Because the hardware is not capable of it. AMD provides a custom scheduler driver that's supposed to inject in games through the Xbox Game Bar, which sets the affinity of the detected game to run on the more appropriate CCD for best performance. The caveats of this are obvious:
1. It needs the Xbox Game Bar to be installed, activated and updated
2. AMD and Microsoft need to issue updates for it to work (something which is not guaranteed to happen in the very long term)
3. You can't get the benefits of using both sides of the processor at once. It works better as an 8 core or 8 core+3D cache configuration, rarely if ever well together
You may manually assign processor affinity with a tool like Process Lasso, otherwise the only way to activate or deactivate CCDs is with a system reboot. Ryzen Master should be able to automate that process with a one-click I believe, but it is not a seamless experience.
You seem to be contradicting yourself, because you say that there is no problem, and then you proceed to point out the exact problem, lol.
Windows is not optimized for this topology, and it is very likely that it will never be. Intel's thread director works around this issue seamlessly through hardware-based runtime feedback to the operating system by literally telling it where to issue and physically execute each instruction, something that AMD hardware is not currently capable of.
Your reading comprehension is terrible.
I said that Windows and drivers have no problem scheduling tasks.
What doesn't exist is specific software optimization. SOFTWARE <<< SOFTWARE <<< SOFTWARE.
SOFTWARE like Adobe and games Software like Call of Duty.
If you create a game or software to use only 4 or 6 cores, why do you think the game will use more cores?
You want an R9 9950 to use 16 cores in games that weren't made to use 16 cores.
Games on the Ryzen 9800x3D are already optimized to use the 3DX CACHE, which is exactly why it is the best CPU for gaming today.
Now what you think should happen is that the R9 9950X3D would perform better than the 9800X3D just because it has more cores. But games and software do not use more cores, they were not PROGRAMMED TO USE MORE CORES.
There is no scheduling problem in this.
You do not know the slightest about hardware development and driver creation.
What I say is a fact and there is proof of it.
If there was such a problem, the same game on Linux would perform better just because it was running Linux.
But we know that games on Linux do not perform better. If the Linux Kernel were miraculous, games would be made for them, but they are not.
MacOS, which uses a UNIX-based kernel, would perform better, but it does not.
Android, which uses the Linux Kernel, would perform better.
SteamOS, which uses the Linux Kernel, would perform better, but it also does not.