My first Android was an HTC Hero, which was released in ~ October of 2009.
One of the first things I did was swap the location of the Maps and Store icons to make it easier to reach on the edge of the phone.
I recall people complaining that same year that the iPhone 1 couldn’t copy or paste text.
:)
Fun fact: The human battery thing is actually a retcon the Wazowskis did at the last moment because they thought the American public would be too stupid to grok the actual understanding of the Matrix.
Humans are an entropic species, they consume more energy than they produce - any synthetic race that tried to harness energy from a net negative energy producer is an idiot.
What the Matrix is, is actually a distributed simulation MATRIX that uses individual human brains as nodes in a shared, hallucinogenic dream, indistinguishable from reality.
The real simulation isn’t so primitive, it doesn’t require people to be popsicle tubes in some crazy dystopian cyberpunk black and red tower attended by insectoid robots.
Instead the entire universe is contained on a single state machine, compromising a [redacted] amount of memory, running in [redacted]. Simulants are never aware of being inside of the simulation, except for rare instances where outsiders occasionally post on Lemmy.
Why they do that, we don’t know. We suspect that it is [all further content redacted].
I’m an AI Developer.
TLDR: CUDA.
Getting ROCM to work properly is like herding cats.
You need a custom implementation for the specific operating system, the driver version must be locked and compatible, especially with a Workstation / WRX card, the Pro drivers are especially prone to breaking, you need the specific dependencies to be compiled for your variant of HIPBlas, or zLUDA, if that doesn’t work, you need ONNX transition graphs, but then find out PyTorch doesn’t support ONNX unless it’s 1.2.0 which breaks another dependency of X-Transformers, which then breaks because the version of HIPBlas is incompatible with that older version of Python and …
Inhales
And THEN MAYBE it’ll work at 85% of the speed of CUDA. If it doesn’t crash first due to an arbitrary error such as CUDA_UNIMPEMENTED_FUNCTION_HALF
You get the picture. On Nvidia, it’s click, open, CUDA working? Yes?, done. You don’t spend 120 hours fucking around and recompiling for your specific usecase.