In discrete GPUs, VRAM is RAM memory that lives on the GPU's PCB.
They are located in separate chips to the GPU's compute, since just like for CPUs, you can't put both on the same chip as the manufacturing processes are different and incompatible.
Integrated GPUs don't have VRAM and just instead use the same RAM as the CPU.
Tested on Ubuntu 23.10 with P14s:
sudo apt install hipcc
git clone https://github.com/ROCm/HIP-Examples
cd HIP-Examples/HIP-Examples-Applications/HelloWorld
make
TODO fails with:
/bin/hipcc -g   -c -o HelloWorld.o HelloWorld.cpp
clang: error: cannot find ROCm device library for gfx1103; provide its path via '--rocm-path' or '--rocm-device-lib-path', or pass '-nogpulib' to build without ROCm device library
make: *** [<builtin>: HelloWorld.o] Error 1

Articles by others on the same topic (1)

A Graphics Processing Unit (GPU) is a specialized electronic circuit designed to accelerate the processing of images and videos for output to a display. While CPUs (Central Processing Units) are optimized for general-purpose computing tasks, GPUs are tailored for rendering graphics and performing complex mathematical calculations efficiently, particularly those that can be processed in parallel.