Problem 206

Concealed Square: find n where n^2 = 1_2_3_4_5_6_7_8_9_0.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 206
# Concealed Square: find n where n^2 = 1_2_3_4_5_6_7_8_9_0.

function form_ok(x: i64) -> bool {
    let mut square: i64 = x * x
    if square % 10 != 0 { return false }
    square = square / 100
    let mut want: i64 = 9
    while want >= 1 {
        if square % 10 != want { return false }
        square = square / 100
        want = want - 1
    }
    return true
}

function main() -> i32 {
    let mut x: i64 = 1389026620
    while x >= 1010101010 {
        if form_ok(x) {
            printf("%lld\n", x)
            return 0
        }
        x = x - 10
    }
    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; }

bool form_ok_i64(int64_t x);
int32_t main(void);

bool form_ok_i64(int64_t x) {
    int64_t square = (x * x);
    if (FLOW_CHECKED_MOD((square), (10)) != 0) {
        return 0;
    }
    square = FLOW_CHECKED_DIV((square), (100));
    int64_t want = 9;
    while (want >= 1) {
        if (FLOW_CHECKED_MOD((square), (10)) != want) {
            return 0;
        }
        square = FLOW_CHECKED_DIV((square), (100));
        want = (want - 1);
    }
    return 1;
}

int32_t main(void) {
    int64_t x = 1389026620;
    while (x >= 1010101010) {
        if (form_ok_i64(x)) {
            printf("%lld\n", x);
            return 0;
        }
        x = (x - 10);
    }
    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 @form_ok(%arg0: i64) -> i1 {
    %0 = arith.muli %arg0, %arg0 : i64
    %1 = llvm.mlir.constant(1 : i64) : i64
    %2 = llvm.alloca %1 x i64 : (i64) -> !llvm.ptr
    llvm.store %0, %2 : i64, !llvm.ptr
    %3 = llvm.load %2 : !llvm.ptr -> i64
    %4 = arith.constant 10 : i32
    %6 = arith.extsi %4 : i32 to i64
    %5 = arith.remsi %3, %6 : i64
    %7 = arith.constant 0 : i32
    %9 = arith.extsi %7 : i32 to i64
    %8 = arith.cmpi ne, %5, %9 : i64
    cf.cond_br %8, ^bb0, ^bb1
    ^bb0:
      %10 = arith.constant 0 : i1
      func.return %10 : i1
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %11 = llvm.load %2 : !llvm.ptr -> i64
    %12 = arith.constant 100 : i32
    %14 = arith.extsi %12 : i32 to i64
    %13 = arith.divsi %11, %14 : i64
    llvm.store %13, %2 : i64, !llvm.ptr
    %15 = arith.constant 9 : i32
    %16 = arith.extsi %15 : i32 to i64
    %17 = llvm.mlir.constant(1 : i64) : i64
    %18 = llvm.alloca %17 x i64 : (i64) -> !llvm.ptr
    llvm.store %16, %18 : i64, !llvm.ptr
    cf.br ^bb3
    ^bb3:
    %19 = llvm.load %18 : !llvm.ptr -> i64
    %20 = arith.constant 1 : i32
    %22 = arith.extsi %20 : i32 to i64
    %21 = arith.cmpi sge, %19, %22 : i64
    cf.cond_br %21, ^bb4, ^bb5
    ^bb4:
      %23 = llvm.load %2 : !llvm.ptr -> i64
      %24 = arith.constant 10 : i32
      %26 = arith.extsi %24 : i32 to i64
      %25 = arith.remsi %23, %26 : i64
      %27 = llvm.load %18 : !llvm.ptr -> i64
      %28 = arith.cmpi ne, %25, %27 : i64
      cf.cond_br %28, ^bb6, ^bb7
      ^bb6:
        %29 = arith.constant 0 : i1
        func.return %29 : i1
      ^bb7:
        cf.br ^bb8
      ^bb8:
      %30 = llvm.load %2 : !llvm.ptr -> i64
      %31 = arith.constant 100 : i32
      %33 = arith.extsi %31 : i32 to i64
      %32 = arith.divsi %30, %33 : i64
      llvm.store %32, %2 : i64, !llvm.ptr
      %34 = llvm.load %18 : !llvm.ptr -> i64
      %35 = arith.constant 1 : i32
      %37 = arith.extsi %35 : i32 to i64
      %36 = arith.subi %34, %37 : i64
      llvm.store %36, %18 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %38 = arith.constant 1 : i1
    func.return %38 : i1
  }
  func.func @main() -> i32 {
    %39 = arith.constant 1389026620 : i32
    %40 = arith.extsi %39 : i32 to i64
    %41 = llvm.mlir.constant(1 : i64) : i64
    %42 = llvm.alloca %41 x i64 : (i64) -> !llvm.ptr
    llvm.store %40, %42 : i64, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %43 = llvm.load %42 : !llvm.ptr -> i64
    %44 = arith.constant 1010101010 : i32
    %46 = arith.extsi %44 : i32 to i64
    %45 = arith.cmpi sge, %43, %46 : i64
    cf.cond_br %45, ^bb10, ^bb11
    ^bb10:
      %48 = llvm.load %42 : !llvm.ptr -> i64
      %47 = func.call @form_ok(%48) : (i64) -> i1
      cf.cond_br %47, ^bb12, ^bb13
      ^bb12:
        %49 = llvm.mlir.addressof @str_0 : !llvm.ptr
        %50 = llvm.load %42 : !llvm.ptr -> i64
        %51 = llvm.call @printf(%49, %50) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
        %52 = arith.constant 0 : i32
        func.return %52 : i32
      ^bb13:
        cf.br ^bb14
      ^bb14:
      %53 = llvm.load %42 : !llvm.ptr -> i64
      %54 = arith.constant 10 : i32
      %56 = arith.extsi %54 : i32 to i64
      %55 = arith.subi %53, %56 : i64
      llvm.store %55, %42 : i64, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    %57 = arith.constant 1 : i32
    func.return %57 : i32
  }
}