rharriso

diff --git a/.ebextensions/https-redirect-docker-sc.config b/.ebextensions/https-redirect-docker-sc.config
index fa2fbdc..5ea1cae 100644
--- a/.ebextensions/https-redirect-docker-sc.config
+++ b/.ebextensions/https-redirect-docker-sc.config
@@ -33,14 +33,14 @@ files:
            default        "upgrade";
            ""            "";
        }
-       
+

rharriso

###################################################################################################
#### Copyright 2016 Amazon.com, Inc. or its affiliates. All Rights Reserved.
####
#### Licensed under the Apache License, Version 2.0 (the "License"). You may not use this file
#### except in compliance with the License. A copy of the License is located at
####
####     http://aws.amazon.com/apache2.0/
####
#### or in the "license" file accompanying this file. This file is distributed on an "AS IS"
#### BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the

rharriso

const int PIXEL_COUNT = 500 * 500;
auto pixelCounts = thrust::device_vector<unsigned int>(PIXEL_COUNT);
thrust::fill(pixelCounts.begin(), pixelCounts.end(), 0);

rharriso

const int PIXEL_COUNT = 500 * 500;
usigned int *pixelBodyCounts;
cudaMallocManaged(&pixelBodyCounts, PIXEL_COUNT * sizeof(unsigned int));
initZeros<<<numBlocks, blockSize>>>(pixelBodyCounts);

// do stuff with the pixels

cudaFree(pixelBodyCounts);

rharriso

==24803== NVPROF is profiling process 24803, command: ./main-thrust
==24803== Profiling application: ./main-thrust
==24803== Profiling result:
            Type  Time(%)      Time     Calls       Avg       Min       Max  Name
 GPU activities:   74.98%  157.91ms         2  78.954ms  77.478ms  80.430ms  void thrust::cuda_cub::core::_kernel_agent<thrust::cuda_cub::__parallel_for::ParallelForAgent<thrust::cuda_cub::__transform::unary_transform_f<thrust::counting_iterator<unsigned int, thrust::use_default, thrust::use_default, thrust::use_default>, thrust::detail::normal_iterator<thrust::device_ptr<float>>, thrust::cuda_cub::__transform::no_stencil_tag, initRandomPrg, thrust::cuda_cub::__transform::always_true_predicate>, long>, thrust::cuda_cub::__transform::unary_transform_f<thrust::counting_iterator<unsigned int, thrust::use_default, thrust::use_default, thrust::use_default>, thrust::detail::normal_iterator<thrust::device_ptr<float>>, thrust::cuda_cub::__transform::no_stencil_tag, initRandomPrg, thrust::cuda_cub

rharriso

struct initRandomPrg {
  float minValue, maxValue;

  __host__ __device__
  initRandomPrg(float _mnV=0.f, float _mxV=1.f) : minValue(_mnV), maxValue(_mxV) {};

  __host__ __device__
  float operator()(const unsigned int n) const {
    thrust::default_random_engine rng;
    thrust::uniform_real_distribution<float> dist(minValue, maxValue);

rharriso

==21393== Profiling application: ./main-cuda.cuda
==21393== Profiling result:
            Type  Time(%)      Time     Calls       Avg       Min       Max  Name
 GPU activities:   73.42%  227.76ms         2  113.88ms  113.37ms  114.40ms  void initRandom<float>(float*, float, float)
                   26.58%  82.465ms        10  8.2465ms  4.8140ms  39.102ms  void add<float>(float*, float*, float*)
      API calls:   61.88%  310.24ms        11  28.204ms  4.8177ms  227.74ms  cudaDeviceSynchronize
                   32.78%  164.31ms         3  54.771ms  49.680us  164.11ms  cudaMallocManaged
                    4.87%  24.428ms         3  8.1428ms  8.1220ms  8.1685ms  cudaFree
                    0.32%  1.6288ms        12  135.74us  19.770us  559.03us  cudaLaunch
                    0.11%  528.26us        94  5.6190us     240ns  229.88us  cuDeviceGetAttribute

rharriso

// add two arrays
template<typename T>
__global__ void add(T *output, T *inputA, T *inputB) {
  int idx = (blockIdx.x * blockDim.x) + threadIdx.x;
  output[idx] = inputA[idx] + inputB[idx];
}

template<typename T>
__global__ void initRandom(T *arr, float minValue, float maxValue) {
  int idx = blockIdx.x * blockDim.x + threadIdx.x;

rharriso

#include <iostream>
#include <math.h>
#include <cstdlib>
#include <vector>

int main () {
  int N = 8000 * 8000; // 800px x 800px image
  int iterations = 10;
  int size = N*sizeof(float);
  float *x, *y, *output;

rharriso

CC:=/usr/local/cuda-9.1/bin/nvcc
PROF:=/usr/local/cuda-9.1/bin/nvprof

run: main-cuda.cuda 
	./main-cuda.cuda

prof: main-cuda main-cpu main-thrust
	${PROF} ./main-cuda
	@echo
	${PROF} ./main-thrust