bits from different posts that are relevant, my emphasis also added
Originally Posted by The Stilt
Phase count isn´t everything. Most of the cheaper high phase count boards have doubled low(er) quality VRM (i.e. quantity over quality). You need to be able to dissipate the heat the VRM produces, regardless the phase count.
the UD3 R5 and UD3P have digital and doubled VRM. … The IR drivers paired with the VRM are also stronger than the ones on analog boards.
970A-UD3P has 8 virtual phases for the CPU each with a single NEC / Renesas uPA2724 low-side mosfet in it. Each of these fets are rated for 29A continuous current at 25°C meaning in total you can pull 232A from the VRM at this temperature.
ASUS boards have two sets of low-side fets, where as Gigabyte has only one.The current through each of the fets will be 33% lower than on Gigabyte. Each low-side fet on ASUS (OnSemi 4937N) is rated for 44A @ 85°C where as the NEC uPA2724 fets on Gigabyte are rated to ~16A in the same temperature.
For the UD3P mosfet model the de-rating factor is following:
100% @ 25°C (29A per fet or 232A in total)
80% @ 50°C (23.2A per fet or 185.6A in total)
60% @ 75°C (17.4A per fet or 139.2A in total)
40% @ 100°C (11.6A per fet or 92.8A in total)
24% @ 120°C (6.96A per fet or 55.7A in total)
0% @ 150°C (tJMax, absolutely no load allowed)
- On 220W TDP models the VRM must be able to provide 140A of current
- On 125W TDP models the VRM must be able to provide 110A of current
- On 95W TDP model the VRM must be able to provide 85A of current
The values below are rough estimates and only apply on this specific board.
They are safe values and within these limits the board should provide it´s normal expected life time.
- FX-8K series CPUs, < 4.4GHz, <1.3875V = Up to 80 - 85°C VRM temperature
- FX-8K series CPUs, < 4.7GHz, < 1.4500V = Up to 65°C VRM temperature
- FX-9K series CPUs, stock = Up to 75°C VRM temperature (impossible without water cooling)
Compared to Rev 1.0 they have removed several polymer capacitors used for filtering (they are expensive on ODM standards) and removed some of the digital NCT3933 sink / source controllers from the board too. Having the "extra" NCT3933´s removed explains why the voltage control options in Rev 2.0 are worse.
There doesn´t appear to be any change to the VRM despite the fewer load-line options would indicate that. However I base that just on the low resolution images available so I cannot really tell for sure unless someone provides higher res. images of the VRM (with heatsink removed) and of the VRM controller itself. If they have replaced the gate drivers for example with cheaper ones that definitely can effect on the maximum overclockability.
Gigabyte seems to use the same base bios for ALL of their motherboards, regardless if they are Intel or AMD based. They just seem to apply patch on top of a patch until they get it even remotely "working". There is no way you can otherwise explain the hidden bios options these bioses have. There is plenty of Intel exclusive stuff in the bios of these boards