an appropriate place to ask:

What is the most likely culprit of vDroop? If I notice that vcore is fluttering or dropping a little bit while stress testing, would the most likely culprit by my power supply (Corsair HX620) or my motherboard's voltage regulation (Gigabye GA-MA785GPMT-UD2H)? I think my motherboard uses a 4+1 phase voltage regulator. It's a mATX board with no heatsink over the mosfets, so it's no an enthusiast-class board... then again, despite Corsair's known quality, it is possible that I happened to get a less-than-stellar one.

Is vDroop more likely to be caused by poor motherboard voltage regulation or poor power supply voltage regulation?

Fluttering is not the same as vdroop. Fluttering has to do with a constant fluctuation between vcore values. Vdroop is normal under load, just as it's normal for engine RPMs to fall off when you put a car in gear. How much vdroop are you talking about? If it is excessive, that can be a problem. Having said all that, some boards automatically add juice to compensate for vdroop so you that you may even see the vcore increase under load. Other boards provide an option in bios called "Load Line Calibration" that allows you to manually program in an offset.

 

[Archer0915]

The number of phases is more than just a simple the more the better. The control mechanism and possible user manipulation is also very important.

Phase shifting is a way to keep any one mosfet from reaching a terminal temperature. By splitting the load, the maximum overall temperature is lowered, thus allowing for better power management and less heat to be generated.
This information : http://www.hardwaresecrets.com/arti...e-Motherboard-Voltage-Regulator-Circuit/616/6 : from an article at Hardware Secrets appears to be in order; a few errors here and there, but as I said, in order.

http://www.techreaction.net/2010/08/09/review-asus-p7p55d-e-pro-a-step-above/

Though above does not go deep into it, any name brand (with a good reputation) board that offers a high number of phases will be less likely to go pop. Though this may not be a limiting factor in your top OC it will be a limiting factor in a high sustained OC.

 

[QuietIce]

currently have a 785G mobo w/ 8 power phases

looking at the biostar ta890fxe as an "upgrade" but noticed it only has 6 power phases

what affect will this have on overclocking? should i look for another 890fx board but with 8 power phases? im trying to take my 955 under water and get ~4.2ghz

As Archer pointed out this question is about more than just how many power phases there are. In general, assuming all the components for the power circuits are the same quality, the board with more power phases will deliver cleaner power for a longer period of time. However, we all know board component quality isn't the same and manufacturers don't publish any kind of quality rating for their boards. All we have to go by is the experience of board owners we can trust.

I don't know which 785G board you have but I can tell you Xoke (and a few others) has been running his Biostar 890FXE into the ground and it seems to have delivered everything he's asked of it. From that information I'd guess that board has a quality power system. Whether it's "better" than your current board, who knows? Personally, I'd OC your current board to see what it will do. If you have problems with it there's no reason you can't upgrade to another board at a later date ...