Problem 720

Unpredictable permutations: rank of the doubling construction mod 1e9+7. Ported from native C to pure Flow. Builds doubling arrays for 2^24.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 720
# Unpredictable permutations: rank of the doubling construction mod 1e9+7.
# Ported from native C to pure Flow. Builds doubling arrays for 2^24.

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

const MOD: i64 = 1000000007

function mulmod(a: i64, b: i64) -> i64 {
    return (a % MOD) * (b % MOD) % MOD
}

function main() -> i32 {
    let k: i32 = 25
    if k == 0 {
        printf("%lld\n", 1 as i64)
        return 0
    }
    if k == 1 {
        printf("%lld\n", 1 as i64)
        return 0
    }
    if k == 2 {
        printf("%lld\n", 3 as i64)
        return 0
    }

    let target: i64 = (1 as i64) << (k - 1)

    # build_arrays: use mutable pointers so we can reassign
    let mut vals: ptr<i32> = malloc(target * 4) as ptr<i32>
    let mut codes: ptr<i32> = malloc(target * 4) as ptr<i32>

    vals[0] = 1
    vals[1] = 3
    vals[2] = 2
    vals[3] = 4
    codes[0] = 0
    codes[1] = 1
    codes[2] = 0
    codes[3] = 0
    let mut size: i64 = 4

    while size < target {
        let m: i64 = size
        let n: i64 = m << 1
        let nv: ptr<i32> = malloc(n * 4) as ptr<i32>
        let nc: ptr<i32> = malloc(n * 4) as ptr<i32>

        let mut i: i64 = 0
        while i < m - 1 {
            let v: i32 = vals[i]
            nv[i] = (v << 1) - 1
            nc[i] = (v - 1) + codes[i]
            i = i + 1
        }
        nv[m - 1] = 2
        nc[m - 1] = 0
        let v_last: i32 = vals[m - 1]
        nv[m] = (v_last << 1) - 1
        nc[m] = (m - 2) as i32
        let mut j: i64 = 1
        while j < m {
            let v: i32 = vals[j]
            nv[m + j] = v << 1
            nc[m + j] = codes[j]
            j = j + 1
        }

        free(vals)
        free(codes)
        vals = nv
        codes = nc
        size = n
    }

    # rank_from_prev(vals, codes, target)
    let m: i64 = target
    let mut rank: i64 = 0
    let mut fact: i64 = 1
    let mut step: i64 = 1

    let mut j: i64 = m - 1
    while j > 0 {
        let l: i64 = codes[j] as i64
        rank = (rank + mulmod(l, fact)) % MOD
        fact = mulmod(fact, step)
        step = step + 1
        j = j - 1
    }
    let l_mid: i64 = m - 2
    rank = (rank + mulmod(l_mid, fact)) % MOD
    fact = mulmod(fact, step)
    step = step + 1
    fact = mulmod(fact, step)
    step = step + 1

    let mut i: i64 = m - 2
    while i >= 0 {
        let v: i32 = vals[i]
        let l: i64 = (v - 1) as i64 + (codes[i] as i64)
        rank = (rank + mulmod(l, fact)) % MOD
        fact = mulmod(fact, step)
        step = step + 1
        i = i - 1
    }

    let rank0: i64 = rank
    let result: i64 = (rank0 + 1) % MOD

    free(vals)
    free(codes)
    printf("%lld\n", result)
    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 mulmod_i64_i64(int64_t a, int64_t b);
int32_t main(void);

static const int64_t MOD = 1000000007;




int64_t mulmod_i64_i64(int64_t a, int64_t b) {
    return FLOW_CHECKED_MOD(((FLOW_CHECKED_MOD((a), (MOD)) * FLOW_CHECKED_MOD((b), (MOD)))), (MOD));
}

int32_t main(void) {
    int32_t k = 25;
    if (k == 0) {
        printf("%lld\n", ((int64_t)(1)));
        return 0;
    }
    if (k == 1) {
        printf("%lld\n", ((int64_t)(1)));
        return 0;
    }
    if (k == 2) {
        printf("%lld\n", ((int64_t)(3)));
        return 0;
    }
    int64_t target = FLOW_CHECKED_SHL((((int64_t)(1))), ((k - 1)));
    int32_t* vals = (int32_t*)(((int32_t*)(malloc((target * 4)))));
    int32_t* codes = (int32_t*)(((int32_t*)(malloc((target * 4)))));
    vals[0] = 1;
    vals[1] = 3;
    vals[2] = 2;
    vals[3] = 4;
    codes[0] = 0;
    codes[1] = 1;
    codes[2] = 0;
    codes[3] = 0;
    int64_t size = 4;
    while (size < target) {
        int64_t m = size;
        int64_t n = FLOW_CHECKED_SHL((m), (1));
        int32_t* nv = (int32_t*)(((int32_t*)(malloc((n * 4)))));
        int32_t* nc = (int32_t*)(((int32_t*)(malloc((n * 4)))));
        int64_t i = 0;
        while (i < (m - 1)) {
            int32_t v = vals[i];
            nv[i] = (FLOW_CHECKED_SHL((v), (1)) - 1);
            nc[i] = ((v - 1) + codes[i]);
            i = (i + 1);
        }
        nv[(m - 1)] = 2;
        nc[(m - 1)] = 0;
        int32_t v_last = vals[(m - 1)];
        nv[m] = (FLOW_CHECKED_SHL((v_last), (1)) - 1);
        nc[m] = ((int32_t)((m - 2)));
        int64_t j = 1;
        while (j < m) {
            int32_t v = vals[j];
            nv[(m + j)] = FLOW_CHECKED_SHL((v), (1));
            nc[(m + j)] = codes[j];
            j = (j + 1);
        }
        free(vals);
        free(codes);
        vals = nv;
        codes = nc;
        size = n;
    }
    int64_t m = target;
    int64_t rank = 0;
    int64_t fact = 1;
    int64_t step = 1;
    int64_t j = (m - 1);
    while (j > 0) {
        int64_t l = ((int64_t)(codes[j]));
        rank = FLOW_CHECKED_MOD(((rank + mulmod_i64_i64(l, fact))), (MOD));
        fact = mulmod_i64_i64(fact, step);
        step = (step + 1);
        j = (j - 1);
    }
    int64_t l_mid = (m - 2);
    rank = FLOW_CHECKED_MOD(((rank + mulmod_i64_i64(l_mid, fact))), (MOD));
    fact = mulmod_i64_i64(fact, step);
    step = (step + 1);
    fact = mulmod_i64_i64(fact, step);
    step = (step + 1);
    int64_t i = (m - 2);
    while (i >= 0) {
        int32_t v = vals[i];
        int64_t l = (((int64_t)((v - 1))) + ((int64_t)(codes[i])));
        rank = FLOW_CHECKED_MOD(((rank + mulmod_i64_i64(l, fact))), (MOD));
        fact = mulmod_i64_i64(fact, step);
        step = (step + 1);
        i = (i - 1);
    }
    int64_t rank0 = rank;
    int64_t result = FLOW_CHECKED_MOD(((rank0 + 1)), (MOD));
    free(vals);
    free(codes);
    printf("%lld\n", result);
    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 private @calloc(i64, i64) -> !llvm.ptr
  func.func private @free(!llvm.ptr) -> ()
  func.func private @malloc(i64) -> !llvm.ptr
  // Constant: MOD
  llvm.mlir.global internal constant @MOD(1000000007 : i64) : i64
  func.func @mulmod(%arg0: i64, %arg1: i64) -> i64 {
    %0 = llvm.mlir.addressof @MOD : !llvm.ptr
    %1 = llvm.load %0 : !llvm.ptr -> i64
    %2 = arith.remsi %arg0, %1 : i64
    %3 = llvm.mlir.addressof @MOD : !llvm.ptr
    %4 = llvm.load %3 : !llvm.ptr -> i64
    %5 = arith.remsi %arg1, %4 : i64
    %6 = arith.muli %2, %5 : i64
    %7 = llvm.mlir.addressof @MOD : !llvm.ptr
    %8 = llvm.load %7 : !llvm.ptr -> i64
    %9 = arith.remsi %6, %8 : i64
    func.return %9 : i64
  }
  func.func @main() -> i32 {
    %10 = arith.constant 25 : i32
    %11 = arith.constant 0 : i32
    %12 = arith.cmpi eq, %10, %11 : i32
    cf.cond_br %12, ^bb0, ^bb1
    ^bb0:
      %13 = llvm.mlir.addressof @str_0 : !llvm.ptr
      %14 = arith.constant 1 : i32
      %15 = arith.extsi %14 : i32 to i64
      %16 = llvm.call @printf(%13, %15) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
      %17 = arith.constant 0 : i32
      func.return %17 : i32
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %18 = arith.constant 1 : i32
    %19 = arith.cmpi eq, %10, %18 : i32
    cf.cond_br %19, ^bb3, ^bb4
    ^bb3:
      %20 = llvm.mlir.addressof @str_0 : !llvm.ptr
      %21 = arith.constant 1 : i32
      %22 = arith.extsi %21 : i32 to i64
      %23 = llvm.call @printf(%20, %22) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
      %24 = arith.constant 0 : i32
      func.return %24 : i32
    ^bb4:
      cf.br ^bb5
    ^bb5:
    %25 = arith.constant 2 : i32
    %26 = arith.cmpi eq, %10, %25 : i32
    cf.cond_br %26, ^bb6, ^bb7
    ^bb6:
      %27 = llvm.mlir.addressof @str_0 : !llvm.ptr
      %28 = arith.constant 3 : i32
      %29 = arith.extsi %28 : i32 to i64
      %30 = llvm.call @printf(%27, %29) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
      %31 = arith.constant 0 : i32
      func.return %31 : i32
    ^bb7:
      cf.br ^bb8
    ^bb8:
    %32 = arith.constant 1 : i32
    %33 = arith.extsi %32 : i32 to i64
    %34 = arith.constant 1 : i32
    %35 = arith.subi %10, %34 : i32
    %37 = arith.extsi %35 : i32 to i64
    %36 = arith.shli %33, %37 : i64
    %39 = arith.constant 4 : i32
    %41 = arith.extsi %39 : i32 to i64
    %40 = arith.muli %36, %41 : i64
    %38 = func.call @malloc(%40) : (i64) -> !llvm.ptr
    %42 = llvm.mlir.constant(1 : i64) : i64
    %43 = llvm.alloca %42 x !llvm.ptr : (i64) -> !llvm.ptr
    llvm.store %38, %43 : !llvm.ptr, !llvm.ptr
    %45 = arith.constant 4 : i32
    %47 = arith.extsi %45 : i32 to i64
    %46 = arith.muli %36, %47 : i64
    %44 = func.call @malloc(%46) : (i64) -> !llvm.ptr
    %48 = llvm.mlir.constant(1 : i64) : i64
    %49 = llvm.alloca %48 x !llvm.ptr : (i64) -> !llvm.ptr
    llvm.store %44, %49 : !llvm.ptr, !llvm.ptr
    %50 = arith.constant 1 : i32
    %51 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
    %52 = arith.constant 0 : i32
    %53 = arith.extsi %52 : i32 to i64
    %54 = llvm.getelementptr %51[%53] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %50, %54 : i32, !llvm.ptr
    %55 = arith.constant 3 : i32
    %56 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
    %57 = arith.constant 1 : i32
    %58 = arith.extsi %57 : i32 to i64
    %59 = llvm.getelementptr %56[%58] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %55, %59 : i32, !llvm.ptr
    %60 = arith.constant 2 : i32
    %61 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
    %62 = arith.constant 2 : i32
    %63 = arith.extsi %62 : i32 to i64
    %64 = llvm.getelementptr %61[%63] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %60, %64 : i32, !llvm.ptr
    %65 = arith.constant 4 : i32
    %66 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
    %67 = arith.constant 3 : i32
    %68 = arith.extsi %67 : i32 to i64
    %69 = llvm.getelementptr %66[%68] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %65, %69 : i32, !llvm.ptr
    %70 = arith.constant 0 : i32
    %71 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
    %72 = arith.constant 0 : i32
    %73 = arith.extsi %72 : i32 to i64
    %74 = llvm.getelementptr %71[%73] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %70, %74 : i32, !llvm.ptr
    %75 = arith.constant 1 : i32
    %76 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
    %77 = arith.constant 1 : i32
    %78 = arith.extsi %77 : i32 to i64
    %79 = llvm.getelementptr %76[%78] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %75, %79 : i32, !llvm.ptr
    %80 = arith.constant 0 : i32
    %81 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
    %82 = arith.constant 2 : i32
    %83 = arith.extsi %82 : i32 to i64
    %84 = llvm.getelementptr %81[%83] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %80, %84 : i32, !llvm.ptr
    %85 = arith.constant 0 : i32
    %86 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
    %87 = arith.constant 3 : i32
    %88 = arith.extsi %87 : i32 to i64
    %89 = llvm.getelementptr %86[%88] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %85, %89 : i32, !llvm.ptr
    %90 = arith.constant 4 : i32
    %91 = arith.extsi %90 : i32 to i64
    %92 = llvm.mlir.constant(1 : i64) : i64
    %93 = llvm.alloca %92 x i64 : (i64) -> !llvm.ptr
    llvm.store %91, %93 : i64, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %94 = llvm.load %93 : !llvm.ptr -> i64
    %95 = arith.cmpi slt, %94, %36 : i64
    cf.cond_br %95, ^bb10, ^bb11
    ^bb10:
      %96 = llvm.load %93 : !llvm.ptr -> i64
      %97 = arith.constant 1 : i32
      %99 = arith.extsi %97 : i32 to i64
      %98 = arith.shli %96, %99 : i64
      %101 = arith.constant 4 : i32
      %103 = arith.extsi %101 : i32 to i64
      %102 = arith.muli %98, %103 : i64
      %100 = func.call @malloc(%102) : (i64) -> !llvm.ptr
      %105 = arith.constant 4 : i32
      %107 = arith.extsi %105 : i32 to i64
      %106 = arith.muli %98, %107 : i64
      %104 = func.call @malloc(%106) : (i64) -> !llvm.ptr
      %108 = arith.constant 0 : i32
      %109 = arith.extsi %108 : i32 to i64
      %110 = llvm.mlir.constant(1 : i64) : i64
      %111 = llvm.alloca %110 x i64 : (i64) -> !llvm.ptr
      llvm.store %109, %111 : i64, !llvm.ptr
      cf.br ^bb12
      ^bb12:
      %112 = llvm.load %111 : !llvm.ptr -> i64
      %113 = arith.constant 1 : i32
      %115 = arith.extsi %113 : i32 to i64
      %114 = arith.subi %96, %115 : i64
      %116 = arith.cmpi slt, %112, %114 : i64
      cf.cond_br %116, ^bb13, ^bb14
      ^bb13:
        %118 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
        %119 = llvm.load %111 : !llvm.ptr -> i64
        %120 = llvm.getelementptr %118[%119] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %117 = llvm.load %120 : !llvm.ptr -> i32
        %121 = arith.constant 1 : i32
        %122 = arith.shli %117, %121 : i32
        %123 = arith.constant 1 : i32
        %124 = arith.subi %122, %123 : i32
        %125 = llvm.load %111 : !llvm.ptr -> i64
        %126 = llvm.getelementptr %100[%125] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %124, %126 : i32, !llvm.ptr
        %127 = arith.constant 1 : i32
        %128 = arith.subi %117, %127 : i32
        %130 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
        %131 = llvm.load %111 : !llvm.ptr -> i64
        %132 = llvm.getelementptr %130[%131] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %129 = llvm.load %132 : !llvm.ptr -> i32
        %133 = arith.addi %128, %129 : i32
        %134 = llvm.load %111 : !llvm.ptr -> i64
        %135 = llvm.getelementptr %104[%134] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %133, %135 : i32, !llvm.ptr
        %136 = llvm.load %111 : !llvm.ptr -> i64
        %137 = arith.constant 1 : i32
        %139 = arith.extsi %137 : i32 to i64
        %138 = arith.addi %136, %139 : i64
        llvm.store %138, %111 : i64, !llvm.ptr
        cf.br ^bb12
      ^bb14:
      %140 = arith.constant 2 : i32
      %141 = arith.constant 1 : i32
      %143 = arith.extsi %141 : i32 to i64
      %142 = arith.subi %96, %143 : i64
      %144 = llvm.getelementptr %100[%142] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %140, %144 : i32, !llvm.ptr
      %145 = arith.constant 0 : i32
      %146 = arith.constant 1 : i32
      %148 = arith.extsi %146 : i32 to i64
      %147 = arith.subi %96, %148 : i64
      %149 = llvm.getelementptr %104[%147] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %145, %149 : i32, !llvm.ptr
      %151 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
      %152 = arith.constant 1 : i32
      %154 = arith.extsi %152 : i32 to i64
      %153 = arith.subi %96, %154 : i64
      %155 = llvm.getelementptr %151[%153] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %150 = llvm.load %155 : !llvm.ptr -> i32
      %156 = arith.constant 1 : i32
      %157 = arith.shli %150, %156 : i32
      %158 = arith.constant 1 : i32
      %159 = arith.subi %157, %158 : i32
      %160 = llvm.getelementptr %100[%96] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %159, %160 : i32, !llvm.ptr
      %161 = arith.constant 2 : i32
      %163 = arith.extsi %161 : i32 to i64
      %162 = arith.subi %96, %163 : i64
      %164 = arith.trunci %162 : i64 to i32
      %165 = llvm.getelementptr %104[%96] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %164, %165 : i32, !llvm.ptr
      %166 = arith.constant 1 : i32
      %167 = arith.extsi %166 : i32 to i64
      %168 = llvm.mlir.constant(1 : i64) : i64
      %169 = llvm.alloca %168 x i64 : (i64) -> !llvm.ptr
      llvm.store %167, %169 : i64, !llvm.ptr
      cf.br ^bb15
      ^bb15:
      %170 = llvm.load %169 : !llvm.ptr -> i64
      %171 = arith.cmpi slt, %170, %96 : i64
      cf.cond_br %171, ^bb16, ^bb17
      ^bb16:
        %173 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
        %174 = llvm.load %169 : !llvm.ptr -> i64
        %175 = llvm.getelementptr %173[%174] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %172 = llvm.load %175 : !llvm.ptr -> i32
        %176 = arith.constant 1 : i32
        %177 = arith.shli %172, %176 : i32
        %178 = llvm.load %169 : !llvm.ptr -> i64
        %179 = arith.addi %96, %178 : i64
        %180 = llvm.getelementptr %100[%179] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %177, %180 : i32, !llvm.ptr
        %182 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
        %183 = llvm.load %169 : !llvm.ptr -> i64
        %184 = llvm.getelementptr %182[%183] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %181 = llvm.load %184 : !llvm.ptr -> i32
        %185 = llvm.load %169 : !llvm.ptr -> i64
        %186 = arith.addi %96, %185 : i64
        %187 = llvm.getelementptr %104[%186] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %181, %187 : i32, !llvm.ptr
        %188 = llvm.load %169 : !llvm.ptr -> i64
        %189 = arith.constant 1 : i32
        %191 = arith.extsi %189 : i32 to i64
        %190 = arith.addi %188, %191 : i64
        llvm.store %190, %169 : i64, !llvm.ptr
        cf.br ^bb15
      ^bb17:
      %193 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
      func.call @free(%193) : (!llvm.ptr) -> ()
      %195 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
      func.call @free(%195) : (!llvm.ptr) -> ()
      llvm.store %100, %43 : !llvm.ptr, !llvm.ptr
      llvm.store %104, %49 : !llvm.ptr, !llvm.ptr
      llvm.store %98, %93 : i64, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    %196 = arith.constant 0 : i32
    %197 = arith.extsi %196 : i32 to i64
    %198 = llvm.mlir.constant(1 : i64) : i64
    %199 = llvm.alloca %198 x i64 : (i64) -> !llvm.ptr
    llvm.store %197, %199 : i64, !llvm.ptr
    %200 = arith.constant 1 : i32
    %201 = arith.extsi %200 : i32 to i64
    %202 = llvm.mlir.constant(1 : i64) : i64
    %203 = llvm.alloca %202 x i64 : (i64) -> !llvm.ptr
    llvm.store %201, %203 : i64, !llvm.ptr
    %204 = arith.constant 1 : i32
    %205 = arith.extsi %204 : i32 to i64
    %206 = llvm.mlir.constant(1 : i64) : i64
    %207 = llvm.alloca %206 x i64 : (i64) -> !llvm.ptr
    llvm.store %205, %207 : i64, !llvm.ptr
    %208 = arith.constant 1 : i32
    %210 = arith.extsi %208 : i32 to i64
    %209 = arith.subi %36, %210 : i64
    %211 = llvm.mlir.constant(1 : i64) : i64
    %212 = llvm.alloca %211 x i64 : (i64) -> !llvm.ptr
    llvm.store %209, %212 : i64, !llvm.ptr
    cf.br ^bb18
    ^bb18:
    %213 = llvm.load %212 : !llvm.ptr -> i64
    %214 = arith.constant 0 : i32
    %216 = arith.extsi %214 : i32 to i64
    %215 = arith.cmpi sgt, %213, %216 : i64
    cf.cond_br %215, ^bb19, ^bb20
    ^bb19:
      %218 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
      %219 = llvm.load %212 : !llvm.ptr -> i64
      %220 = llvm.getelementptr %218[%219] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %217 = llvm.load %220 : !llvm.ptr -> i32
      %221 = arith.extsi %217 : i32 to i64
      %222 = llvm.load %199 : !llvm.ptr -> i64
      %224 = llvm.load %203 : !llvm.ptr -> i64
      %223 = func.call @mulmod(%221, %224) : (i64, i64) -> i64
      %225 = arith.addi %222, %223 : i64
      %226 = llvm.mlir.addressof @MOD : !llvm.ptr
      %227 = llvm.load %226 : !llvm.ptr -> i64
      %228 = arith.remsi %225, %227 : i64
      llvm.store %228, %199 : i64, !llvm.ptr
      %230 = llvm.load %203 : !llvm.ptr -> i64
      %231 = llvm.load %207 : !llvm.ptr -> i64
      %229 = func.call @mulmod(%230, %231) : (i64, i64) -> i64
      llvm.store %229, %203 : i64, !llvm.ptr
      %232 = llvm.load %207 : !llvm.ptr -> i64
      %233 = arith.constant 1 : i32
      %235 = arith.extsi %233 : i32 to i64
      %234 = arith.addi %232, %235 : i64
      llvm.store %234, %207 : i64, !llvm.ptr
      %236 = llvm.load %212 : !llvm.ptr -> i64
      %237 = arith.constant 1 : i32
      %239 = arith.extsi %237 : i32 to i64
      %238 = arith.subi %236, %239 : i64
      llvm.store %238, %212 : i64, !llvm.ptr
      cf.br ^bb18
    ^bb20:
    %240 = arith.constant 2 : i32
    %242 = arith.extsi %240 : i32 to i64
    %241 = arith.subi %36, %242 : i64
    %243 = llvm.load %199 : !llvm.ptr -> i64
    %245 = llvm.load %203 : !llvm.ptr -> i64
    %244 = func.call @mulmod(%241, %245) : (i64, i64) -> i64
    %246 = arith.addi %243, %244 : i64
    %247 = llvm.mlir.addressof @MOD : !llvm.ptr
    %248 = llvm.load %247 : !llvm.ptr -> i64
    %249 = arith.remsi %246, %248 : i64
    llvm.store %249, %199 : i64, !llvm.ptr
    %251 = llvm.load %203 : !llvm.ptr -> i64
    %252 = llvm.load %207 : !llvm.ptr -> i64
    %250 = func.call @mulmod(%251, %252) : (i64, i64) -> i64
    llvm.store %250, %203 : i64, !llvm.ptr
    %253 = llvm.load %207 : !llvm.ptr -> i64
    %254 = arith.constant 1 : i32
    %256 = arith.extsi %254 : i32 to i64
    %255 = arith.addi %253, %256 : i64
    llvm.store %255, %207 : i64, !llvm.ptr
    %258 = llvm.load %203 : !llvm.ptr -> i64
    %259 = llvm.load %207 : !llvm.ptr -> i64
    %257 = func.call @mulmod(%258, %259) : (i64, i64) -> i64
    llvm.store %257, %203 : i64, !llvm.ptr
    %260 = llvm.load %207 : !llvm.ptr -> i64
    %261 = arith.constant 1 : i32
    %263 = arith.extsi %261 : i32 to i64
    %262 = arith.addi %260, %263 : i64
    llvm.store %262, %207 : i64, !llvm.ptr
    %264 = arith.constant 2 : i32
    %266 = arith.extsi %264 : i32 to i64
    %265 = arith.subi %36, %266 : i64
    %267 = llvm.mlir.constant(1 : i64) : i64
    %268 = llvm.alloca %267 x i64 : (i64) -> !llvm.ptr
    llvm.store %265, %268 : i64, !llvm.ptr
    cf.br ^bb21
    ^bb21:
    %269 = llvm.load %268 : !llvm.ptr -> i64
    %270 = arith.constant 0 : i32
    %272 = arith.extsi %270 : i32 to i64
    %271 = arith.cmpi sge, %269, %272 : i64
    cf.cond_br %271, ^bb22, ^bb23
    ^bb22:
      %274 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
      %275 = llvm.load %268 : !llvm.ptr -> i64
      %276 = llvm.getelementptr %274[%275] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %273 = llvm.load %276 : !llvm.ptr -> i32
      %277 = arith.constant 1 : i32
      %278 = arith.subi %273, %277 : i32
      %279 = arith.extsi %278 : i32 to i64
      %281 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
      %282 = llvm.load %268 : !llvm.ptr -> i64
      %283 = llvm.getelementptr %281[%282] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %280 = llvm.load %283 : !llvm.ptr -> i32
      %284 = arith.extsi %280 : i32 to i64
      %285 = arith.addi %279, %284 : i64
      %286 = llvm.load %199 : !llvm.ptr -> i64
      %288 = llvm.load %203 : !llvm.ptr -> i64
      %287 = func.call @mulmod(%285, %288) : (i64, i64) -> i64
      %289 = arith.addi %286, %287 : i64
      %290 = llvm.mlir.addressof @MOD : !llvm.ptr
      %291 = llvm.load %290 : !llvm.ptr -> i64
      %292 = arith.remsi %289, %291 : i64
      llvm.store %292, %199 : i64, !llvm.ptr
      %294 = llvm.load %203 : !llvm.ptr -> i64
      %295 = llvm.load %207 : !llvm.ptr -> i64
      %293 = func.call @mulmod(%294, %295) : (i64, i64) -> i64
      llvm.store %293, %203 : i64, !llvm.ptr
      %296 = llvm.load %207 : !llvm.ptr -> i64
      %297 = arith.constant 1 : i32
      %299 = arith.extsi %297 : i32 to i64
      %298 = arith.addi %296, %299 : i64
      llvm.store %298, %207 : i64, !llvm.ptr
      %300 = llvm.load %268 : !llvm.ptr -> i64
      %301 = arith.constant 1 : i32
      %303 = arith.extsi %301 : i32 to i64
      %302 = arith.subi %300, %303 : i64
      llvm.store %302, %268 : i64, !llvm.ptr
      cf.br ^bb21
    ^bb23:
    %304 = llvm.load %199 : !llvm.ptr -> i64
    %305 = arith.constant 1 : i32
    %307 = arith.extsi %305 : i32 to i64
    %306 = arith.addi %304, %307 : i64
    %308 = llvm.mlir.addressof @MOD : !llvm.ptr
    %309 = llvm.load %308 : !llvm.ptr -> i64
    %310 = arith.remsi %306, %309 : i64
    %312 = llvm.load %43 : !llvm.ptr -> !llvm.ptr
    func.call @free(%312) : (!llvm.ptr) -> ()
    %314 = llvm.load %49 : !llvm.ptr -> !llvm.ptr
    func.call @free(%314) : (!llvm.ptr) -> ()
    %315 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %316 = llvm.call @printf(%315, %310) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %317 = arith.constant 0 : i32
    func.return %317 : i32
  }
}