Get comfortable being uncomfortable

Sandeep Kumar Behera sandeepkumar-skb

Get comfortable being uncomfortable

Deep Learning acceleration and computer architecture enthusiast.

sandeepkumar-skb / py_multiprocess_computeB.py

Last active March 31, 2021 03:49

This is a multiprocessing example speeding up a compute bound problem.

	import multiprocessing as mp
	import time as time

	def square():
	for i in range(1000000):
	x = pow(i, 2)

	if __name__ == "__main__":
	num_iter = 10
	start = time.time()

sandeepkumar-skb / py_thread_compute_bound.py

Created March 31, 2021 03:34

This python threading example demonstrates python threading in compute bound situation.

sandeepkumar-skb / reduce.cu

Created February 17, 2021 05:22

	#include <iostream>
	#include <chrono>

	#define BLOCK_SIZE 256

	inline void gpuAssert(cudaError_t err, const char *file, int line)
	{
	if (err != cudaSuccess){
	printf("%s in %s at line %d\n", cudaGetErrorString(err), file, line);
	exit(EXIT_FAILURE);

sandeepkumar-skb / transpose.cu

Created January 30, 2021 04:32

	#include <cuda.h>
	#include <stdio.h>
	#define BLOCK_SIZE 32
	#define NUM_REPS 100

	inline void gpuAssert(cudaError_t err, const char *file, int line)
	{
	if (err != cudaSuccess){
	printf("%s in %s at line %d\n", cudaGetErrorString(err), file, line);
	exit(EXIT_FAILURE);

sandeepkumar-skb / cooperative_groups.cu

Last active January 29, 2021 23:29

	#include <cooperative_groups.h>
	#include <algorithm>
	#include <cuda.h>
	#include<stdio.h>

	using namespace cooperative_groups;

	inline void gpuAssert(cudaError_t err, const char *file, int line)
	{
	if (err != cudaSuccess){

sandeepkumar-skb / pointer_aliasing_demo.cu

Created December 21, 2020 01:36

	#include <stdio.h>
	#include <chrono>
	#include <iostream>

	#define BLOCK_SIZE 128

	inline void gpuAssert(cudaError_t err, const char *file, int line)
	{
	if (err != cudaSuccess){
	printf("%s in %s at line %d\n", cudaGetErrorString(err), __FILE__, __LINE__);

sandeepkumar-skb / histogram_shmem.cu

Last active January 30, 2021 04:35

	#include <stdio.h>
	#include <iostream>
	#include <chrono>

	#define BLOCK_SIZE 256
	#define GRID_SIZE 72 //Turing Titan RTX
	#define OUT_SIZE 256

	inline void gpuAssert(cudaError_t err, const char *file, int line)
	{

sandeepkumar-skb / conv_bn_folding.py

Last active January 10, 2021 21:17

Folding BN into convolution

	import torch
	import torch.nn as nn
	import copy
	import torchvision.models as models


	class BN_Folder():
	def fold(self, model):
	mymodel = copy.deepcopy(model)
	mymodel.eval()

sandeepkumar-skb / openmp_thread_pool.cpp

Created November 6, 2020 23:20

mkolod / redirect_streams_and_cuda_checks.cu

Last active September 27, 2020 04:12

Redirect Streams and CUDA checks