c1e168fdbe
+ Microarchitecture optimizations + 64-bit support + Xilinx FPGA support + LLVM-16 support + Refactoring and quality control fixes minor update minor update minor update minor update minor update minor update cleanup cleanup cache bindings and memory perf refactory minor update minor update hw unit tests fixes minor update minor update minor update minor update minor update minor udpate minor update minor update minor update minor update minor update minor update minor update minor updates minor updates minor update minor update minor update minor update minor update minor update minor updates minor updates minor updates minor updates minor update minor update
26 lines
678 B
C++
26 lines
678 B
C++
#include <stdint.h>
|
|
#include <vx_intrinsics.h>
|
|
#include <vx_spawn.h>
|
|
#include "common.h"
|
|
|
|
void kernel_body(int task_id, kernel_arg_t* __UNIFORM__ arg) {
|
|
uint32_t num_points = arg->num_points;
|
|
TYPE* src_ptr = (TYPE*)arg->src_addr;
|
|
TYPE* dst_ptr = (TYPE*)arg->dst_addr;
|
|
|
|
TYPE ref_value = src_ptr[task_id];
|
|
|
|
uint32_t pos = 0;
|
|
for (uint32_t i = 0; i < num_points; ++i) {
|
|
TYPE cur_value = src_ptr[i];
|
|
pos += (cur_value < ref_value) || ((cur_value == ref_value) && (i < task_id));
|
|
}
|
|
dst_ptr[pos] = ref_value;
|
|
}
|
|
|
|
int main() {
|
|
kernel_arg_t* arg = (kernel_arg_t*)KERNEL_ARG_DEV_MEM_ADDR;
|
|
vx_spawn_tasks(arg->num_points, (vx_spawn_tasks_cb)kernel_body, arg);
|
|
return 0;
|
|
}
|