the split power plane was a requirement afterall.
I've checked Gigabyte's site, they tell very clearly which mobos support AM3 and which don't ; all their AM2+ mobos support it, and a handful of AM2 mobos do.
Actually the split power planes isn't really a requirement. I think that's more of a marketing thing imposed by AMD, if you want to sell it as a AM2+ board it needs to have split power planes, but not necessarily HT 3.0 (for instance boards with AMD 740G which is a renamed rs690).
In fact I was wrong that no "true" AM2 boards have bios updates for AM2+/AM3 cpus. Case in point gigabyte. The quite popular GA-MA69GM-S2H / GA-MA69GM-S3H boards - nowhere does it say AM3 ready, but if you go look at the supported cpu list (
http://www.gigabyte.com.tw/Support/Motherboard/CPUSupport_Model.aspx?ProductID=2579#anchor_os) it in fact lists all AM3 cpus (except the A II regors which is a pity but it's a start).
Guess I should have gone with gigabyte instead of asus...
Actually AMD says all cpus work with AM2 right here:
http://support.amd.com/us/Processor_TechDocs/43375.pdf - look at the single plane stuff.
Some interesting observations:
- NB voltage and frequency are fixed with dual plane boards
- NB frequency is still fixed in single plane boards, but voltage obviously will change with core voltage. To achieve this AMD both lowered the NB frequency and limited the amount of downvolt in lower cpu p states. In case of the 45nm cpus the voltage difference can be quite large.
- TDP will therefore be higher in single plane boards. In highest p-state because NB is overvolted, in lower p-states (particularly the lowest one) because cpu voltage can't be lowered that much. The difference can be quite large (factor of 2) and in fact even larger for instance for the A II X2 when cpu is in c1e state.
Anyway, clearly this shows it should work, even officially, and the board manufacturers are just unwilling to adapt their bioses.