1 /*
2 * Copyright (c) 2019 Synopsys, Inc.
3 *
4 * SPDX-License-Identifier: Apache-2.0
5 */
6
7 #include <zephyr/kernel.h>
8 #include <stdio.h>
9
10 /* Amount of execution threads to create and run */
11 #define THREADS_NUM 16
12
13 /*
14 * Amount of digits of Pi to calculate, must be a multiple of 4,
15 * as used algorithm spits 4 digits on every iteration.
16 */
17 #define DIGITS_NUM 240
18
19 #define LENGTH ((DIGITS_NUM / 4) * 14)
20 #define STACK_SIZE ((LENGTH * sizeof(int) + 1280))
21
22 #ifdef CONFIG_SMP
23 #define CORES_NUM arch_num_cpus()
24 #else
25 #define CORES_NUM 1
26 #endif
27
28 static K_THREAD_STACK_ARRAY_DEFINE(tstack, THREADS_NUM, STACK_SIZE);
29 static struct k_thread tthread[THREADS_NUM];
30 static char th_buffer[THREADS_NUM][DIGITS_NUM + 1];
31 static atomic_t th_counter = THREADS_NUM;
32
33 struct k_sem main_sem;
34
test_thread(void * arg1,void * arg2,void * arg3)35 void test_thread(void *arg1, void *arg2, void *arg3)
36 {
37 atomic_t *counter = (atomic_t *)arg1;
38 char *buffer = (char *)arg2;
39
40 ARG_UNUSED(arg3);
41
42 /*
43 * Adapted and improved (for random number of digits) version of Pi
44 * calculation program initially proposed by Dik T. Winter as:
45 * -------------------------------->8--------------------------------
46 * int a=10000,b,c=2800,d,e,f[2801],g;main(){for(;b-c;)f[b++]=a/5;
47 * for(;d=0,g=c*2;c-=14,printf("%.4d",e+d/a),e=d%a)for(b=c;d+=f[b]*a,
48 * f[b]=d%--g,d/=g--,--b;d*=b);}
49 * -------------------------------->8--------------------------------
50 */
51 #define NEW_BASE 10000
52 #define ARRAY_INIT 2000
53
54 int array[LENGTH + 1] = {};
55 int carry = 0;
56 int i, j;
57
58 for (i = 0; i < LENGTH; i++) {
59 array[i] = ARRAY_INIT;
60 }
61
62 for (i = LENGTH; i > 0; i -= 14) {
63 int sum = 0, value;
64
65 for (j = i; j > 0; --j) {
66 sum = sum * j + NEW_BASE * array[j];
67 array[j] = sum % (j * 2 - 1);
68 sum /= j * 2 - 1;
69 }
70
71 value = carry + sum / NEW_BASE;
72 carry = sum % NEW_BASE;
73
74 /* Convert 4-digit int to string */
75 sprintf(buffer, "%.4d", value);
76 buffer += 4;
77 }
78
79 if (atomic_dec(counter) == 1) {
80 k_sem_give(&main_sem);
81 }
82 }
83
main(void)84 int main(void)
85 {
86 uint32_t start_time, stop_time, cycles_spent, nanoseconds_spent;
87 int i;
88
89 if (k_sem_init(&main_sem, 0, 1)) {
90 printk("Failed initialization!\n");
91 return -1;
92 }
93
94 printk("Calculate first %d digits of Pi independently by %d threads.\n",
95 DIGITS_NUM, THREADS_NUM);
96
97 /* Capture initial time stamp */
98 start_time = k_cycle_get_32();
99
100 for (i = 0; i < THREADS_NUM; i++) {
101 k_thread_create(&tthread[i], tstack[i], STACK_SIZE,
102 test_thread,
103 (void *)&th_counter, (void *)th_buffer[i], NULL,
104 K_PRIO_COOP(10), 0, K_NO_WAIT);
105 }
106
107 /* Wait for all workers to finish their calculations */
108 k_sem_take(&main_sem, K_FOREVER);
109
110 /* Capture final time stamp */
111 stop_time = k_cycle_get_32();
112
113 cycles_spent = stop_time - start_time;
114 nanoseconds_spent = (uint32_t)k_cyc_to_ns_floor64(cycles_spent);
115
116 for (i = 0; i < THREADS_NUM; i++) {
117 printk("Pi value calculated by thread #%d: %s\n", i, th_buffer[i]);
118 }
119
120 printk("All %d threads executed by %d cores in %d msec\n", THREADS_NUM,
121 CORES_NUM, nanoseconds_spent / 1000 / 1000);
122 return 0;
123 }
124
