Problem 012

First triangle number with over 500 divisors.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 012
# First triangle number with over 500 divisors.

function divisor_count(n: i64) -> i64 {
    let mut count: i64 = 0
    let mut i: i64 = 1
    while i * i <= n {
        if n % i == 0 {
            if i * i == n {
                count = count + 1
            } else {
                count = count + 2
            }
        }
        i = i + 1
    }
    return count
}

function solve(min_divisors: i64) -> i64 {
    let mut n: i64 = 1
    let mut tri: i64 = 0
    while true {
        tri = tri + n
        if divisor_count(tri) > min_divisors {
            return tri
        }
        n = n + 1
    }
    return 0
}

function main() -> i32 {
    printf("%lld\n", solve(500))
    return 0
}

Generated C

#include <stdint.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

/* Flow runtime helpers */
typedef struct flow_temp_node { struct flow_temp_node* next; } flow_temp_node;
static flow_temp_node* flow_temp_head = NULL;
static int flow_temp_atexit_set = 0;
__attribute__((unused)) static void flow_temp_free_all(void) {
    while (flow_temp_head) {
        flow_temp_node* n = flow_temp_head;
        flow_temp_head = n->next;
        free(n);
    }
}
__attribute__((unused)) static void* flow_temp_alloc(size_t nbytes) {
    flow_temp_node* node = (flow_temp_node*)malloc(sizeof(flow_temp_node) + nbytes);
    if (!node) return NULL;
    node->next = flow_temp_head;
    flow_temp_head = node;
    if (!flow_temp_atexit_set) {
        flow_temp_atexit_set = 1;
        atexit(flow_temp_free_all);
    }
    return (void*)(node + 1);
}
#ifndef FLOW_DIAG
#define FLOW_DIAG(msg) fprintf(stderr, "%s", (msg))
#endif
#ifndef FLOW_LOG
#define FLOW_LOG(fmt, ...) printf(fmt, __VA_ARGS__)
#endif
#ifndef FLOW_LOG_EMPTY
#define FLOW_LOG_EMPTY(fmt) printf(fmt)
#endif
static char* flow_strcat(const char* a, const char* b) {
    size_t la = strlen(a ? a : ""), lb = strlen(b ? b : "");
    char* r = (char*)flow_temp_alloc(la + lb + 1);
    if (!r) return NULL;
    if (la) memcpy(r, a, la);
    if (lb) memcpy(r + la, b, lb);
    r[la + lb] = '\0';
    return r;
}

#define __flow_in_arr(arr, val) __extension__ ({ \
    int _found = 0; \
    size_t _n = sizeof(arr)/sizeof((arr)[0]); \
    for (size_t _i = 0; _i < _n; _i++) { \
        if ((arr)[_i] == (val)) { _found = 1; break; } \
    } _found; })

/* Unified fault handler (MISRA #279) — override with -DFLOW_FAULT_HANDLER=fn */
#ifndef FLOW_FAULT_HANDLER
__attribute__((unused)) static inline void flow_fault_handler(const char* msg) {
    fprintf(stderr, "flow: %s\n", msg ? msg : "fault");
    abort();
#if defined(__GNUC__) || defined(__clang__)
    __builtin_unreachable();
#endif
}
#else
#define flow_fault_handler FLOW_FAULT_HANDLER
#endif
#define flow_div_by_zero_handler() flow_fault_handler("division by zero")
#define flow_shift_ub_handler() flow_fault_handler("invalid shift (amount out of range or left-shift of negative)")

#ifndef FLOW_CHECKED_DIV
#define FLOW_CHECKED_DIV(L, R) (((R) != 0) ? ((L) / (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_MOD
#define FLOW_CHECKED_MOD(L, R) (((R) != 0) ? ((L) % (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHL
#define FLOW_CHECKED_SHL(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull)) && ((L) >= 0)) ? ((L) << (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHR
#define FLOW_CHECKED_SHR(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull))) ? ((L) >> (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif

#include <math.h>

void* _ui_state = NULL;

static inline float i32_to_f32(int32_t v) { return (float)v; }

/* Host stub for @gpu kernels (device codegen replaces this). */
static inline int32_t gpu_thread_id(void) { return 0; }

int64_t divisor_count_i64(int64_t n);
int64_t solve_i64(int64_t min_divisors);
int32_t main(void);

int64_t divisor_count_i64(int64_t n) {
    int64_t count = 0;
    int64_t i = 1;
    while ((i * i) <= n) {
        if (FLOW_CHECKED_MOD((n), (i)) == 0) {
            if ((i * i) == n) {
                count = (count + 1);
            } else {
                count = (count + 2);
            }
        }
        i = (i + 1);
    }
    return count;
}

int64_t solve_i64(int64_t min_divisors) {
    int64_t n = 1;
    int64_t tri = 0;
    while (1) {
        tri = (tri + n);
        if (divisor_count_i64(tri) > min_divisors) {
            return tri;
        }
        n = (n + 1);
    }
    return 0;
}

int32_t main(void) {
    printf("%lld\n", solve_i64(500));
    return 0;
}

Generated MLIR

module {
  llvm.func @printf(!llvm.ptr, ...) -> i32
  llvm.mlir.global internal constant @str_0("%lld\n\00") {addr_space = 0 : i32} : !llvm.array<6 x i8>
  func.func @divisor_count(%arg0: i64) -> i64 {
    %0 = arith.constant 0 : i32
    %1 = arith.extsi %0 : i32 to i64
    %2 = llvm.mlir.constant(1 : i64) : i64
    %3 = llvm.alloca %2 x i64 : (i64) -> !llvm.ptr
    llvm.store %1, %3 : i64, !llvm.ptr
    %4 = arith.constant 1 : i32
    %5 = arith.extsi %4 : i32 to i64
    %6 = llvm.mlir.constant(1 : i64) : i64
    %7 = llvm.alloca %6 x i64 : (i64) -> !llvm.ptr
    llvm.store %5, %7 : i64, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %8 = llvm.load %7 : !llvm.ptr -> i64
    %9 = llvm.load %7 : !llvm.ptr -> i64
    %10 = arith.muli %8, %9 : i64
    %11 = arith.cmpi sle, %10, %arg0 : i64
    cf.cond_br %11, ^bb1, ^bb2
    ^bb1:
      %12 = llvm.load %7 : !llvm.ptr -> i64
      %13 = arith.remsi %arg0, %12 : i64
      %14 = arith.constant 0 : i32
      %16 = arith.extsi %14 : i32 to i64
      %15 = arith.cmpi eq, %13, %16 : i64
      cf.cond_br %15, ^bb3, ^bb4
      ^bb3:
        %17 = llvm.load %7 : !llvm.ptr -> i64
        %18 = llvm.load %7 : !llvm.ptr -> i64
        %19 = arith.muli %17, %18 : i64
        %20 = arith.cmpi eq, %19, %arg0 : i64
        cf.cond_br %20, ^bb6, ^bb7
        ^bb6:
          %21 = llvm.load %3 : !llvm.ptr -> i64
          %22 = arith.constant 1 : i32
          %24 = arith.extsi %22 : i32 to i64
          %23 = arith.addi %21, %24 : i64
          llvm.store %23, %3 : i64, !llvm.ptr
          cf.br ^bb8
        ^bb7:
          %25 = llvm.load %3 : !llvm.ptr -> i64
          %26 = arith.constant 2 : i32
          %28 = arith.extsi %26 : i32 to i64
          %27 = arith.addi %25, %28 : i64
          llvm.store %27, %3 : i64, !llvm.ptr
          cf.br ^bb8
        ^bb8:
        cf.br ^bb5
      ^bb4:
        cf.br ^bb5
      ^bb5:
      %29 = llvm.load %7 : !llvm.ptr -> i64
      %30 = arith.constant 1 : i32
      %32 = arith.extsi %30 : i32 to i64
      %31 = arith.addi %29, %32 : i64
      llvm.store %31, %7 : i64, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %33 = llvm.load %3 : !llvm.ptr -> i64
    func.return %33 : i64
  }
  func.func @solve(%arg0: i64) -> i64 {
    %34 = arith.constant 1 : i32
    %35 = arith.extsi %34 : i32 to i64
    %36 = llvm.mlir.constant(1 : i64) : i64
    %37 = llvm.alloca %36 x i64 : (i64) -> !llvm.ptr
    llvm.store %35, %37 : i64, !llvm.ptr
    %38 = arith.constant 0 : i32
    %39 = arith.extsi %38 : i32 to i64
    %40 = llvm.mlir.constant(1 : i64) : i64
    %41 = llvm.alloca %40 x i64 : (i64) -> !llvm.ptr
    llvm.store %39, %41 : i64, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %42 = arith.constant 1 : i1
    cf.cond_br %42, ^bb10, ^bb11
    ^bb10:
      %43 = llvm.load %41 : !llvm.ptr -> i64
      %44 = llvm.load %37 : !llvm.ptr -> i64
      %45 = arith.addi %43, %44 : i64
      llvm.store %45, %41 : i64, !llvm.ptr
      %47 = llvm.load %41 : !llvm.ptr -> i64
      %46 = func.call @divisor_count(%47) : (i64) -> i64
      %48 = arith.cmpi sgt, %46, %arg0 : i64
      cf.cond_br %48, ^bb12, ^bb13
      ^bb12:
        %49 = llvm.load %41 : !llvm.ptr -> i64
        func.return %49 : i64
      ^bb13:
        cf.br ^bb14
      ^bb14:
      %50 = llvm.load %37 : !llvm.ptr -> i64
      %51 = arith.constant 1 : i32
      %53 = arith.extsi %51 : i32 to i64
      %52 = arith.addi %50, %53 : i64
      llvm.store %52, %37 : i64, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    %54 = arith.constant 0 : i32
    %55 = arith.extsi %54 : i32 to i64
    func.return %55 : i64
  }
  func.func @main() -> i32 {
    %56 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %58 = arith.constant 500 : i32
    %59 = arith.extsi %58 : i32 to i64
    %57 = func.call @solve(%59) : (i64) -> i64
    %60 = llvm.call @printf(%56, %57) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %61 = arith.constant 0 : i32
    func.return %61 : i32
  }
}