Thread_analyzer and cpu_load_get results are mismatch for multiple preemptive threads

[keykur111]

while running tm_preemptive_scheduling_test.c i am testing

.
.
.

272		tm_preemptive_thread_4_counter);
273		}

thread_analyzer_print(true);
int cpu_load=cpu_load_get(true);  // true = reset after reading

275          /* Show the time period total.  */
276		printf("Time Period Total:  %lu\n\n", total - last_total);
.
.
.

to print the cpu_load_get total values i am changing the cpu_load_get.c like below

118	uint32_t res;
119	uint64_t active_us;

	printf("ticks_idle: %llu, total_cycles: %llu", ticks_idle, total);

121	if (IS_ENABLED(CONFIG_CPU_LOAD_USE_COUNTER)) {
122		if (ticks_idle > (uint64_t)UINT32_MAX) {

output:

thread_1*** Booting Zephyr OS build v4.2.0-rc3 ***
**** Thread-Metric Preemptive Scheduling Test **** Relative Time: 30
Thread analyze:
TM_PreEmp_Schedul_4 : STACK: unused 952 usage 72 / 1024 (7 %); CPU: 10 %
                     : Total CPU cycles used: 131348255
TM_PreEmp_Schedul_3 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 24 %
                     : Total CPU cycles used: 302652845
TM_PreEmp_Schedul_2 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 26 %
                     : Total CPU cycles used: 323239611
TM_PreEmp_Schedul_1 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 24 %
                     : Total CPU cycles used: 302903497
TM_PreEmp_Schedul_0 : STACK: unused 888 usage 136 / 1024 (13 %); CPU: 13 %
                     : Total CPU cycles used: 166560415
idle                : STACK: unused 288 usage 32 / 320 (10 %); CPU: 0 %
                     : Total CPU cycles used: 0
main                : STACK: unused 368 usage 656 / 1024 (64 %); CPU: 0 %
                     : Total CPU cycles used: 3835885
ISR0                : STACK: unused 1872 usage 176 / 2048 (8 %)
ticks_idle: 0, total_cycles: 2404707942 CPU Load: 1000‰

 
**** Thread-Metric Preemptive Scheduling Test **** Relative Time: 60
Thread analyze:
TM_PreEmp_Schedul_4 : STACK: unused 952 usage 72 / 1024 (7 %); CPU: 10 %
                     : Total CPU cycles used: 268604211
TM_PreEmp_Schedul_3 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 24 %
                     : Total CPU cycles used: 611737041
TM_PreEmp_Schedul_2 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 25 %
                     : Total CPU cycles used: 632283998
TM_PreEmp_Schedul_1 : STACK: unused 880 usage 144 / 1024 (14 %); CPU: 24 %
                     : Total CPU cycles used: 611768326
TM_PreEmp_Schedul_0 : STACK: unused 888 usage 136 / 1024 (13 %); CPU: 13 %
                     : Total CPU cycles used: 332466878
idle                : STACK: unused 288 usage 32 / 320 (10 %); CPU: 0 %
                     : Total CPU cycles used: 0
main                : STACK: unused 368 usage 656 / 1024 (64 %); CPU: 0 %
                     : Total CPU cycles used: 8809921
ISR0                : STACK: unused 1872 usage 176 / 2048 (8 %)
ticks_idle: 0, total_cycles: 2405039433 CPU Load: 1000‰

<style> </style>

	Iteration 1	Iteration 2 (thread_analyze without reset)
TM_PreEmp_Schedul_4	131348255	268604211
TM_PreEmp_Schedul_3	302652845	611737041
TM_PreEmp_Schedul_2	323239611	632283998
TM_PreEmp_Schedul_1	302903497	611768326
TM_PreEmp_Schedul_0	166560415	332466878
main	3835885	8809921
Total	1230540508	2465670375
previous total	0	1230540508
Iteration total	1230540508	1235129867
	2404707942	2405039433
Difference	1174167434	1169909566

This shows the difference of almost double.
Note: while doing with 1 task alone the difference is very less.
I would like to know where the cycles are getting increased.

setup environment details:

Zephyr version: v4.2.0-rc3
CONFIG_SYS_CLOCK_TICKS_PER_SEC=1000
target_compile_options(app PRIVATE -O2 -g3 -Werror -Wall -Wextra -Wvla -Wno-error=unused-parameter  -Wno-error=stringop-overflow -Wno-error=maybe-uninitialized)