@dvb_eck@ankkala There's a lot of latency introduced in making memory swappable
GPUs are extremely sensitive to latency and calculate workloads much faster than any CPU would in the same task, but their scope is limited while CPUs have broader instructions/usability for general computing
Personally, as someone who was deployed to the Woke 1 wars in 2014, i take full responsibility for what it ended up looking like a decade later. "It's not my job to educate you" was actually the most insane, detrimental and endlessly parroted line of that early Woke 1 era