Will you move to OpenGL V as soon as possible? (for client)

It´s called that PA is first of all CPU-bound. As the plans for the next iteration of OpenGL are to cut down overhead a lot it would make sense to offer support for that one ASAP. (better: ASAIL= AsSoonAsIt´sLaunched)
Will you do that? Or do you want to stick with the current OGL-Versions to don´t hit compatibility?
(proberbly OGL V even is compatible with most DX 11 and OGL 4-Hardware...)

PA barely uses any CPU though?

Squishy client side is cpu bound. Mainly due to fact that ogl 3 can only use 1 main thread for rendering. Varrak has managed to offload some stuff onto other threads which helps immensely (pa now uses 3 threads quite well) however with a high end gpu the cpu (main thread) is still what limits performance.

Ogl v (thought it was called ogl 'n' for next?) Is a new modern rewrite that removes this limitation, although to do this it breaks backwards compatibility so may require significant changes to the render engine to support and currently we don't know what hardware will be supported...

Ah alrighty, which is /why/ it utilizes less CPU, a bad thing in this case.

Let's say fair at moment PA client performance is already great on any reasonable CPU/GPU and main bottleneck is on server side. After developers implemented feature that stops unit rendering once icons is visible I have great FPS even of my integrated Intel GPU.

Does that mean Open GL doesn't have multi-core support up to this day?

It has partial multi core support... Dx11 is somewhat better but also limited. Amd mantle is the only available fully multi threaded api out right now. Dx12 and open gl n are both looking to implement a similar system. Whilst there is minimal benefit to mantle in gpu bound games like shooters, rts is an area that had potential huge gains.

Uber discussed using mantle but decided against it as its currently only available on windows and is vendor specific so would only benefit a small section of the community (ie too little rearward for the time investment). On the other hand ogl n will be fully cross platform and pa is an ogl game, so that makes a lot of sense as a future development.

Nope, it's have nothing to do with multi-core support itself, but with limitations applied to renderer code and engine because API wasn't designed to push data to GPU from multiple threads. PA client isn't really CPU-intensive application so it's only affect guys who have powerful GPU, but slow CPU that might be unable to load GPU due to API overhead.

In same time OpenGL implementation from different vendors might able to use more than one core and there is also some advanced features like Nvidia's "Threaded Optimization". There also number of trick Uber already use to achieve lower CPU overhead.

Ty, that stuff is always interesting to read about. I was ofc speaking of efficient use of multiple cores, not the bare bones idea of 'support.'^^

The lack of multicore-support in general just seems so strange to me. Even today many games/engines/api's
are lacking in that regard, while quad-cores are quite common since like 10+ years. Is it so hard to get a program to scale with cores/threads? And even when it is, the groundwork for tackling those issues should've been done long ago...