Problem 047

First of four consecutive integers each with four distinct prime factors.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 047
# First of four consecutive integers each with four distinct prime factors.

extern {
    function calloc(n: i64, size: i64) -> ptr<void>
    function free(p: ptr<void>) -> void
}

function main() -> i32 {
    let limit: i64 = 200000
    let factors: ptr<i32> = calloc(limit, 4)
    if factors == null { return 1 }

    let mut i: i64 = 2
    while i < limit {
        if factors[i] == 0 {
            # i is prime
            let mut m: i64 = i
            while m < limit {
                factors[m] = factors[m] + 1
                m = m + i
            }
        }
        i = i + 1
    }

    let mut n: i64 = 2
    while n + 3 < limit {
        if factors[n] == 4 && factors[n + 1] == 4 && factors[n + 2] == 4 && factors[n + 3] == 4 {
            printf("%lld\n", n)
            free(factors)
            return 0
        }
        n = n + 1
    }
    free(factors)
    return 1
}

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; }

int32_t main(void);



int32_t main(void) {
    int64_t limit = 200000;
    int32_t* factors = (int32_t*)(calloc(limit, 4));
    if (factors == NULL) {
        return 1;
    }
    int64_t i = 2;
    while (i < limit) {
        if (factors[i] == 0) {
            int64_t m = i;
            while (m < limit) {
                factors[m] = (factors[m] + 1);
                m = (m + i);
            }
        }
        i = (i + 1);
    }
    int64_t n = 2;
    while ((n + 3) < limit) {
        if ((((factors[n] == 4 && factors[(n + 1)] == 4) && factors[(n + 2)] == 4) && factors[(n + 3)] == 4)) {
            printf("%lld\n", n);
            free(factors);
            return 0;
        }
        n = (n + 1);
    }
    free(factors);
    return 1;
}

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 private @calloc(i64, i64) -> !llvm.ptr
  func.func private @free(!llvm.ptr) -> ()
  func.func @main() -> i32 {
    %0 = arith.constant 200000 : i32
    %1 = arith.extsi %0 : i32 to i64
    %3 = arith.constant 4 : i32
    %4 = arith.extsi %3 : i32 to i64
    %2 = func.call @calloc(%1, %4) : (i64, i64) -> !llvm.ptr
    %5 = llvm.mlir.zero : !llvm.ptr
    %6 = llvm.icmp "eq" %2, %5 : !llvm.ptr
    cf.cond_br %6, ^bb0, ^bb1
    ^bb0:
      %7 = arith.constant 1 : i32
      func.return %7 : i32
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %8 = arith.constant 2 : i32
    %9 = arith.extsi %8 : i32 to i64
    %10 = llvm.mlir.constant(1 : i64) : i64
    %11 = llvm.alloca %10 x i64 : (i64) -> !llvm.ptr
    llvm.store %9, %11 : i64, !llvm.ptr
    cf.br ^bb3
    ^bb3:
    %12 = llvm.load %11 : !llvm.ptr -> i64
    %13 = arith.cmpi slt, %12, %1 : i64
    cf.cond_br %13, ^bb4, ^bb5
    ^bb4:
      %15 = llvm.load %11 : !llvm.ptr -> i64
      %16 = llvm.getelementptr %2[%15] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %14 = llvm.load %16 : !llvm.ptr -> i32
      %17 = arith.constant 0 : i32
      %18 = arith.cmpi eq, %14, %17 : i32
      cf.cond_br %18, ^bb6, ^bb7
      ^bb6:
        %19 = llvm.load %11 : !llvm.ptr -> i64
        %20 = llvm.mlir.constant(1 : i64) : i64
        %21 = llvm.alloca %20 x i64 : (i64) -> !llvm.ptr
        llvm.store %19, %21 : i64, !llvm.ptr
        cf.br ^bb9
        ^bb9:
        %22 = llvm.load %21 : !llvm.ptr -> i64
        %23 = arith.cmpi slt, %22, %1 : i64
        cf.cond_br %23, ^bb10, ^bb11
        ^bb10:
          %25 = llvm.load %21 : !llvm.ptr -> i64
          %26 = llvm.getelementptr %2[%25] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          %24 = llvm.load %26 : !llvm.ptr -> i32
          %27 = arith.constant 1 : i32
          %28 = arith.addi %24, %27 : i32
          %29 = llvm.load %21 : !llvm.ptr -> i64
          %30 = llvm.getelementptr %2[%29] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          llvm.store %28, %30 : i32, !llvm.ptr
          %31 = llvm.load %21 : !llvm.ptr -> i64
          %32 = llvm.load %11 : !llvm.ptr -> i64
          %33 = arith.addi %31, %32 : i64
          llvm.store %33, %21 : i64, !llvm.ptr
          cf.br ^bb9
        ^bb11:
        cf.br ^bb8
      ^bb7:
        cf.br ^bb8
      ^bb8:
      %34 = llvm.load %11 : !llvm.ptr -> i64
      %35 = arith.constant 1 : i32
      %37 = arith.extsi %35 : i32 to i64
      %36 = arith.addi %34, %37 : i64
      llvm.store %36, %11 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %38 = arith.constant 2 : 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 ^bb12
    ^bb12:
    %42 = llvm.load %41 : !llvm.ptr -> i64
    %43 = arith.constant 3 : i32
    %45 = arith.extsi %43 : i32 to i64
    %44 = arith.addi %42, %45 : i64
    %46 = arith.cmpi slt, %44, %1 : i64
    cf.cond_br %46, ^bb13, ^bb14
    ^bb13:
      %48 = llvm.load %41 : !llvm.ptr -> i64
      %49 = llvm.getelementptr %2[%48] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %47 = llvm.load %49 : !llvm.ptr -> i32
      %50 = arith.constant 4 : i32
      %51 = arith.cmpi eq, %47, %50 : i32
      %52 = scf.if %51 -> (i1) {
        %54 = llvm.load %41 : !llvm.ptr -> i64
        %55 = arith.constant 1 : i32
        %57 = arith.extsi %55 : i32 to i64
        %56 = arith.addi %54, %57 : i64
        %58 = llvm.getelementptr %2[%56] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %53 = llvm.load %58 : !llvm.ptr -> i32
        %59 = arith.constant 4 : i32
        %60 = arith.cmpi eq, %53, %59 : i32
        scf.yield %60 : i1
      } else {
        %61 = arith.constant false
        scf.yield %61 : i1
      }
      %62 = scf.if %52 -> (i1) {
        %64 = llvm.load %41 : !llvm.ptr -> i64
        %65 = arith.constant 2 : i32
        %67 = arith.extsi %65 : i32 to i64
        %66 = arith.addi %64, %67 : i64
        %68 = llvm.getelementptr %2[%66] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %63 = llvm.load %68 : !llvm.ptr -> i32
        %69 = arith.constant 4 : i32
        %70 = arith.cmpi eq, %63, %69 : i32
        scf.yield %70 : i1
      } else {
        %71 = arith.constant false
        scf.yield %71 : i1
      }
      %72 = scf.if %62 -> (i1) {
        %74 = llvm.load %41 : !llvm.ptr -> i64
        %75 = arith.constant 3 : i32
        %77 = arith.extsi %75 : i32 to i64
        %76 = arith.addi %74, %77 : i64
        %78 = llvm.getelementptr %2[%76] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %73 = llvm.load %78 : !llvm.ptr -> i32
        %79 = arith.constant 4 : i32
        %80 = arith.cmpi eq, %73, %79 : i32
        scf.yield %80 : i1
      } else {
        %81 = arith.constant false
        scf.yield %81 : i1
      }
      cf.cond_br %72, ^bb15, ^bb16
      ^bb15:
        %82 = llvm.mlir.addressof @str_0 : !llvm.ptr
        %83 = llvm.load %41 : !llvm.ptr -> i64
        %84 = llvm.call @printf(%82, %83) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
        func.call @free(%2) : (!llvm.ptr) -> ()
        %86 = arith.constant 0 : i32
        func.return %86 : i32
      ^bb16:
        cf.br ^bb17
      ^bb17:
      %87 = llvm.load %41 : !llvm.ptr -> i64
      %88 = arith.constant 1 : i32
      %90 = arith.extsi %88 : i32 to i64
      %89 = arith.addi %87, %90 : i64
      llvm.store %89, %41 : i64, !llvm.ptr
      cf.br ^bb12
    ^bb14:
    func.call @free(%2) : (!llvm.ptr) -> ()
    %92 = arith.constant 1 : i32
    func.return %92 : i32
  }
}