Code fragments for Satori Speed Up Green Up Hackathon during IAP 2020

cuml-tsne.py

# demo for profiling python using command
#
# nvprof --print-gpu-trace python cuml_tsne.py

from sklearn.datasets import load_digits
X, y = load_digits().data, load_digits().target
from cuml.manifold import TSNE
tsne = TSNE(n_components = 2)
X_hat = tsne.fit_transform(X)

# code is based on https://medium.com/rapids-ai/tsne-with-gpus-hours-to-seconds-9d9c17c941db
# and needs sklearn and cuml, which are both available for ppc64le in conda channels. 
# 

satori-green-up-hackathon-iap-2020

Gists for speed-up, green-up Satori hackathon IAP 2020

o vector-add.cu 
  Andrew Kirby code for nvprof example

vector-add.cu

#include <stdio.h>
#define N 1048576

__global__ void add_vectors(int *a, int *b, int *c){
    int id = blockDim.x * blockIdx.x + threadIdx.x;
    if(id < N) c[id] = a[id] + b[id];
}

int main(){
    size_t bytes = N*sizeof(int);
    int *A = (int*)malloc(bytes);
    int *B = (int*)malloc(bytes);
    int *C = (int*)malloc(bytes);
    int *d_A, *d_B, *d_C;
    
    cudaMalloc(&d_A, bytes);
    cudaMalloc(&d_B, bytes);
    cudaMalloc(&d_C, bytes);
    for(int i=0; i<N; i++){
        A[i] = 1;
        B[i] = 2;
    }
    
    cudaMemcpy(d_A, A, bytes, cudaMemcpyHostToDevice);
    cudaMemcpy(d_B, B, bytes, cudaMemcpyHostToDevice);
    
    int thr_per_blk = 256;
    int blk_in_grid = ceil( float(N) / thr_per_blk );
    
    add_vectors<<< blk_in_grid, thr_per_blk >>>(d_A, d_B, d_C);
    cudaMemcpy(C, d_C, bytes, cudaMemcpyDeviceToHost);
    
    free(A);
    free(B);
    free(C);
    cudaFree(d_A);
    cudaFree(d_B);
    cudaFree(d_C);
    return 0; 
}