In 2005 C coders were like "IF Only".
What's happening after Doom3 becoming open-sourced....Nothing.
By the time it becomes playable on high settings with multi-player mode, you and me will have lost half our hair....or fallout 4 might get opensourced or one plus ten might work as backup memory for your brain.
There is only so much money that can be made with fossilised technology and code, all the major players including Microsoft see a future with linux.
Will Microsoft make their own linux 'gaming' platform?
A Microsoft gaming PC of the Future if Steam fails--- A windows platform(linux based) with 8GB DDR4 or some other form of future memory, ARM
Octa-core CPU and GPU pumping a locked 30fps on a 4K panel.
A Blender rig being powered by an Octa-core HSA enabled, ARM solution.
Where is HSA for Blender, AMD? Although 2.76 has OpenCL support, its flimsy and requires Ubuntu.
A C coder can make a decent living submitting patches for the Kernel and getting paid for it by major companies. Linaro is already making strides and
hiring 18 year old kids, something corporate organized, good behaviour, degree boasting halo of brightness could never make anyone productive, the 80's have passed and the train has left the station. Microsoft and Apple know this very well.
Companies are realizing, Code needs to be optimized, AMD understands this as does any other brand. Did you know you can run
a full game on a CPU if compiled properly using LLVM and OpenCL, apart from the fact a buffer overflow is imminent...debugging could take a year. Compilers are really dumb, until the day arrives you could just emulate everything instead of risking real time faults....well LLVM does that, kind of.
Octa-core desktops processors really mean NOTHING and if your code paths for multithreaded are still emulating processing patterns from an initial code
base. Its serious work that takes YEARS
. The proof of the pudding is in your modern laptop/ desktop.....ask yourself if you own an Intel 2500k or an FX6300 and if they are worth upgrading? The fastest chip of the last 2 generations are twice as fast as a medium segment chip of the current generation.
How hard do you think rewriting the GPU code for future pipelines in an optimized fashion will take? No x86 company can afford 2 years writing code when they have to release silicon every 2 years. ARM chips on the other hand are built around software technology(content/code/multimedia) standards.
This is clearly evident with ARM devices that run crude code like java stuff on GPU and instructions on CPU that can leverage instead of bottlenecking, with IPC boasting and Clock speeds on desktop parts.
Running multiple chrome browsers on a 1080 mobile device having only 2GB Ram!
Today a new compiler is like new hardware... comparing a dual core with an octa-core.