not to mention. Whilst not required, having a Crossfire or SLI system as a developer is handy, as it simplifies bug-fixing on the developers side.
If he/she is planning to code for Vulcan (most likely unless he picks up a off the shelf engine like Unreal Engine 4).
However, the amount of hardware needed to make a game is entirely up to the developer, and whether he will need 8 cores 16 threads, or 2 cores 2 threads is simply up to how complicated he/she is planning to make the game/engine.
It is usually a really bad idea to run two lower end cards over a single high end card.
Sometimes Crossfire/SLI doesn't work at all, and in those cases you are stuck with the performance of a single low end card.
A lot of times scaling is very bad. Not so much an issue weeks/months after a AAA game is released, but at launch and for games where the developers don't care that much about performance (read: most of them) you usually end up with pretty bad scaling. Two cards might only perform like one and a half. You will most likely experience more glitches in games with multi-GPU setups as well.
Single GPU setups are just far more consistent than multi GPU ones.
If you need the system within 6 months then wait 6 months and then decide which parts to buy. There is no point in trying to decide which parts you should buy half a year from now. Nobody knows what the prices will be like, which parts will be out or even how they will perform. Anyone saying anything else is a liar, ignorant or a wizard with a crystal ball that can see into the future.
Again, buy what you need, when you need it. Don't plan your build 6 months ahead of time because your plans will most likely change as prices and the hardware available changes.
What do you mean "spend all of that on something too expensive"? Your budget is 6000... Trust me, you can get a computer for under 6000 without only buying AMD parts. You should focus less on what brand you buy from, and more on what the product you are buying can/can't do.
You are complicating things way too much. You said that you have just started to learn game development. Having support for dual GPUs should be the least of your priorities right now. You have to learn to walk before you can run.
AMD cards can't do PhysX on the GPU, but they can do it on the CPU (although that is not recommended for very PhysX heavy games). You can also use other physics engines like Havok.
He is just starting out. Corssfire and SLI support should be very low on the priority list. Learning how to make a game run on a single GPU comes first.
I think you forgot something in the mid section there. Seems like there should be something after "if he is planning to code for Vulcan."
Yes, how much hardware you need entirely depends on the game. However, this is someone who has just started learning game development. When someone who is new in a field tries to estimate what will be needed in 6 months they usually end up being completely wrong. That's why I keep saying that you should only buy what you need, and when you need it. Anything else is basically gambling with your money.
You can not evaluate the performance of a library using a single implementation.
Ashes of the Singularity uses one aspect of DX12 that works well for that game. That does not mean every AAA developer will rush out and use it because it may not offer any real benefit to them.
Along with those threads I would suggest that you watch these videos from JayzTwoCents. He is considered the "Watercooling God of YouTube" and he knows his shit. If you still have questions after that there are plenty of people on the forum that would be willing to help you figure it out (myself included).