Problem 168

Right-rotations that are multiples; sum last 5 digits of all such numbers with 2..100 digits.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 168
# Right-rotations that are multiples; sum last 5 digits of all such numbers
# with 2..100 digits.

function search(num_digits0: i32, multiplier: i32, last_digit: i32, modulo: i64) -> i64 {
    let mut num_digits: i32 = num_digits0
    let mut shift: i64 = 10
    let mut carry: i32 = 0
    let mut current: i32 = last_digit
    let mut result: i64 = last_digit as i64

    while num_digits > 1 {
        num_digits = num_digits - 1
        let next_val: i32 = multiplier * current + carry
        carry = next_val / 10
        current = next_val % 10
        if shift < modulo {
            result = result + (current as i64) * shift
            shift = shift * 10
        }
    }
    let first: i32 = multiplier * current + carry
    if current == 0 || first != last_digit {
        return 0
    }
    return result
}

function main() -> i32 {
    let modulo: i64 = 100000
    let mut result: i64 = 0
    let mut num_digits: i32 = 2
    while num_digits <= 100 {
        let mut multiplier: i32 = 1
        while multiplier <= 9 {
            let mut last_digit: i32 = 1
            while last_digit <= 9 {
                result = result + search(num_digits, multiplier, last_digit, modulo)
                last_digit = last_digit + 1
            }
            multiplier = multiplier + 1
        }
        num_digits = num_digits + 1
    }
    printf("%lld\n", result % modulo)
    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 search_i32_i32_i32_i64(int32_t num_digits0, int32_t multiplier, int32_t last_digit, int64_t modulo);
int32_t main(void);

int64_t search_i32_i32_i32_i64(int32_t num_digits0, int32_t multiplier, int32_t last_digit, int64_t modulo) {
    int32_t num_digits = num_digits0;
    int64_t shift = 10;
    int32_t carry = 0;
    int32_t current = last_digit;
    int64_t result = ((int64_t)(last_digit));
    while (num_digits > 1) {
        num_digits = (num_digits - 1);
        int32_t next_val = ((multiplier * current) + carry);
        carry = FLOW_CHECKED_DIV((next_val), (10));
        current = FLOW_CHECKED_MOD((next_val), (10));
        if (shift < modulo) {
            result = (result + (((int64_t)(current)) * shift));
            shift = (shift * 10);
        }
    }
    int32_t first = ((multiplier * current) + carry);
    if ((current == 0 || first != last_digit)) {
        return 0;
    }
    return result;
}

int32_t main(void) {
    int64_t modulo = 100000;
    int64_t result = 0;
    int32_t num_digits = 2;
    while (num_digits <= 100) {
        int32_t multiplier = 1;
        while (multiplier <= 9) {
            int32_t last_digit = 1;
            while (last_digit <= 9) {
                result = (result + search_i32_i32_i32_i64(num_digits, multiplier, last_digit, modulo));
                last_digit = (last_digit + 1);
            }
            multiplier = (multiplier + 1);
        }
        num_digits = (num_digits + 1);
    }
    printf("%lld\n", FLOW_CHECKED_MOD((result), (modulo)));
    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 @search(%arg0: i32, %arg1: i32, %arg2: i32, %arg3: i64) -> i64 {
    %0 = llvm.mlir.constant(1 : i64) : i64
    %1 = llvm.alloca %0 x i32 : (i64) -> !llvm.ptr
    llvm.store %arg0, %1 : i32, !llvm.ptr
    %2 = arith.constant 10 : i32
    %3 = arith.extsi %2 : i32 to i64
    %4 = llvm.mlir.constant(1 : i64) : i64
    %5 = llvm.alloca %4 x i64 : (i64) -> !llvm.ptr
    llvm.store %3, %5 : i64, !llvm.ptr
    %6 = arith.constant 0 : i32
    %7 = llvm.mlir.constant(1 : i64) : i64
    %8 = llvm.alloca %7 x i32 : (i64) -> !llvm.ptr
    llvm.store %6, %8 : i32, !llvm.ptr
    %9 = llvm.mlir.constant(1 : i64) : i64
    %10 = llvm.alloca %9 x i32 : (i64) -> !llvm.ptr
    llvm.store %arg2, %10 : i32, !llvm.ptr
    %11 = arith.extsi %arg2 : i32 to i64
    %12 = llvm.mlir.constant(1 : i64) : i64
    %13 = llvm.alloca %12 x i64 : (i64) -> !llvm.ptr
    llvm.store %11, %13 : i64, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %14 = llvm.load %1 : !llvm.ptr -> i32
    %15 = arith.constant 1 : i32
    %16 = arith.cmpi sgt, %14, %15 : i32
    cf.cond_br %16, ^bb1, ^bb2
    ^bb1:
      %17 = llvm.load %1 : !llvm.ptr -> i32
      %18 = arith.constant 1 : i32
      %19 = arith.subi %17, %18 : i32
      llvm.store %19, %1 : i32, !llvm.ptr
      %20 = llvm.load %10 : !llvm.ptr -> i32
      %21 = arith.muli %arg1, %20 : i32
      %22 = llvm.load %8 : !llvm.ptr -> i32
      %23 = arith.addi %21, %22 : i32
      %24 = arith.constant 10 : i32
      %25 = arith.divsi %23, %24 : i32
      llvm.store %25, %8 : i32, !llvm.ptr
      %26 = arith.constant 10 : i32
      %27 = arith.remsi %23, %26 : i32
      llvm.store %27, %10 : i32, !llvm.ptr
      %28 = llvm.load %5 : !llvm.ptr -> i64
      %29 = arith.cmpi slt, %28, %arg3 : i64
      cf.cond_br %29, ^bb3, ^bb4
      ^bb3:
        %30 = llvm.load %13 : !llvm.ptr -> i64
        %31 = llvm.load %10 : !llvm.ptr -> i32
        %32 = arith.extsi %31 : i32 to i64
        %33 = llvm.load %5 : !llvm.ptr -> i64
        %34 = arith.muli %32, %33 : i64
        %35 = arith.addi %30, %34 : i64
        llvm.store %35, %13 : i64, !llvm.ptr
        %36 = llvm.load %5 : !llvm.ptr -> i64
        %37 = arith.constant 10 : i32
        %39 = arith.extsi %37 : i32 to i64
        %38 = arith.muli %36, %39 : i64
        llvm.store %38, %5 : i64, !llvm.ptr
        cf.br ^bb5
      ^bb4:
        cf.br ^bb5
      ^bb5:
      cf.br ^bb0
    ^bb2:
    %40 = llvm.load %10 : !llvm.ptr -> i32
    %41 = arith.muli %arg1, %40 : i32
    %42 = llvm.load %8 : !llvm.ptr -> i32
    %43 = arith.addi %41, %42 : i32
    %44 = llvm.load %10 : !llvm.ptr -> i32
    %45 = arith.constant 0 : i32
    %46 = arith.cmpi eq, %44, %45 : i32
    %47 = scf.if %46 -> (i1) {
      %48 = arith.constant true
      scf.yield %48 : i1
    } else {
      %49 = arith.cmpi ne, %43, %arg2 : i32
      scf.yield %49 : i1
    }
    cf.cond_br %47, ^bb6, ^bb7
    ^bb6:
      %50 = arith.constant 0 : i32
      %51 = arith.extsi %50 : i32 to i64
      func.return %51 : i64
    ^bb7:
      cf.br ^bb8
    ^bb8:
    %52 = llvm.load %13 : !llvm.ptr -> i64
    func.return %52 : i64
  }
  func.func @main() -> i32 {
    %53 = arith.constant 100000 : i32
    %54 = arith.extsi %53 : i32 to i64
    %55 = arith.constant 0 : i32
    %56 = arith.extsi %55 : i32 to i64
    %57 = llvm.mlir.constant(1 : i64) : i64
    %58 = llvm.alloca %57 x i64 : (i64) -> !llvm.ptr
    llvm.store %56, %58 : i64, !llvm.ptr
    %59 = arith.constant 2 : i32
    %60 = llvm.mlir.constant(1 : i64) : i64
    %61 = llvm.alloca %60 x i32 : (i64) -> !llvm.ptr
    llvm.store %59, %61 : i32, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %62 = llvm.load %61 : !llvm.ptr -> i32
    %63 = arith.constant 100 : i32
    %64 = arith.cmpi sle, %62, %63 : i32
    cf.cond_br %64, ^bb10, ^bb11
    ^bb10:
      %65 = arith.constant 1 : i32
      %66 = llvm.mlir.constant(1 : i64) : i64
      %67 = llvm.alloca %66 x i32 : (i64) -> !llvm.ptr
      llvm.store %65, %67 : i32, !llvm.ptr
      cf.br ^bb12
      ^bb12:
      %68 = llvm.load %67 : !llvm.ptr -> i32
      %69 = arith.constant 9 : i32
      %70 = arith.cmpi sle, %68, %69 : i32
      cf.cond_br %70, ^bb13, ^bb14
      ^bb13:
        %71 = arith.constant 1 : i32
        %72 = llvm.mlir.constant(1 : i64) : i64
        %73 = llvm.alloca %72 x i32 : (i64) -> !llvm.ptr
        llvm.store %71, %73 : i32, !llvm.ptr
        cf.br ^bb15
        ^bb15:
        %74 = llvm.load %73 : !llvm.ptr -> i32
        %75 = arith.constant 9 : i32
        %76 = arith.cmpi sle, %74, %75 : i32
        cf.cond_br %76, ^bb16, ^bb17
        ^bb16:
          %77 = llvm.load %58 : !llvm.ptr -> i64
          %79 = llvm.load %61 : !llvm.ptr -> i32
          %80 = llvm.load %67 : !llvm.ptr -> i32
          %81 = llvm.load %73 : !llvm.ptr -> i32
          %78 = func.call @search(%79, %80, %81, %54) : (i32, i32, i32, i64) -> i64
          %82 = arith.addi %77, %78 : i64
          llvm.store %82, %58 : i64, !llvm.ptr
          %83 = llvm.load %73 : !llvm.ptr -> i32
          %84 = arith.constant 1 : i32
          %85 = arith.addi %83, %84 : i32
          llvm.store %85, %73 : i32, !llvm.ptr
          cf.br ^bb15
        ^bb17:
        %86 = llvm.load %67 : !llvm.ptr -> i32
        %87 = arith.constant 1 : i32
        %88 = arith.addi %86, %87 : i32
        llvm.store %88, %67 : i32, !llvm.ptr
        cf.br ^bb12
      ^bb14:
      %89 = llvm.load %61 : !llvm.ptr -> i32
      %90 = arith.constant 1 : i32
      %91 = arith.addi %89, %90 : i32
      llvm.store %91, %61 : i32, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    %92 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %93 = llvm.load %58 : !llvm.ptr -> i64
    %94 = arith.remsi %93, %54 : i64
    %95 = llvm.call @printf(%92, %94) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %96 = arith.constant 0 : i32
    func.return %96 : i32
  }
}