Given the content of dataA and dataB: DataA: 1 4 9 8 2 5 11 5 7 1 2 2 1871975 DataB: 7 0 4 8 4 5 7 1 2 6 4 3 2 656 8294 And the following kernel: kernel void kernelA(_global int* dataA, global int2* dataB, global int4* dataC) { } For the work-item with global ID=2, what will the content of the following be? V • v1 • v2 int gid= get_global_id(0); int4 v vload4 (gid, dataA + 3); int4 v1 = (int4) (dataB[gid + 1], dataB[gid]); int4 v2 (int4) (v.s22, v1.s30); int4 m v1 > V; dataC[gid] select (v1, v2, m); .m • dataC[gid] Given the following information: Kernel name: importantKernel. The kernel accepts a single integer, an input array of char4s and an output array of int8s. There should be 1024 work-items. There should be 4 work-groups. Global IDs are to start from 13. 7/8 Provide OpenCL code for the following: The signature of the kernel. Setting up buffers and kernel arguments. Setting up and enqueueing the kernel. You are free to choose the variable names. The content of arguments passed to the kernel is not important.

Given the content of dataA and dataB: DataA: 1 4 9 8 2 5 11 5 7 1 2 2 1871975 DataB: 7 0 4 8 4 5 7 1 2 6 4 3 2 656 8294 And the following kernel: kernel void kernelA(_global int* dataA, global int2* dataB, global int4* dataC) { } For the work-item with global ID=2, what will the content of the following be? V • v1 • v2 int gid= get_global_id(0); int4 v vload4 (gid, dataA + 3); int4 v1 = (int4) (dataB[gid + 1], dataB[gid]); int4 v2 (int4) (v.s22, v1.s30); int4 m v1 > V; dataC[gid] select (v1, v2, m); .m • dataC[gid] Given the following information: Kernel name: importantKernel. The kernel accepts a single integer, an input array of char4s and an output array of int8s. There should be 1024 work-items. There should be 4 work-groups. Global IDs are to start from 13. 7/8 Provide OpenCL code for the following: The signature of the kernel. Setting up buffers and kernel arguments. Setting up and enqueueing the kernel. You are free to choose the variable names. The content of arguments passed to the kernel is not important.

Microsoft Visual C#

7th Edition

ISBN:9781337102100

Author:Joyce, Farrell.

Publisher:Joyce, Farrell.

Chapter8: Advanced Method Concepts

Section: Chapter Questions

Problem 7RQ

See similar textbooks

Similar questions

On a 32-bit processor, a baggy bounds scheme is expected to set all of the bounds table entries to 31 at initialization time. Assume that a buggy implementation with a slot size of 32 bytes performs bounds table initialization inappropriately, resulting in random entries being incorrectly set to 1. Assume that a networked server processes network messages using an uninstrumented library. Assume that this library does not include some buffer overflow bugs (i.e., it never calls unsafe functions such as gets()). However, the server suffers from the above-mentioned bounds table initialization issue, and an attacker may submit messages to the server that force the library to dynamically assign and write memory in an attacker-controlled amount using uninstrumented code that looks like this: / N is the size of the buffer that the intruder gets to choose.for (int I = 0; I N/4; i++, p += 4) char *p = malloc(N); *p = 'a'; *(p+1) = 'b'; *(p+2) = 'c'; *(p+3) = 'd'; Assume the server is using a…
You are tasked with the development of an E-Scooter ride-sharesystem. It allows registered commuters to approach an idle E-Scooter andreserve it, following which they use the E-Scooter to commute a certaindistance (that is not known prior to use). Finally, after the commuter reachestheir destination, they end the ride, which prompts an automaticcomputation of the ride fees. which is automatically debited using thecommuters registered payment details.
A picture is filtered by a 3*3 kernel of's, and then the result is filtered again. What size kernel can execute this function in one pass?
Which of the following kernels are developmental kernels? (Choose all that apply.) a. 2.3.4 b. 3.5.5 c. 4.1-rc5 d. 4.4.4
Analyze the code solution below and discuss any gain in performance. To gather measurement values, use any CUDA or OpenMP library API that is used to measure performance (time, speedup, etc..) or by using what is offered by the compiler (C, C++). #include <stdlib.h> //malloc and free #include <stdio.h> //printf #include <omp.h> //OpenMP // Very small values for this simple illustrative example #define ARRAY_SIZE 8 //Size of arrays whose elements will be added together. #define NUM_THREADS 4 //Number of threads to use for vector addition. /* int main (int argc, char *argv[]) { // elements of arrays a and b will be added // and placed in array c int * a; int * b; int * c; intn = ARRAY_SIZE; // number of array elements intn_per_thread; // elements per thread inttotal_threads = NUM_THREADS; // number of threads to use inti; // loop index // allocate space for the arrays a = (int *) malloc(sizeof(int)*n); b = (int *) malloc(sizeof(int)*n); c = (int *)…
Suppose that we have Parent.c as follows. It is compiled and linked to generate Parent executable. L1: #include <sys/wait.h> L2: #define NULL 0 L3: int main (void) L4: { L5: if (fork() == 0){ L6: //execve("Child",NULL,NULL); L7: printf("I am child process with PID: %d\n", getpid()); L8: // exit(0); L9: } else { L10: printf("Process[%d]: in execution ...\n", getppid()); L11: sleep(5); L12: if(wait(NULL) > 0) L13: printf("Process[%d]: \n", getpid()); L14: printf("Process[%d]: ...\n", getppid()); L15: } L16: printf("HHHH:%d\n", getpid()); L17: } L1 denotes Line 1, L2 denotes L2, etc. If we run the Parent program: a: How many times will L16 execute? That is, how many lines of “HHHH” will be printed? Please briefly explain. b: What information will Line 13 print out? c: Which…
A ring buffer, often known as a circular queue, is a fixed-size N FIFO data structure. It can be used to transport data between asynchronous processes or to store log files. When the buffer is empty, the consumer must wait for data to be deposited; when the buffer is full, the producer must wait for data to be deposited. Create a RingBuffer API as well as an implementation that uses an array representation (with circular wrap-around).
/* GCC *//* gcc -Wa,-adhln -g -masm=intel -m32 "Project 2.c" > "Project 2-g.asm" *//* gcc -Wa,-adhln -O -masm=intel -m32 "Project 2.c" > "Project 2-o.asm" */ #include <stdio.h> #define NOINLINE __attribute__ ((noinline)) static NOINLINE int function1(int x, int y){ int i;int sum;int values[10];sum = 0;for (i = 0; i < 10; i++) {values[i] = 10 * i + x * y;sum += values[i];} return (sum);} static int NOINLINE function2(int *values, int valuesLen){ int i;int sum; sum = 0;for (i = 0; i < valuesLen; i++) {sum += values[i];} return (sum);} static NOINLINE int function3(int x){ int y; y = x / 10;return (y);} static NOINLINE int function4(int a, int b, int c, int d){ int r; if (a > b)r = a;else if (a > c)r = 2 * a;else if (a > d)r = 3 * a;elser = -1;return (r);} int main(int argc, char **argv){ int i;int j;int k;int values[10]; i = 1;j = 2;k = function1(i, j);printf("function1: i = %d, j = %d, k = %d\n", i, j, k); for (i = 0; i < 10; i++) {values[i] = i;}k =…
/* GCC *//* gcc -Wa,-adhln -g -masm=intel -m32 "Project 2.c" > "Project 2-g.asm" *//* gcc -Wa,-adhln -O -masm=intel -m32 "Project 2.c" > "Project 2-o.asm" */ #include <stdio.h> #define NOINLINE __attribute__ ((noinline)) static NOINLINE int function1(int x, int y){ int i;int sum;int values[10];sum = 0;for (i = 0; i < 10; i++) {values[i] = 10 * i + x * y;sum += values[i];} return (sum);} static int NOINLINE function2(int *values, int valuesLen){ int i;int sum; sum = 0;for (i = 0; i < valuesLen; i++) {sum += values[i];} return (sum);} static NOINLINE int function3(int x){ int y; y = x / 10;return (y);} static NOINLINE int function4(int a, int b, int c, int d){ int r; if (a > b)r = a;else if (a > c)r = 2 * a;else if (a > d)r = 3 * a;elser = -1;return (r);} int main(int argc, char **argv){ int i;int j;int k;int values[10]; i = 1;j = 2;k = function1(i, j);printf("function1: i = %d, j = %d, k = %d\n", i, j, k); for (i = 0; i < 10; i++) {values[i] = i;}k =…
As a programmer, you should attempt to avoid making your case for deep access's superiority versus shallow access based only on performance.
Object code files are generated by compilers and assemblers from source code. The object files need to be linked together before they can be executed. Compilers could, in principle, produce executables without linking. Can you explain the problems with the second method?
Please provide correct solution. Task1: Solve the producer and consumer problem with inter thread communication (join(), wait(), sleep() etc.) modifying the given C code. #include <pthread.h> #include <stdio.h> #include <string.h> #define MAX 10 //producers and consumers can produce and consume upto MAX #define BUFLEN 6 #define NUMTHREAD 2 /* number of threads */ void * consumer(int *id); void * producer(int *id); char buffer[BUFLEN]; char source[BUFLEN]; //from this array producer will store it's production into buffer int pCount = 0; int cCount = 0; int buflen; //initializing pthread mutex and condition variables pthread_mutex_t count_mutex = PTHREAD_MUTEX_INITIALIZER; pthread_cond_t nonEmpty = PTHREAD_COND_INITIALIZER; pthread_cond_t full = PTHREAD_COND_INITIALIZER; int thread_id[NUMTHREAD] = {0,1}; int i = 0; int j = 0; main() { int i; /* define the type to be pthread */ pthread_t thread[NUMTHREAD];…