Problem 296

Count triangles with BE = ac/(a+b) integer, perimeter <= 100000.

Answer1137208419
Output1137208419
StatusPASS
Native helperno
Runtime40 ms
Peak memory4736 KB
Time complexityO(n^2) (estimated)
Space complexityO(n^2) (estimated)

Performance comparison

MetricOur solutionBest known
Time complexityO(n^2)O(n^2)
Space complexityO(n^2)O(n^2)
ApproachFlow solutionBottom-up DP
VerdictOptimal

Flow source

# Project Euler 296
# Count triangles with BE = ac/(a+b) integer, perimeter <= 100000.

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

function floor_sum(n0: i64, m0: i64, a0: i64, b0: i64) -> i64 {
    let mut total: i64 = 0
    let mut n: i64 = n0
    let mut m: i64 = m0
    let mut a: i64 = a0
    let mut b: i64 = b0
    while true {
        if a >= m {
            total = total + (n - 1) * n * (a / m) / 2
            a = a % m
        }
        if b >= m {
            total = total + n * (b / m)
            b = b % m
        }
        let y: i64 = a * n + b
        if y < m {
            return total
        }
        n = y / m
        b = y % m
        let tmp: i64 = m
        m = a
        a = tmp
    }
    return total
}

function main() -> i32 {
    let limit: i64 = 100000
    let max_q: i64 = limit / 3
    let mu: ptr<i32> = calloc(max_q + 1, 4)
    let is_comp: ptr<i8> = calloc(max_q + 1, 1)
    let primes: ptr<i64> = calloc(max_q / 5 + 10, 8)
    mu[1] = 1
    let mut pc: i64 = 0
    for n in 2..(max_q + 1) {
        if is_comp[n] == 0 {
            primes[pc] = n
            pc = pc + 1
            mu[n] = -1
        }
        let mi: i32 = mu[n]
        let mut j: i64 = 0
        while j < pc {
            let p: i64 = primes[j]
            if p > max_q / n { break }
            let ip: i64 = n * p
            is_comp[ip] = 1
            if n % p == 0 {
                mu[ip] = 0
                break
            }
            mu[ip] = 0 - mi
            j = j + 1
        }
    }

    # divisors[q] stored as flat: first count, then pairs (d, mu) in separate arrays via CSR
    let offs: ptr<i64> = calloc(max_q + 2, 8)
    for d in 1..(max_q + 1) {
        if mu[d] != 0 {
            for mult in d..(max_q + 1) step d {
                offs[mult + 1] = offs[mult + 1] + 1
            }
        }
    }
    for i in 1..(max_q + 2) {
        offs[i] = offs[i] + offs[i - 1]
    }
    let total_pairs: i64 = offs[max_q + 1]
    let ds: ptr<i64> = calloc(total_pairs, 8)
    let ms: ptr<i32> = calloc(total_pairs, 4)
    let cur: ptr<i64> = calloc(max_q + 1, 8)
    for i in 0..(max_q + 1) {
        cur[i] = offs[i]
    }
    for d in 1..(max_q + 1) {
        if mu[d] != 0 {
            for mult in d..(max_q + 1) step d {
                let at: i64 = cur[mult]
                ds[at] = d
                ms[at] = mu[d]
                cur[mult] = at + 1
            }
        }
    }

    let mut count: i64 = 0
    for q in 2..(max_q + 1) {
        let max_g_plus_k: i64 = limit / q
        let half_q: i64 = q / 2
        let start: i64 = offs[q]
        let end: i64 = offs[q + 1]

        # reduced_up_to_half = sum mu*(half_q//d)
        let mut reduced_up_to_half: i64 = 0
        for t in start..end {
            reduced_up_to_half = reduced_up_to_half + (ms[t] as i64) * (half_q / ds[t])
        }

        for g in 2..max_g_plus_k {
            let mut min_h: i64 = 1
            let alt: i64 = 2 * g - max_g_plus_k
            if alt > min_h { min_h = alt }
            let before: i64 = (min_h * q + g - 1) / g - 1
            if before < half_q {
                let mut coprime_before: i64 = 0
                let mut floor_half: i64 = 0
                let mut floor_before: i64 = 0
                for t in start..end {
                    let dd: i64 = ds[t]
                    let mmu: i64 = ms[t] as i64
                    coprime_before = coprime_before + mmu * (before / dd)
                    let n1: i64 = half_q / dd
                    if n1 > 0 {
                        floor_half = floor_half + mmu * floor_sum(n1 + 1, q / dd, g, 0)
                    }
                    let n2: i64 = before / dd
                    if n2 > 0 {
                        floor_before = floor_before + mmu * floor_sum(n2 + 1, q / dd, g, 0)
                    }
                }
                let reduced_count: i64 = reduced_up_to_half - coprime_before
                let floor_part: i64 = floor_half - floor_before
                count = count + floor_part - (min_h - 1) * reduced_count
            }
        }
    }
    printf("%lld\n", count)
    free(mu); free(is_comp); free(primes); free(offs); free(ds); free(ms); free(cur)
    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 floor_sum_i64_i64_i64_i64(int64_t n0, int64_t m0, int64_t a0, int64_t b0);
int32_t main(void);



int64_t floor_sum_i64_i64_i64_i64(int64_t n0, int64_t m0, int64_t a0, int64_t b0) {
    int64_t total = 0;
    int64_t n = n0;
    int64_t m = m0;
    int64_t a = a0;
    int64_t b = b0;
    while (1) {
        if (a >= m) {
            total = (total + FLOW_CHECKED_DIV(((((n - 1) * n) * FLOW_CHECKED_DIV((a), (m)))), (2)));
            a = FLOW_CHECKED_MOD((a), (m));
        }
        if (b >= m) {
            total = (total + (n * FLOW_CHECKED_DIV((b), (m))));
            b = FLOW_CHECKED_MOD((b), (m));
        }
        int64_t y = ((a * n) + b);
        if (y < m) {
            return total;
        }
        n = FLOW_CHECKED_DIV((y), (m));
        b = FLOW_CHECKED_MOD((y), (m));
        int64_t tmp = m;
        m = a;
        a = tmp;
    }
    return total;
}

int32_t main(void) {
    int64_t limit = 100000;
    int64_t max_q = FLOW_CHECKED_DIV((limit), (3));
    int32_t* mu = (int32_t*)(calloc((max_q + 1), 4));
    int8_t* is_comp = (int8_t*)(calloc((max_q + 1), 1));
    int64_t* primes = (int64_t*)(calloc((FLOW_CHECKED_DIV((max_q), (5)) + 10), 8));
    mu[1] = 1;
    int64_t pc = 0;
    int32_t __flow_step_1 = 1;
    for (int32_t n = 2; (2 <= (max_q + 1)) ? n < (max_q + 1) : n > (max_q + 1); n += (2 <= (max_q + 1)) ? 1 : -1) {
        if (is_comp[n] == 0) {
            primes[pc] = n;
            pc = (pc + 1);
            mu[n] = (-1);
        }
        int32_t mi = mu[n];
        int64_t j = 0;
        while (j < pc) {
            int64_t p = primes[j];
            if (p > FLOW_CHECKED_DIV((max_q), (n))) {
                break;
            }
            int64_t ip = (n * p);
            is_comp[ip] = 1;
            if (FLOW_CHECKED_MOD((n), (p)) == 0) {
                mu[ip] = 0;
                break;
            }
            mu[ip] = (0 - mi);
            j = (j + 1);
        }
    }
    int64_t* offs = (int64_t*)(calloc((max_q + 2), 8));
    int32_t __flow_step_2 = 1;
    for (int32_t d = 1; (1 <= (max_q + 1)) ? d < (max_q + 1) : d > (max_q + 1); d += (1 <= (max_q + 1)) ? 1 : -1) {
        if (mu[d] != 0) {
            int32_t __flow_step_3 = d;
            #pragma clang loop vectorize(enable) interleave(enable)
            #pragma GCC ivdep
            for (int32_t mult = d; (__flow_step_3 > 0) ? mult < (max_q + 1) : mult > (max_q + 1); mult += __flow_step_3) {
                offs[(mult + 1)] = (offs[(mult + 1)] + 1);
            }
        }
    }
    int32_t __flow_step_4 = 1;
    for (int32_t i = 1; (1 <= (max_q + 2)) ? i < (max_q + 2) : i > (max_q + 2); i += (1 <= (max_q + 2)) ? 1 : -1) {
        offs[i] = (offs[i] + offs[(i - 1)]);
    }
    int64_t total_pairs = offs[(max_q + 1)];
    int64_t* ds = (int64_t*)(calloc(total_pairs, 8));
    int32_t* ms = (int32_t*)(calloc(total_pairs, 4));
    int64_t* cur = (int64_t*)(calloc((max_q + 1), 8));
    int32_t __flow_step_5 = 1;
    for (int32_t i = 0; (0 <= (max_q + 1)) ? i < (max_q + 1) : i > (max_q + 1); i += (0 <= (max_q + 1)) ? 1 : -1) {
        cur[i] = offs[i];
    }
    int32_t __flow_step_6 = 1;
    for (int32_t d = 1; (1 <= (max_q + 1)) ? d < (max_q + 1) : d > (max_q + 1); d += (1 <= (max_q + 1)) ? 1 : -1) {
        if (mu[d] != 0) {
            int32_t __flow_step_7 = d;
            #pragma clang loop vectorize(enable) interleave(enable)
            #pragma GCC ivdep
            for (int32_t mult = d; (__flow_step_7 > 0) ? mult < (max_q + 1) : mult > (max_q + 1); mult += __flow_step_7) {
                int64_t at = cur[mult];
                ds[at] = d;
                ms[at] = mu[d];
                cur[mult] = (at + 1);
            }
        }
    }
    int64_t count = 0;
    int32_t __flow_step_8 = 1;
    for (int32_t q = 2; (2 <= (max_q + 1)) ? q < (max_q + 1) : q > (max_q + 1); q += (2 <= (max_q + 1)) ? 1 : -1) {
        int64_t max_g_plus_k = FLOW_CHECKED_DIV((limit), (q));
        int64_t half_q = FLOW_CHECKED_DIV((q), (2));
        int64_t start = offs[q];
        int64_t end = offs[(q + 1)];
        int64_t reduced_up_to_half = 0;
        int32_t __flow_step_9 = 1;
        for (int32_t t = start; (start <= end) ? t < end : t > end; t += (start <= end) ? 1 : -1) {
            reduced_up_to_half = (reduced_up_to_half + (((int64_t)(ms[t])) * FLOW_CHECKED_DIV((half_q), (ds[t]))));
        }
        int32_t __flow_step_10 = 1;
        for (int32_t g = 2; (2 <= max_g_plus_k) ? g < max_g_plus_k : g > max_g_plus_k; g += (2 <= max_g_plus_k) ? 1 : -1) {
            int64_t min_h = 1;
            int64_t alt = ((2 * g) - max_g_plus_k);
            if (alt > min_h) {
                min_h = alt;
            }
            int64_t before = (FLOW_CHECKED_DIV(((((min_h * q) + g) - 1)), (g)) - 1);
            if (before < half_q) {
                int64_t coprime_before = 0;
                int64_t floor_half = 0;
                int64_t floor_before = 0;
                int32_t __flow_step_11 = 1;
                for (int32_t t = start; (start <= end) ? t < end : t > end; t += (start <= end) ? 1 : -1) {
                    int64_t dd = ds[t];
                    int64_t mmu = ((int64_t)(ms[t]));
                    coprime_before = (coprime_before + (mmu * FLOW_CHECKED_DIV((before), (dd))));
                    int64_t n1 = FLOW_CHECKED_DIV((half_q), (dd));
                    if (n1 > 0) {
                        floor_half = (floor_half + (mmu * floor_sum_i64_i64_i64_i64((n1 + 1), FLOW_CHECKED_DIV((q), (dd)), g, 0)));
                    }
                    int64_t n2 = FLOW_CHECKED_DIV((before), (dd));
                    if (n2 > 0) {
                        floor_before = (floor_before + (mmu * floor_sum_i64_i64_i64_i64((n2 + 1), FLOW_CHECKED_DIV((q), (dd)), g, 0)));
                    }
                }
                int64_t reduced_count = (reduced_up_to_half - coprime_before);
                int64_t floor_part = (floor_half - floor_before);
                count = ((count + floor_part) - ((min_h - 1) * reduced_count));
            }
        }
    }
    printf("%lld\n", count);
    free(mu);
    free(is_comp);
    free(primes);
    free(offs);
    free(ds);
    free(ms);
    free(cur);
    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 @floor_sum(%arg0: i64, %arg1: i64, %arg2: i64, %arg3: 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 = llvm.mlir.constant(1 : i64) : i64
    %5 = llvm.alloca %4 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg0, %5 : i64, !llvm.ptr
    %6 = llvm.mlir.constant(1 : i64) : i64
    %7 = llvm.alloca %6 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg1, %7 : i64, !llvm.ptr
    %8 = llvm.mlir.constant(1 : i64) : i64
    %9 = llvm.alloca %8 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg2, %9 : i64, !llvm.ptr
    %10 = llvm.mlir.constant(1 : i64) : i64
    %11 = llvm.alloca %10 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg3, %11 : i64, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %12 = arith.constant 1 : i1
    cf.cond_br %12, ^bb1, ^bb2
    ^bb1:
      %13 = llvm.load %9 : !llvm.ptr -> i64
      %14 = llvm.load %7 : !llvm.ptr -> i64
      %15 = arith.cmpi sge, %13, %14 : i64
      cf.cond_br %15, ^bb3, ^bb4
      ^bb3:
        %16 = llvm.load %3 : !llvm.ptr -> i64
        %17 = llvm.load %5 : !llvm.ptr -> i64
        %18 = arith.constant 1 : i32
        %20 = arith.extsi %18 : i32 to i64
        %19 = arith.subi %17, %20 : i64
        %21 = llvm.load %5 : !llvm.ptr -> i64
        %22 = arith.muli %19, %21 : i64
        %23 = llvm.load %9 : !llvm.ptr -> i64
        %24 = llvm.load %7 : !llvm.ptr -> i64
        %25 = arith.divsi %23, %24 : i64
        %26 = arith.muli %22, %25 : i64
        %27 = arith.constant 2 : i32
        %29 = arith.extsi %27 : i32 to i64
        %28 = arith.divsi %26, %29 : i64
        %30 = arith.addi %16, %28 : i64
        llvm.store %30, %3 : i64, !llvm.ptr
        %31 = llvm.load %9 : !llvm.ptr -> i64
        %32 = llvm.load %7 : !llvm.ptr -> i64
        %33 = arith.remsi %31, %32 : i64
        llvm.store %33, %9 : i64, !llvm.ptr
        cf.br ^bb5
      ^bb4:
        cf.br ^bb5
      ^bb5:
      %34 = llvm.load %11 : !llvm.ptr -> i64
      %35 = llvm.load %7 : !llvm.ptr -> i64
      %36 = arith.cmpi sge, %34, %35 : i64
      cf.cond_br %36, ^bb6, ^bb7
      ^bb6:
        %37 = llvm.load %3 : !llvm.ptr -> i64
        %38 = llvm.load %5 : !llvm.ptr -> i64
        %39 = llvm.load %11 : !llvm.ptr -> i64
        %40 = llvm.load %7 : !llvm.ptr -> i64
        %41 = arith.divsi %39, %40 : i64
        %42 = arith.muli %38, %41 : i64
        %43 = arith.addi %37, %42 : i64
        llvm.store %43, %3 : i64, !llvm.ptr
        %44 = llvm.load %11 : !llvm.ptr -> i64
        %45 = llvm.load %7 : !llvm.ptr -> i64
        %46 = arith.remsi %44, %45 : i64
        llvm.store %46, %11 : i64, !llvm.ptr
        cf.br ^bb8
      ^bb7:
        cf.br ^bb8
      ^bb8:
      %47 = llvm.load %9 : !llvm.ptr -> i64
      %48 = llvm.load %5 : !llvm.ptr -> i64
      %49 = arith.muli %47, %48 : i64
      %50 = llvm.load %11 : !llvm.ptr -> i64
      %51 = arith.addi %49, %50 : i64
      %52 = llvm.load %7 : !llvm.ptr -> i64
      %53 = arith.cmpi slt, %51, %52 : i64
      cf.cond_br %53, ^bb9, ^bb10
      ^bb9:
        %54 = llvm.load %3 : !llvm.ptr -> i64
        func.return %54 : i64
      ^bb10:
        cf.br ^bb11
      ^bb11:
      %55 = llvm.load %7 : !llvm.ptr -> i64
      %56 = arith.divsi %51, %55 : i64
      llvm.store %56, %5 : i64, !llvm.ptr
      %57 = llvm.load %7 : !llvm.ptr -> i64
      %58 = arith.remsi %51, %57 : i64
      llvm.store %58, %11 : i64, !llvm.ptr
      %59 = llvm.load %7 : !llvm.ptr -> i64
      %60 = llvm.load %9 : !llvm.ptr -> i64
      llvm.store %60, %7 : i64, !llvm.ptr
      llvm.store %59, %9 : i64, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %61 = llvm.load %3 : !llvm.ptr -> i64
    func.return %61 : i64
  }
  func.func @main() -> i32 {
    %62 = arith.constant 100000 : i32
    %63 = arith.extsi %62 : i32 to i64
    %64 = arith.constant 3 : i32
    %66 = arith.extsi %64 : i32 to i64
    %65 = arith.divsi %63, %66 : i64
    %68 = arith.constant 1 : i32
    %70 = arith.extsi %68 : i32 to i64
    %69 = arith.addi %65, %70 : i64
    %71 = arith.constant 4 : i32
    %72 = arith.extsi %71 : i32 to i64
    %67 = func.call @calloc(%69, %72) : (i64, i64) -> !llvm.ptr
    %74 = arith.constant 1 : i32
    %76 = arith.extsi %74 : i32 to i64
    %75 = arith.addi %65, %76 : i64
    %77 = arith.constant 1 : i32
    %78 = arith.extsi %77 : i32 to i64
    %73 = func.call @calloc(%75, %78) : (i64, i64) -> !llvm.ptr
    %80 = arith.constant 5 : i32
    %82 = arith.extsi %80 : i32 to i64
    %81 = arith.divsi %65, %82 : i64
    %83 = arith.constant 10 : i32
    %85 = arith.extsi %83 : i32 to i64
    %84 = arith.addi %81, %85 : i64
    %86 = arith.constant 8 : i32
    %87 = arith.extsi %86 : i32 to i64
    %79 = func.call @calloc(%84, %87) : (i64, i64) -> !llvm.ptr
    %88 = arith.constant 1 : i32
    %89 = arith.constant 1 : i32
    %90 = arith.extsi %89 : i32 to i64
    %91 = llvm.getelementptr %67[%90] : (!llvm.ptr, i64) -> !llvm.ptr, i32
    llvm.store %88, %91 : i32, !llvm.ptr
    %92 = arith.constant 0 : i32
    %93 = arith.extsi %92 : i32 to i64
    %94 = llvm.mlir.constant(1 : i64) : i64
    %95 = llvm.alloca %94 x i64 : (i64) -> !llvm.ptr
    llvm.store %93, %95 : i64, !llvm.ptr
    %96 = arith.constant 2 : i32
    %97 = arith.constant 1 : i32
    %99 = arith.extsi %97 : i32 to i64
    %98 = arith.addi %65, %99 : i64
    %100 = arith.index_cast %96 : i32 to index
    %101 = arith.index_cast %98 : i32 to index
    %103 = arith.constant 1 : index
    %104 = arith.constant -1 : index
    %105 = arith.cmpi sle, %100, %101 : index
    %102 = arith.select %105, %103, %104 : index
    cf.br ^bb12(%100 : index)
    ^bb12(%106: index):
    %107 = arith.cmpi slt, %106, %101 : index
    %108 = arith.cmpi sgt, %106, %101 : index
    %109 = arith.select %105, %107, %108 : i1
    cf.cond_br %109, ^bb13(%106 : index), ^bb14(%106 : index)
    ^bb13(%110: index):
      %112 = arith.index_cast %110 : index to i64
      %113 = llvm.getelementptr %73[%112] : (!llvm.ptr, i64) -> !llvm.ptr, i8
      %111 = llvm.load %113 : !llvm.ptr -> i8
      %114 = arith.constant 0 : i32
      %116 = arith.extsi %111 : i8 to i32
      %115 = arith.cmpi eq, %116, %114 : i32
      cf.cond_br %115, ^bb15, ^bb16
      ^bb15:
        %117 = llvm.load %95 : !llvm.ptr -> i64
        %118 = arith.index_cast %110 : index to i64
        %119 = llvm.getelementptr %79[%117] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        llvm.store %118, %119 : i64, !llvm.ptr
        %120 = llvm.load %95 : !llvm.ptr -> i64
        %121 = arith.constant 1 : i32
        %123 = arith.extsi %121 : i32 to i64
        %122 = arith.addi %120, %123 : i64
        llvm.store %122, %95 : i64, !llvm.ptr
        %124 = arith.constant 1 : i32
        %126 = arith.constant 0 : i32
        %125 = arith.subi %126, %124 : i32
        %127 = arith.index_cast %110 : index to i64
        %128 = llvm.getelementptr %67[%127] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %125, %128 : i32, !llvm.ptr
        cf.br ^bb17
      ^bb16:
        cf.br ^bb17
      ^bb17:
      %130 = arith.index_cast %110 : index to i64
      %131 = llvm.getelementptr %67[%130] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %129 = llvm.load %131 : !llvm.ptr -> i32
      %132 = arith.constant 0 : i32
      %133 = arith.extsi %132 : i32 to i64
      %134 = llvm.mlir.constant(1 : i64) : i64
      %135 = llvm.alloca %134 x i64 : (i64) -> !llvm.ptr
      llvm.store %133, %135 : i64, !llvm.ptr
      cf.br ^bb18
      ^bb18:
      %136 = llvm.load %135 : !llvm.ptr -> i64
      %137 = llvm.load %95 : !llvm.ptr -> i64
      %138 = arith.cmpi slt, %136, %137 : i64
      cf.cond_br %138, ^bb19, ^bb20
      ^bb19:
        %140 = llvm.load %135 : !llvm.ptr -> i64
        %141 = llvm.getelementptr %79[%140] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %139 = llvm.load %141 : !llvm.ptr -> i64
        %143 = arith.trunci %65 : i64 to i32
        %144 = arith.index_cast %110 : index to i32
        %142 = arith.divsi %143, %144 : i32
        %146 = arith.extsi %142 : i32 to i64
        %145 = arith.cmpi sgt, %139, %146 : i64
        cf.cond_br %145, ^bb21, ^bb22
        ^bb21:
          cf.br ^bb20
        ^bb22:
          cf.br ^bb23
        ^bb23:
        %148 = arith.index_cast %110 : index to i32
        %149 = arith.trunci %139 : i64 to i32
        %147 = arith.muli %148, %149 : i32
        %150 = arith.extsi %147 : i32 to i64
        %151 = arith.constant 1 : i32
        %152 = arith.trunci %151 : i32 to i8
        %153 = llvm.getelementptr %73[%150] : (!llvm.ptr, i64) -> !llvm.ptr, i8
        llvm.store %152, %153 : i8, !llvm.ptr
        %155 = arith.index_cast %110 : index to i32
        %156 = arith.trunci %139 : i64 to i32
        %154 = arith.remsi %155, %156 : i32
        %157 = arith.constant 0 : i32
        %158 = arith.cmpi eq, %154, %157 : i32
        cf.cond_br %158, ^bb24, ^bb25
        ^bb24:
          %159 = arith.constant 0 : i32
          %160 = llvm.getelementptr %67[%150] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          llvm.store %159, %160 : i32, !llvm.ptr
          cf.br ^bb20
        ^bb25:
          cf.br ^bb26
        ^bb26:
        %161 = arith.constant 0 : i32
        %162 = arith.subi %161, %129 : i32
        %163 = llvm.getelementptr %67[%150] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %162, %163 : i32, !llvm.ptr
        %164 = llvm.load %135 : !llvm.ptr -> i64
        %165 = arith.constant 1 : i32
        %167 = arith.extsi %165 : i32 to i64
        %166 = arith.addi %164, %167 : i64
        llvm.store %166, %135 : i64, !llvm.ptr
        cf.br ^bb18
      ^bb20:
      %168 = arith.addi %110, %102 : index
      cf.br ^bb12(%168 : index)
    ^bb14(%169: index):
    %171 = arith.constant 2 : i32
    %173 = arith.extsi %171 : i32 to i64
    %172 = arith.addi %65, %173 : i64
    %174 = arith.constant 8 : i32
    %175 = arith.extsi %174 : i32 to i64
    %170 = func.call @calloc(%172, %175) : (i64, i64) -> !llvm.ptr
    %176 = arith.constant 1 : i32
    %177 = arith.constant 1 : i32
    %179 = arith.extsi %177 : i32 to i64
    %178 = arith.addi %65, %179 : i64
    %180 = arith.index_cast %176 : i32 to index
    %181 = arith.index_cast %178 : i32 to index
    %183 = arith.constant 1 : index
    %184 = arith.constant -1 : index
    %185 = arith.cmpi sle, %180, %181 : index
    %182 = arith.select %185, %183, %184 : index
    cf.br ^bb27(%180 : index)
    ^bb27(%186: index):
    %187 = arith.cmpi slt, %186, %181 : index
    %188 = arith.cmpi sgt, %186, %181 : index
    %189 = arith.select %185, %187, %188 : i1
    cf.cond_br %189, ^bb28(%186 : index), ^bb29(%186 : index)
    ^bb28(%190: index):
      %192 = arith.index_cast %190 : index to i64
      %193 = llvm.getelementptr %67[%192] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %191 = llvm.load %193 : !llvm.ptr -> i32
      %194 = arith.constant 0 : i32
      %195 = arith.cmpi ne, %191, %194 : i32
      cf.cond_br %195, ^bb30, ^bb31
      ^bb30:
        %196 = arith.constant 1 : i32
        %198 = arith.extsi %196 : i32 to i64
        %197 = arith.addi %65, %198 : i64
        %199 = arith.index_cast %190 : i32 to index
        %200 = arith.index_cast %197 : i32 to index
        %201 = arith.index_cast %190 : i32 to index
        %202 = arith.constant 0 : index
        %203 = arith.cmpi sgt, %201, %202 : index
        cf.br ^bb33(%199 : index)
        ^bb33(%204: index):
        %205 = arith.cmpi slt, %204, %200 : index
        %206 = arith.cmpi sgt, %204, %200 : index
        %207 = arith.select %203, %205, %206 : i1
        cf.cond_br %207, ^bb34(%204 : index), ^bb35(%204 : index)
        ^bb34(%208: index):
          %210 = arith.constant 1 : i32
          %212 = arith.index_cast %208 : index to i32
          %211 = arith.addi %212, %210 : i32
          %213 = arith.extsi %211 : i32 to i64
          %214 = llvm.getelementptr %170[%213] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %209 = llvm.load %214 : !llvm.ptr -> i64
          %215 = arith.constant 1 : i32
          %217 = arith.extsi %215 : i32 to i64
          %216 = arith.addi %209, %217 : i64
          %218 = arith.constant 1 : i32
          %220 = arith.index_cast %208 : index to i32
          %219 = arith.addi %220, %218 : i32
          %221 = arith.extsi %219 : i32 to i64
          %222 = llvm.getelementptr %170[%221] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          llvm.store %216, %222 : i64, !llvm.ptr
          %223 = arith.addi %208, %201 : index
          cf.br ^bb33(%223 : index)
        ^bb35(%224: index):
        cf.br ^bb32
      ^bb31:
        cf.br ^bb32
      ^bb32:
      %225 = arith.addi %190, %182 : index
      cf.br ^bb27(%225 : index)
    ^bb29(%226: index):
    %227 = arith.constant 1 : i32
    %228 = arith.constant 2 : i32
    %230 = arith.extsi %228 : i32 to i64
    %229 = arith.addi %65, %230 : i64
    %231 = arith.index_cast %227 : i32 to index
    %232 = arith.index_cast %229 : i32 to index
    %234 = arith.constant 1 : index
    %235 = arith.constant -1 : index
    %236 = arith.cmpi sle, %231, %232 : index
    %233 = arith.select %236, %234, %235 : index
    cf.br ^bb36(%231 : index)
    ^bb36(%237: index):
    %238 = arith.cmpi slt, %237, %232 : index
    %239 = arith.cmpi sgt, %237, %232 : index
    %240 = arith.select %236, %238, %239 : i1
    cf.cond_br %240, ^bb37(%237 : index), ^bb38(%237 : index)
    ^bb37(%241: index):
      %243 = arith.index_cast %241 : index to i64
      %244 = llvm.getelementptr %170[%243] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %242 = llvm.load %244 : !llvm.ptr -> i64
      %246 = arith.constant 1 : i32
      %248 = arith.index_cast %241 : index to i32
      %247 = arith.subi %248, %246 : i32
      %249 = arith.extsi %247 : i32 to i64
      %250 = llvm.getelementptr %170[%249] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %245 = llvm.load %250 : !llvm.ptr -> i64
      %251 = arith.addi %242, %245 : i64
      %252 = arith.index_cast %241 : index to i64
      %253 = llvm.getelementptr %170[%252] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      llvm.store %251, %253 : i64, !llvm.ptr
      %254 = arith.addi %241, %233 : index
      cf.br ^bb36(%254 : index)
    ^bb38(%255: index):
    %257 = arith.constant 1 : i32
    %259 = arith.extsi %257 : i32 to i64
    %258 = arith.addi %65, %259 : i64
    %260 = llvm.getelementptr %170[%258] : (!llvm.ptr, i64) -> !llvm.ptr, i64
    %256 = llvm.load %260 : !llvm.ptr -> i64
    %262 = arith.constant 8 : i32
    %263 = arith.extsi %262 : i32 to i64
    %261 = func.call @calloc(%256, %263) : (i64, i64) -> !llvm.ptr
    %265 = arith.constant 4 : i32
    %266 = arith.extsi %265 : i32 to i64
    %264 = func.call @calloc(%256, %266) : (i64, i64) -> !llvm.ptr
    %268 = arith.constant 1 : i32
    %270 = arith.extsi %268 : i32 to i64
    %269 = arith.addi %65, %270 : i64
    %271 = arith.constant 8 : i32
    %272 = arith.extsi %271 : i32 to i64
    %267 = func.call @calloc(%269, %272) : (i64, i64) -> !llvm.ptr
    %273 = arith.constant 0 : i32
    %274 = arith.constant 1 : i32
    %276 = arith.extsi %274 : i32 to i64
    %275 = arith.addi %65, %276 : i64
    %277 = arith.index_cast %273 : i32 to index
    %278 = arith.index_cast %275 : i32 to index
    %280 = arith.constant 1 : index
    %281 = arith.constant -1 : index
    %282 = arith.cmpi sle, %277, %278 : index
    %279 = arith.select %282, %280, %281 : index
    cf.br ^bb39(%277 : index)
    ^bb39(%283: index):
    %284 = arith.cmpi slt, %283, %278 : index
    %285 = arith.cmpi sgt, %283, %278 : index
    %286 = arith.select %282, %284, %285 : i1
    cf.cond_br %286, ^bb40(%283 : index), ^bb41(%283 : index)
    ^bb40(%287: index):
      %289 = arith.index_cast %287 : index to i64
      %290 = llvm.getelementptr %170[%289] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %288 = llvm.load %290 : !llvm.ptr -> i64
      %291 = arith.index_cast %287 : index to i64
      %292 = llvm.getelementptr %267[%291] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      llvm.store %288, %292 : i64, !llvm.ptr
      %293 = arith.addi %287, %279 : index
      cf.br ^bb39(%293 : index)
    ^bb41(%294: index):
    %295 = arith.constant 1 : i32
    %296 = arith.constant 1 : i32
    %298 = arith.extsi %296 : i32 to i64
    %297 = arith.addi %65, %298 : i64
    %299 = arith.index_cast %295 : i32 to index
    %300 = arith.index_cast %297 : i32 to index
    %302 = arith.constant 1 : index
    %303 = arith.constant -1 : index
    %304 = arith.cmpi sle, %299, %300 : index
    %301 = arith.select %304, %302, %303 : index
    cf.br ^bb42(%299 : index)
    ^bb42(%305: index):
    %306 = arith.cmpi slt, %305, %300 : index
    %307 = arith.cmpi sgt, %305, %300 : index
    %308 = arith.select %304, %306, %307 : i1
    cf.cond_br %308, ^bb43(%305 : index), ^bb44(%305 : index)
    ^bb43(%309: index):
      %311 = arith.index_cast %309 : index to i64
      %312 = llvm.getelementptr %67[%311] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %310 = llvm.load %312 : !llvm.ptr -> i32
      %313 = arith.constant 0 : i32
      %314 = arith.cmpi ne, %310, %313 : i32
      cf.cond_br %314, ^bb45, ^bb46
      ^bb45:
        %315 = arith.constant 1 : i32
        %317 = arith.extsi %315 : i32 to i64
        %316 = arith.addi %65, %317 : i64
        %318 = arith.index_cast %309 : i32 to index
        %319 = arith.index_cast %316 : i32 to index
        %320 = arith.index_cast %309 : i32 to index
        %321 = arith.constant 0 : index
        %322 = arith.cmpi sgt, %320, %321 : index
        cf.br ^bb48(%318 : index)
        ^bb48(%323: index):
        %324 = arith.cmpi slt, %323, %319 : index
        %325 = arith.cmpi sgt, %323, %319 : index
        %326 = arith.select %322, %324, %325 : i1
        cf.cond_br %326, ^bb49(%323 : index), ^bb50(%323 : index)
        ^bb49(%327: index):
          %329 = arith.index_cast %327 : index to i64
          %330 = llvm.getelementptr %267[%329] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %328 = llvm.load %330 : !llvm.ptr -> i64
          %331 = arith.index_cast %309 : index to i64
          %332 = llvm.getelementptr %261[%328] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          llvm.store %331, %332 : i64, !llvm.ptr
          %334 = arith.index_cast %309 : index to i64
          %335 = llvm.getelementptr %67[%334] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          %333 = llvm.load %335 : !llvm.ptr -> i32
          %336 = llvm.getelementptr %264[%328] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          llvm.store %333, %336 : i32, !llvm.ptr
          %337 = arith.constant 1 : i32
          %339 = arith.extsi %337 : i32 to i64
          %338 = arith.addi %328, %339 : i64
          %340 = arith.index_cast %327 : index to i64
          %341 = llvm.getelementptr %267[%340] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          llvm.store %338, %341 : i64, !llvm.ptr
          %342 = arith.addi %327, %320 : index
          cf.br ^bb48(%342 : index)
        ^bb50(%343: index):
        cf.br ^bb47
      ^bb46:
        cf.br ^bb47
      ^bb47:
      %344 = arith.addi %309, %301 : index
      cf.br ^bb42(%344 : index)
    ^bb44(%345: index):
    %346 = arith.constant 0 : i32
    %347 = arith.extsi %346 : i32 to i64
    %348 = llvm.mlir.constant(1 : i64) : i64
    %349 = llvm.alloca %348 x i64 : (i64) -> !llvm.ptr
    llvm.store %347, %349 : i64, !llvm.ptr
    %350 = arith.constant 2 : i32
    %351 = arith.constant 1 : i32
    %353 = arith.extsi %351 : i32 to i64
    %352 = arith.addi %65, %353 : i64
    %354 = arith.index_cast %350 : i32 to index
    %355 = arith.index_cast %352 : i32 to index
    %357 = arith.constant 1 : index
    %358 = arith.constant -1 : index
    %359 = arith.cmpi sle, %354, %355 : index
    %356 = arith.select %359, %357, %358 : index
    cf.br ^bb51(%354 : index)
    ^bb51(%360: index):
    %361 = arith.cmpi slt, %360, %355 : index
    %362 = arith.cmpi sgt, %360, %355 : index
    %363 = arith.select %359, %361, %362 : i1
    cf.cond_br %363, ^bb52(%360 : index), ^bb53(%360 : index)
    ^bb52(%364: index):
      %366 = arith.trunci %63 : i64 to i32
      %367 = arith.index_cast %364 : index to i32
      %365 = arith.divsi %366, %367 : i32
      %368 = arith.extsi %365 : i32 to i64
      %369 = arith.constant 2 : i32
      %371 = arith.index_cast %364 : index to i32
      %370 = arith.divsi %371, %369 : i32
      %372 = arith.extsi %370 : i32 to i64
      %374 = arith.index_cast %364 : index to i64
      %375 = llvm.getelementptr %170[%374] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %373 = llvm.load %375 : !llvm.ptr -> i64
      %377 = arith.constant 1 : i32
      %379 = arith.index_cast %364 : index to i32
      %378 = arith.addi %379, %377 : i32
      %380 = arith.extsi %378 : i32 to i64
      %381 = llvm.getelementptr %170[%380] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %376 = llvm.load %381 : !llvm.ptr -> i64
      %382 = arith.constant 0 : i32
      %383 = arith.extsi %382 : i32 to i64
      %384 = llvm.mlir.constant(1 : i64) : i64
      %385 = llvm.alloca %384 x i64 : (i64) -> !llvm.ptr
      llvm.store %383, %385 : i64, !llvm.ptr
      %386 = arith.index_cast %373 : i32 to index
      %387 = arith.index_cast %376 : i32 to index
      %389 = arith.constant 1 : index
      %390 = arith.constant -1 : index
      %391 = arith.cmpi sle, %386, %387 : index
      %388 = arith.select %391, %389, %390 : index
      cf.br ^bb54(%386 : index)
      ^bb54(%392: index):
      %393 = arith.cmpi slt, %392, %387 : index
      %394 = arith.cmpi sgt, %392, %387 : index
      %395 = arith.select %391, %393, %394 : i1
      cf.cond_br %395, ^bb55(%392 : index), ^bb56(%392 : index)
      ^bb55(%396: index):
        %397 = llvm.load %385 : !llvm.ptr -> i64
        %399 = arith.index_cast %396 : index to i64
        %400 = llvm.getelementptr %264[%399] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %398 = llvm.load %400 : !llvm.ptr -> i32
        %401 = arith.extsi %398 : i32 to i64
        %403 = arith.index_cast %396 : index to i64
        %404 = llvm.getelementptr %261[%403] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %402 = llvm.load %404 : !llvm.ptr -> i64
        %405 = arith.divsi %372, %402 : i64
        %406 = arith.muli %401, %405 : i64
        %407 = arith.addi %397, %406 : i64
        llvm.store %407, %385 : i64, !llvm.ptr
        %408 = arith.addi %396, %388 : index
        cf.br ^bb54(%408 : index)
      ^bb56(%409: index):
      %410 = arith.constant 2 : i32
      %411 = arith.index_cast %410 : i32 to index
      %412 = arith.index_cast %368 : i32 to index
      %414 = arith.constant 1 : index
      %415 = arith.constant -1 : index
      %416 = arith.cmpi sle, %411, %412 : index
      %413 = arith.select %416, %414, %415 : index
      cf.br ^bb57(%411 : index)
      ^bb57(%417: index):
      %418 = arith.cmpi slt, %417, %412 : index
      %419 = arith.cmpi sgt, %417, %412 : index
      %420 = arith.select %416, %418, %419 : i1
      cf.cond_br %420, ^bb58(%417 : index), ^bb59(%417 : index)
      ^bb58(%421: index):
        %422 = arith.constant 1 : i32
        %423 = arith.extsi %422 : i32 to i64
        %424 = llvm.mlir.constant(1 : i64) : i64
        %425 = llvm.alloca %424 x i64 : (i64) -> !llvm.ptr
        llvm.store %423, %425 : i64, !llvm.ptr
        %426 = arith.constant 2 : i32
        %428 = arith.index_cast %421 : index to i32
        %427 = arith.muli %426, %428 : i32
        %430 = arith.extsi %427 : i32 to i64
        %429 = arith.subi %430, %368 : i64
        %431 = llvm.load %425 : !llvm.ptr -> i64
        %432 = arith.cmpi sgt, %429, %431 : i64
        cf.cond_br %432, ^bb60, ^bb61
        ^bb60:
          llvm.store %429, %425 : i64, !llvm.ptr
          cf.br ^bb62
        ^bb61:
          cf.br ^bb62
        ^bb62:
        %433 = llvm.load %425 : !llvm.ptr -> i64
        %435 = arith.trunci %433 : i64 to i32
        %436 = arith.index_cast %364 : index to i32
        %434 = arith.muli %435, %436 : i32
        %438 = arith.index_cast %421 : index to i32
        %437 = arith.addi %434, %438 : i32
        %439 = arith.constant 1 : i32
        %440 = arith.subi %437, %439 : i32
        %442 = arith.index_cast %421 : index to i32
        %441 = arith.divsi %440, %442 : i32
        %443 = arith.constant 1 : i32
        %444 = arith.subi %441, %443 : i32
        %445 = arith.extsi %444 : i32 to i64
        %446 = arith.cmpi slt, %445, %372 : i64
        cf.cond_br %446, ^bb63, ^bb64
        ^bb63:
          %447 = arith.constant 0 : i32
          %448 = arith.extsi %447 : i32 to i64
          %449 = llvm.mlir.constant(1 : i64) : i64
          %450 = llvm.alloca %449 x i64 : (i64) -> !llvm.ptr
          llvm.store %448, %450 : i64, !llvm.ptr
          %451 = arith.constant 0 : i32
          %452 = arith.extsi %451 : i32 to i64
          %453 = llvm.mlir.constant(1 : i64) : i64
          %454 = llvm.alloca %453 x i64 : (i64) -> !llvm.ptr
          llvm.store %452, %454 : i64, !llvm.ptr
          %455 = arith.constant 0 : i32
          %456 = arith.extsi %455 : i32 to i64
          %457 = llvm.mlir.constant(1 : i64) : i64
          %458 = llvm.alloca %457 x i64 : (i64) -> !llvm.ptr
          llvm.store %456, %458 : i64, !llvm.ptr
          %459 = arith.index_cast %373 : i32 to index
          %460 = arith.index_cast %376 : i32 to index
          %462 = arith.constant 1 : index
          %463 = arith.constant -1 : index
          %464 = arith.cmpi sle, %459, %460 : index
          %461 = arith.select %464, %462, %463 : index
          cf.br ^bb66(%459 : index)
          ^bb66(%465: index):
          %466 = arith.cmpi slt, %465, %460 : index
          %467 = arith.cmpi sgt, %465, %460 : index
          %468 = arith.select %464, %466, %467 : i1
          cf.cond_br %468, ^bb67(%465 : index), ^bb68(%465 : index)
          ^bb67(%469: index):
            %471 = arith.index_cast %469 : index to i64
            %472 = llvm.getelementptr %261[%471] : (!llvm.ptr, i64) -> !llvm.ptr, i64
            %470 = llvm.load %472 : !llvm.ptr -> i64
            %474 = arith.index_cast %469 : index to i64
            %475 = llvm.getelementptr %264[%474] : (!llvm.ptr, i64) -> !llvm.ptr, i32
            %473 = llvm.load %475 : !llvm.ptr -> i32
            %476 = arith.extsi %473 : i32 to i64
            %477 = llvm.load %450 : !llvm.ptr -> i64
            %478 = arith.divsi %445, %470 : i64
            %479 = arith.muli %476, %478 : i64
            %480 = arith.addi %477, %479 : i64
            llvm.store %480, %450 : i64, !llvm.ptr
            %481 = arith.divsi %372, %470 : i64
            %482 = arith.constant 0 : i32
            %484 = arith.extsi %482 : i32 to i64
            %483 = arith.cmpi sgt, %481, %484 : i64
            cf.cond_br %483, ^bb69, ^bb70
            ^bb69:
              %485 = llvm.load %454 : !llvm.ptr -> i64
              %487 = arith.constant 1 : i32
              %489 = arith.extsi %487 : i32 to i64
              %488 = arith.addi %481, %489 : i64
              %491 = arith.index_cast %364 : index to i32
              %492 = arith.trunci %470 : i64 to i32
              %490 = arith.divsi %491, %492 : i32
              %493 = arith.constant 0 : i32
              %494 = arith.extsi %490 : i32 to i64
              %495 = arith.index_cast %421 : index to i64
              %496 = arith.extsi %493 : i32 to i64
              %486 = func.call @floor_sum(%488, %494, %495, %496) : (i64, i64, i64, i64) -> i64
              %497 = arith.muli %476, %486 : i64
              %498 = arith.addi %485, %497 : i64
              llvm.store %498, %454 : i64, !llvm.ptr
              cf.br ^bb71
            ^bb70:
              cf.br ^bb71
            ^bb71:
            %499 = arith.divsi %445, %470 : i64
            %500 = arith.constant 0 : i32
            %502 = arith.extsi %500 : i32 to i64
            %501 = arith.cmpi sgt, %499, %502 : i64
            cf.cond_br %501, ^bb72, ^bb73
            ^bb72:
              %503 = llvm.load %458 : !llvm.ptr -> i64
              %505 = arith.constant 1 : i32
              %507 = arith.extsi %505 : i32 to i64
              %506 = arith.addi %499, %507 : i64
              %509 = arith.index_cast %364 : index to i32
              %510 = arith.trunci %470 : i64 to i32
              %508 = arith.divsi %509, %510 : i32
              %511 = arith.constant 0 : i32
              %512 = arith.extsi %508 : i32 to i64
              %513 = arith.index_cast %421 : index to i64
              %514 = arith.extsi %511 : i32 to i64
              %504 = func.call @floor_sum(%506, %512, %513, %514) : (i64, i64, i64, i64) -> i64
              %515 = arith.muli %476, %504 : i64
              %516 = arith.addi %503, %515 : i64
              llvm.store %516, %458 : i64, !llvm.ptr
              cf.br ^bb74
            ^bb73:
              cf.br ^bb74
            ^bb74:
            %517 = arith.addi %469, %461 : index
            cf.br ^bb66(%517 : index)
          ^bb68(%518: index):
          %519 = llvm.load %385 : !llvm.ptr -> i64
          %520 = llvm.load %450 : !llvm.ptr -> i64
          %521 = arith.subi %519, %520 : i64
          %522 = llvm.load %454 : !llvm.ptr -> i64
          %523 = llvm.load %458 : !llvm.ptr -> i64
          %524 = arith.subi %522, %523 : i64
          %525 = llvm.load %349 : !llvm.ptr -> i64
          %526 = arith.addi %525, %524 : i64
          %527 = llvm.load %425 : !llvm.ptr -> i64
          %528 = arith.constant 1 : i32
          %530 = arith.extsi %528 : i32 to i64
          %529 = arith.subi %527, %530 : i64
          %531 = arith.muli %529, %521 : i64
          %532 = arith.subi %526, %531 : i64
          llvm.store %532, %349 : i64, !llvm.ptr
          cf.br ^bb65
        ^bb64:
          cf.br ^bb65
        ^bb65:
        %533 = arith.addi %421, %413 : index
        cf.br ^bb57(%533 : index)
      ^bb59(%534: index):
      %535 = arith.addi %364, %356 : index
      cf.br ^bb51(%535 : index)
    ^bb53(%536: index):
    %537 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %538 = llvm.load %349 : !llvm.ptr -> i64
    %539 = llvm.call @printf(%537, %538) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    func.call @free(%67) : (!llvm.ptr) -> ()
    func.call @free(%73) : (!llvm.ptr) -> ()
    func.call @free(%79) : (!llvm.ptr) -> ()
    func.call @free(%170) : (!llvm.ptr) -> ()
    func.call @free(%261) : (!llvm.ptr) -> ()
    func.call @free(%264) : (!llvm.ptr) -> ()
    func.call @free(%267) : (!llvm.ptr) -> ()
    %547 = arith.constant 0 : i32
    func.return %547 : i32
  }
}