4+1 vs. 8+2(8+1) Phase Power Design

[Bladesinger7x]

What is the deal with various 4+1/8+2/8+1 Phase Power Design
on AMD AM2+/AM3 boards.

What are the pros/cons of 8+2(8+1) over 4+1.

Is 4+1 (Asus M3A78-T) still ok for 940BE when overclocking
or should I consider 8+2(8+1) boards?

[chew*]

8 + phase when implemented correctly requires less power to acchieve clocks,its cleaner power, it's not as stressfull on PSU, and is considered to have a longer longevity. I might add there are very few true 8 phase designs. Most are split 4 + 4.

4 + works, if properly implemented and kept cool, if not it will eventually blow up when pushed hard.

[cdawall]

one of the few 8+2 i know of is the Asus Crosshair II formula

[Bladesinger7x]

Well the CPU (and GPU) will be under water so no additional
airflow will be coming from the CPU fan, however, I do have
a 250mm fan on the case side (Sharkoon Seraphim case).

I doubt I will overclock the CPU by much (if at all) so 4+1
should be fine?

What is the verdict on Asus M3A78-T pwm design? Ok or not.
Or should I wait for the M4A78-E which has 8+1 design?

[FoxRacR17]

one of the few 8+2 i know of is the Asus Crosshair II formula

Isnt the M4A79 8+2 as well? Or is that one a 4+4 like mentioned earlier in the thread?

[chew*]

Preety sure they are all 4 + 4, ATI made a true 6 phase back in the 939 days.

[cdawall]

Preety sure they are all 4 + 4, ATI made a true 6 phase back in the 939 days.

asus lists the crosshair II as a 8+2?

[chew*]

asus lists the crosshair II as a 8+2?

yah they list alot of phases as 8 + 2 but I do not think to date anyone has made a "TRUE" 8 phase, most of them are 4 + 4 + 2 and they call them 8 + 2phase.

The largest phase I have ever seen was a "true" phase was 6 phase by ATI.

[FoxRacR17]

so then what is the big difference between 4+4+2 and a TRUE 8+2. I understand that is has to do with voltage ripple and the more phases the less voltage ripple.

[chew*]

so then what is the big difference between 4+4+2 and a TRUE 8+2. I understand that is has to do with voltage ripple and the more phases the less voltage ripple.

there is no difference because 8 + 2 doesn't truly exist.....

[FoxRacR17]

there is no difference because 8 + 2 doesn't truly exist.....

haha, ok let rephrase then, what WOULD be the difference?

[chew*]

haha, ok let rephrase then, what WOULD be the difference?

cleaner power delivery to the cpu and probably no vdroop.

[xPliziT]

Allright, let me clear things up.

I have asked a similar question long time ago over at amdzone forums.

http://www.amdzone.com/phpbb3/viewto...140825#p140825

quote:

Time to whip out my electronic circuits textbook...
The topic is not that simple, an 8-phase or 10-phase supply won't always be better than a 5-phase supply. The switches and other parts used in the 8-phase supply you mentioned are not identical to the ones used in the 5-phase AMD boards. The 8-phase parts probably cost less and are much smaller. Both phase supply setups can meet the same requirements regardless of how many phases you use, it just depends mostly on how much you want to spend per part, how much space you have to work with, heat dissipation, your typical design problems. The 8-phase switchers can be cheaper and smaller parts because they will not be stressed as hard as if they were in a 5-phase setup. Also, the phase IC's AMD uses are special, in that they can dynamically change the number of phases depending on how much juice the cpu is asking for, so there is probably no leeway for motherboard designers to create a gimmick board with a 16 phase power supply to woo the enthusiast crowd. I remember reading that the AMD IC's weren't designed for anything more than a 6-phase setup because anything more than that is completely unnecessary for the K8 and K10, and the only reason for having the option of 6 was for extreme overclocking. Considering this, I'd assume that unless you OC the chip enough, it's going to be using 5-phases under load with a 140 watt chip, so 5 will probably be overkill for 99% of the market. 4 and 3 phase boards are meant be paired with a budget processor or a more power friendly Deneb.
end quote

this should sum it up it perfectly.

[Oese]

the dfi m2r hast 6+1 phase design... and i think its true 6+1

[Hemi345]

Unless the board manuf. states how many phases a board has, how can you tell? Count the mosfets or chokes?

[Mara]

Well.. to really know you need to check what VRM controller chip is being used and then pull out the datasheet.

The Vdroop is shouldn't have anything to do with how many phases there is. There are couple reasons for the droop. Most common is the voltage drop caused by copper (resistance) loss in the motherboard. Second is that some PSU have "programmed droop" where voltage drops linearly with function of current. (IIRC, Intel specifies this method for its CPU's to improve dynamic load behavior.)

Droop can be compensated if you take the VRM voltage feedback at the end of the load chain. (Like directly under the CPU.) The drawback is that during dynamic loading the voltage does overshoot and undershoot more than with "programmed droop". This is caused by the limited loop bandwidth of the VRM control loop where it can not respond fast enough for the changed load conditions. Ultimately the LC-filter used in all VRM outputs has a pole frequency that is very low when compared to the required load change speed.

[Dumo]

8+2 phase power on Asus M4A79T Del. (thats their word)..

[FoxRacR17]

8+2 phase power on Asus M4A79T Del. (thats their word)..

Yeah, that is their word and hopefully its not false advertising. How could we tell for sure?

[Dumo]

Try overclock voltages in bios, see if there vdroop or instability.
Some mobo can't take high vcore for long benching session.

[G.Foyle]

There seems to be a lot of confusion about ASUS 8+2 power circuitry.
Both Crosshair 3 and M4A79T-Deluxe use L6740 PWM controller made by ST Micro.
It's a 4+1 phase controller, which means power regulation is done in 4 phases for core voltage and in 1 phase for NB voltage. ASUS uses double inductors for each phase to split the load to more components and thus increase longevity and decrease temps. It's not a true 8 phase regulation, the same as for example Rampage II Extreme PWM being a 8-phase, double inductor design, not a 16-phase PWM.


Whether it is a blatant lie depends on what does "phase" mean for you. ASUS and Gigabyte seem to think that inductor = phase, but rest of the world knows that "phase" is "one part or portion in recurring or serial activities or occurrences logically connected within a greater process, often resulting in an output or a change." (Wikipedia). Phase is not something that exists on motherboard PCB, it's a parameter of the voltage control process. So for most people, yes, ASUS and Gigabyte and many other manufacturers lie about their PWM designs.

EDIT: The same design is used on M3A79-T Deluxe.

[chew*]

Yep good info man.....

I believe the gigabyte is a 6 phase......but they go about 6 phase a diff way.....

[xdan]

Hello guys, i'm not to expert in phase power design, can one of us please tell me how man phases for CPU have the board in the picture?
5 CPU+ 1chipset ( or4 4+2?)? The board is ASROCK A790GMH/128. I'm planning to buy one, it's PWM seems to be solid even though it has no cooling..
And. i have heard that Asrock MB overvolts the voltage a bit, quite strange...
Opinions if this 90$ can hold a X3 720 at 1.5v?

[chew*]

Judging by the components used I would not plan on breaking any records....An 8 pin input at the PWM would not hurt......seems they only used 4pin....they are likely overvolting to comnpensate for vdroop....

[xdan]

Judging by the components used I would not plan on breaking any records

I'm not planning to break any records , i'm planning a decent overclock 3.7-3.8GHZ no more than 1.5v..
Actually from a 90$ Mb it will be quite good...

[DedEmbryonicCe1]

There seems to be a lot of confusion about ASUS 8+2 power circuitry.
Both Crosshair 3 and M4A79T-Deluxe use L6740 PWM controller made by ST Micro.
It's a 4+1 phase controller, which means power regulation is done in 4 phases for core voltage and in 1 phase for NB voltage.

That is the same controller used by Biostar for the TA790GXE 128M and presumably all the other 4+1 boards they released at the same time. The TA890FXE is now on the L6717 controller. My experience with the L6740 from my TA790GXE 128M was that it was best to disable the active phase switching when overclocking. With phase switching enabled I could crash/shutdown the board stopping a Prime95 run whereas the same frequencies and voltages were perfectly stable in the same scenario with phase switching disabled via the BIOS.