Why the latest mimalloc performance is seriously reduced.

I tried based on the following test case (1.c), the latest master version mimalloc, malloc/free performance degradation ; multiple times malloc/free, using mimalloc took 16.5s, surprisingly much slower than the 6.5s without mimalloc.

root@virtual-machine:/media/shixingya/mimalloc/build# vi 1.c
root@virtual-machine:/media/shixingya/mimalloc/build# gcc 1.c -w && ./a.out
6.547415
root@virtual-machine:/media/shixingya/mimalloc/build# env LD_PRELOAD=./libmimalloc.so ./a.out
16.503070

#include <sys/time.h>
int main() {
struct timeval last;
struct timeval next;
gettimeofday(&last, 0);
for (int i = 0; i < 10000000; ++i) {
int *p = malloc(i * sizeof(int));
free(p);
}
gettimeofday(&next, 0);
printf("%llu.%06llu\n",
(next.tv_usec > last.tv_usec ? next.tv_sec - last.tv_sec : next.tv_sec - 1 - last.tv_sec),
(next.tv_usec > last.tv_usec ? next.tv_usec - last.tv_usec : 1000000 + next.tv_usec - last.tv_usec));
return 0;
}

Reproduce the timings from the embedded 1.c case using gcc and LD_PRELOAD=./libmimalloc.so, then compare the latest master behavior with and without mimalloc. Trace the allocation path involved in repeated malloc/free calls and identify why the mimalloc run is slower; done means the regression is explained and a focused test or benchmark captures the result.