These are few HPC implementation codes/snippets which I am sharing with permission to share them. You can use them if you want. :)
Some snippets/code might be absent/faulty due to the reasons of licensing/trade-secret/good will in the favor of the company. Also, few changes might be required to make executable. Sorry for this! (ー_ー)!!
This is a simple CUDA interpretation of a FEA problem I studied in my bachelors. The reference can be found here.
Euler's equation for incompressible, inviscid, isotropic fluid using the Finite Difference Method. I suggest creating a function to read data from a file and then writing the results to other (Hint: #include fstream).
Pretty much self explanatory. I recommend above hint.
The code solves the Poisson equation using the Jacobi iterative method on a 2D grid. The code assumes a Dirichlet boundary condition, where the potential is specified on the boundary of the domain. The Jacobi method updates each point in the grid using a weighted average of its neighbors, and repeats this process until the solution converges to a specified tolerance or a maximum number of iterations is reached.
I am a beginner in OpenCL. This code was ran on Intel-AMD:CPU-GPU combo with some changes in code and parameters defined at the compilation and runtime which I believe you are smart enough to figure out :)
I guess the name is self explainatory. Several runtime parameters are suggested.
FFT based on simple radix algorithm. I suggest use of other highly optimised libraries avaiable if you don't want to hustle xD
Figure it out on your own!
The header file(.hpp) has class that takes inputs required to solve the vorticity equation for a fluid and solves the equation using input fluid parameters. This class takes in the number of grid points in the x, y, and z directions (Nx, Ny, and Nz), the size of the domain in the x, y, and z directions (Lx, Ly, and Lz), and the kinematic viscosity of the fluid (nu) as constructor arguments. It then initializes the fluid variables and allocates memory for them. The solve method takes in a time step dt and solves the vorticity equation for the given time step using the current values of the fluid variables. It updates the vorticity components omega_x, omega_y, and omega_z, and then uses these updated values to update the velocity components u, v, and w. The update is done using finite differences and explicit time-stepping.
CPU optimised CUDA code for 3D Heat Conduction.