1. #include "cuda_runtime.h"
2. #include "device_launch_parameters.h"
3. #include <stdio.h>
4. #include <iostream>
5. #include <cstdio>
6. #include <ctime>
7.  
8. using namespace std;
9.  
10. cudaError_t addWithCuda(int *is_prime, unsigned long long int *number, unsigned long long int *sqrtt);
11.  
12. __global__ void addKernel(int *is_prime, unsigned long long int *number, unsigned long long int *sqrtt)
13. {
14.         unsigned long long int index = blockIdx.x * blockDim.x + threadIdx.x;
15.         unsigned long long int i = index;
16.         while (i < *sqrtt + 1)
17.         {
18.                 if (i >1)
19.                 {
20.                         if (*number%i == 0)
21.                         {
22.                                 *is_prime = 1;
23.                         }
24.                 }
25.                 i += blockDim.x*gridDim.x;
26.         }
27.         cudaThreadSynchronize;
28. }
29.  
30. int is_prime_number(unsigned long long int *number)
31. {
32.  
33.         for (long long int i = 3; i < sqrt(*number); i += 1)
34.         {
35.  
36.                 if (*number%i == 0)
37.                 {
38.                         cout << "(cpu)liczba nie jest pierwszan";
39.                         return 0;
40.                 }
41.  
42.         }
43.         cout << "(cpu) liczba pierwszan";
44.         return 1;
45.  
46. }
47.  
48. int main()
49. {
50.    
51.         unsigned long long int *number =new unsigned long long int;
52.         unsigned long long int *sqrtt = new unsigned long long int;
53.         int *is_prime = new int;
54.  
55.         *number = ((unsigned long long int)1) << 61;
56.         *number -= 1;
57.         *sqrtt = sqrt(*number);
58.         *is_prime = 0;
59.  
60.         clock_t start1 = clock();
61.         is_prime_number(number);
62.         printf("Czas wykonywania: %lu msn", clock() - start1);
63.  
64.     // Add vectors in parallel.
65.     cudaError_t cudaStatus = addWithCuda(is_prime,number,sqrtt);
66.     if (cudaStatus != cudaSuccess) {
67.         fprintf(stderr, "addWithCuda failed!");
68.                 system("pause");
69.         return 1;
70.     }
71.  
72.     //printf("{1,2,3,4,5} + {10,20,30,40,50} = {%d,%d,%d,%d,%d}n",
73.         //c[0], c[1], c[2], c[3], c[4]);
74.  
75.     // cudaDeviceReset must be called before exiting in order for profiling and
76.     // tracing tools such as Nsight and Visual Profiler to show complete traces.
77.     cudaStatus = cudaDeviceReset();
78.     if (cudaStatus != cudaSuccess) {
79.         fprintf(stderr, "cudaDeviceReset failed!");
80.                 system("pause");
81.         return 1;
82.     }
83.  
84.         system("pause");
85.     return 0;
86. }
87.  
88. // Helper function for using CUDA to add vectors in parallel.
89. cudaError_t addWithCuda(int *is_prime, unsigned long long int *number, unsigned long long  int *sqrtt)
90. {
91.     unsigned long long int *dev_a = 0;
92.     unsigned long long int *dev_b = 0;
93.     int *dev_c = 0;
94.     cudaError_t cudaStatus;
95.  
96.     // Choose which GPU to run on, change this on a multi-GPU system.
97.     cudaStatus = cudaSetDevice(0);
98.     if (cudaStatus != cudaSuccess) {
99.         fprintf(stderr, "cudaSetDevice failed!  Do you have a CUDA-capable GPU installed?");
100.         goto Error;
101.     }
102.  
103.     // Allocate GPU buffers for three vectors (two input, one output)    .
104.     cudaStatus = cudaMalloc((void**)&dev_c, sizeof(int));
105.     if (cudaStatus != cudaSuccess) {
106.         fprintf(stderr, "cudaMalloc failed!");
107.         goto Error;
108.     }
109.  
110.         cudaStatus = cudaMalloc((void**)&dev_a, sizeof(unsigned long long int));
111.     if (cudaStatus != cudaSuccess) {
112.         fprintf(stderr, "cudaMalloc failed!");
113.         goto Error;
114.     }
115.  
116.     cudaStatus = cudaMalloc((void**)&dev_b, sizeof(unsigned long long int));
117.     if (cudaStatus != cudaSuccess) {
118.         fprintf(stderr, "cudaMalloc failed!");
119.         goto Error;
120.     }
121.  
122.     // Copy input vectors from host memory to GPU buffers.
123.     cudaStatus = cudaMemcpy(dev_a, number, sizeof(unsigned long long int), cudaMemcpyHostToDevice);
124.     if (cudaStatus != cudaSuccess) {
125.         fprintf(stderr, "cudaMemcpy failed!");
126.         goto Error;
127.     }
128.  
129.     cudaStatus = cudaMemcpy(dev_b, sqrtt,sizeof(int), cudaMemcpyHostToDevice);
130.     if (cudaStatus != cudaSuccess) {
131.         fprintf(stderr, "cudaMemcpy failed!");
132.         goto Error;
133.     }
134.  
135.         clock_t start = clock();
136.        
137.  
138.     // Launch a kernel on the GPU with one thread for each element.
139.     addKernel<<<12, 512>>>(dev_c, dev_a, dev_b);
140.  
141.     // Check for any errors launching the kernel
142.     cudaStatus = cudaGetLastError();
143.     if (cudaStatus != cudaSuccess) {
144.         fprintf(stderr, "addKernel launch failed: %sn", cudaGetErrorString(cudaStatus));
145.         goto Error;
146.     }
147.    
148.     // cudaDeviceSynchronize waits for the kernel to finish, and returns
149.     // any errors encountered during the launch.
150.    /* cudaStatus = cudaDeviceSynchronize();
151.     if (cudaStatus != cudaSuccess) {
152.         fprintf(stderr, "cudaDeviceSynchronize returned error code %d after launching addKernel!n", cudaStatus);
153.         goto Error;
154.     }*/
155.  
156.     // Copy output vector from GPU buffer to host memory.
157.     cudaStatus = cudaMemcpy(is_prime, dev_c, sizeof(int), cudaMemcpyDeviceToHost);
158.     if (cudaStatus != cudaSuccess) {
159.         fprintf(stderr, "cudaMemcpy failed!");
160.         goto Error;
161.     }
162.  
163.         if (*is_prime == 1) {
164.                 cout << "(GPU)Liczba nie jest pierwsza n";
165.  
166.         }
167.         else {
168.                 cout << "(GPU)liczba pierwsza n";
169.  
170.         }
171.         printf("Czas wykonywania: %lu msn", clock() - start);
172.  
173.  
174. Error:
175.     cudaFree(dev_c);
176.     cudaFree(dev_a);
177.     cudaFree(dev_b);
178.    
179.     return cudaStatus;
180. }
181.