big page, maybe in compiled or parallel specific stuff. cuda, opencl, cuBLAS, rocBLAS. actually probably want to separate stuff in base headered c and related, and GPU stuff. standard portable intermediate representation (SPIR) used in vulk and opencl ROCm Radeon Open Compute platforM oneAPI. open standard for GPUs, ai accelerators etc "and cuda" in big title of the gpu stuff opencl and sycl. same big page as cuda. + integrated graphics gpu memory nvidia: nvidia-smi program
Instruction level parallelism in that page. How much a program can be parallelised simulatneous multithreading as concept relaed to superscalar CPUs (multiple execution units)
c parallel
#include <pthread.h>
pthread\_t threads[2];
// start
pthread\_create(&threads[0], NULL, myFun, (void *) &thread\_args);
// block until thread finished
pthread\_join(threads[0], NULL);in big page on parallel stuff, h3 on linear algebra on parallel: hyperthreading