Problem 887

Bounded Binary Search — sum Q(N,d) for d=0..7, N<=7^10.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 887
# Bounded Binary Search — sum Q(N,d) for d=0..7, N<=7^10.

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

# Memo: key = q*16 + (d+8) since d can be small negatives; use hash
const HSIZE: i64 = 1 << 20
let mut HKEY: ptr<i64> = null
let mut HVAL: ptr<i64> = null
let mut HUSE: ptr<i8> = null

function pack(q: i64, d: i64) -> i64 {
    return (q << 32) ^ (d + 1000000000)
}

function max_n(q: i64, d: i64) -> i64 {
    if d < 0 { return 1 }
    if q == 0 { return 1 }
    if d >= q - 1 {
        return 1 << q
    }
    let key: i64 = pack(q, d)
    let mut h: i64 = (key * 11400714819323198485) & (HSIZE - 1)
    while HUSE[h] != 0 {
        if HKEY[h] == key { return HVAL[h] }
        h = (h + 1) & (HSIZE - 1)
    }
    let left: i64 = max_n(q - 1, d - 1)
    let ans: i64 = left + max_n(q - 1, left + d - 1)
    HUSE[h] = 1
    HKEY[h] = key
    HVAL[h] = ans
    return ans
}

function sum_Q_upto(limit: i64, d: i64) -> i64 {
    if limit <= 0 { return 0 }
    if d == 0 {
        return limit * (limit - 1) / 2
    }
    let mut total: i64 = 0
    let mut prev_cap: i64 = 0
    let mut q: i64 = 0
    while prev_cap < limit {
        let mut cap: i64 = max_n(q, d)
        if cap > limit { cap = limit }
        total = total + q * (cap - prev_cap)
        prev_cap = cap
        q = q + 1
    }
    return total
}

function main() -> i32 {
    HKEY = calloc(HSIZE, 8)
    HVAL = calloc(HSIZE, 8)
    HUSE = calloc(HSIZE, 1)
    let mut limit: i64 = 1
    let mut i: i64 = 0
    while i < 10 {
        limit = limit * 7
        i = i + 1
    }
    let mut ans: i64 = 0
    let mut d: i64 = 0
    while d <= 7 {
        ans = ans + sum_Q_upto(limit, d)
        d = d + 1
    }
    printf("%lld\n", ans)
    free(HKEY)
    free(HVAL)
    free(HUSE)
    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 pack_i64_i64(int64_t q, int64_t d);
int64_t max_n_i64_i64(int64_t q, int64_t d);
int64_t sum_Q_upto_i64_i64(int64_t limit, int64_t d);
int32_t main(void);

static const int64_t HSIZE = FLOW_CHECKED_SHL((1), (20));

/* Module statics */
static int64_t* HKEY = NULL;
static int64_t* HVAL = NULL;
static int8_t* HUSE = NULL;



int64_t pack_i64_i64(int64_t q, int64_t d) {
    return (FLOW_CHECKED_SHL((q), (32)) ^ (d + 1000000000));
}

int64_t max_n_i64_i64(int64_t q, int64_t d) {
    if (d < 0) {
        return 1;
    }
    if (q == 0) {
        return 1;
    }
    if (d >= (q - 1)) {
        return FLOW_CHECKED_SHL((1), (q));
    }
    int64_t key = pack_i64_i64(q, d);
    int64_t h = ((key * ((__int128)0x9E3779B97F4A7C15ULL)) & (HSIZE - 1));
    while (HUSE[h] != 0) {
        if (HKEY[h] == key) {
            return HVAL[h];
        }
        h = ((h + 1) & (HSIZE - 1));
    }
    int64_t left = max_n_i64_i64((q - 1), (d - 1));
    int64_t ans = (left + max_n_i64_i64((q - 1), ((left + d) - 1)));
    HUSE[h] = 1;
    HKEY[h] = key;
    HVAL[h] = ans;
    return ans;
}

int64_t sum_Q_upto_i64_i64(int64_t limit, int64_t d) {
    if (limit <= 0) {
        return 0;
    }
    if (d == 0) {
        return FLOW_CHECKED_DIV(((limit * (limit - 1))), (2));
    }
    int64_t total = 0;
    int64_t prev_cap = 0;
    int64_t q = 0;
    while (prev_cap < limit) {
        int64_t cap = max_n_i64_i64(q, d);
        if (cap > limit) {
            cap = limit;
        }
        total = (total + (q * (cap - prev_cap)));
        prev_cap = cap;
        q = (q + 1);
    }
    return total;
}

int32_t main(void) {
    HKEY = calloc(HSIZE, 8);
    HVAL = calloc(HSIZE, 8);
    HUSE = calloc(HSIZE, 1);
    int64_t limit = 1;
    int64_t i = 0;
    while (i < 10) {
        limit = (limit * 7);
        i = (i + 1);
    }
    int64_t ans = 0;
    int64_t d = 0;
    while (d <= 7) {
        ans = (ans + sum_Q_upto_i64_i64(limit, d));
        d = (d + 1);
    }
    printf("%lld\n", ans);
    free(HKEY);
    free(HVAL);
    free(HUSE);
    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) -> ()
  // Constant: HSIZE
  llvm.mlir.global internal constant @HSIZE(1048576 : i64) : i64
  // Module static: HKEY
  llvm.mlir.global internal @HKEY() {addr_space = 0 : i32} : !llvm.ptr {
    %0 = llvm.mlir.zero : !llvm.ptr
    llvm.return %0 : !llvm.ptr
  }
  // Module static: HVAL
  llvm.mlir.global internal @HVAL() {addr_space = 0 : i32} : !llvm.ptr {
    %1 = llvm.mlir.zero : !llvm.ptr
    llvm.return %1 : !llvm.ptr
  }
  // Module static: HUSE
  llvm.mlir.global internal @HUSE() {addr_space = 0 : i32} : !llvm.ptr {
    %2 = llvm.mlir.zero : !llvm.ptr
    llvm.return %2 : !llvm.ptr
  }
  func.func @pack(%arg0: i64, %arg1: i64) -> i64 {
    %3 = arith.constant 32 : i32
    %5 = arith.extsi %3 : i32 to i64
    %4 = arith.shli %arg0, %5 : i64
    %6 = arith.constant 1000000000 : i32
    %8 = arith.extsi %6 : i32 to i64
    %7 = arith.addi %arg1, %8 : i64
    %9 = arith.xori %4, %7 : i64
    func.return %9 : i64
  }
  func.func @max_n(%arg0: i64, %arg1: i64) -> i64 {
    %10 = arith.constant 0 : i32
    %12 = arith.extsi %10 : i32 to i64
    %11 = arith.cmpi slt, %arg1, %12 : i64
    cf.cond_br %11, ^bb0, ^bb1
    ^bb0:
      %13 = arith.constant 1 : i32
      %14 = arith.extsi %13 : i32 to i64
      func.return %14 : i64
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %15 = arith.constant 0 : i32
    %17 = arith.extsi %15 : i32 to i64
    %16 = arith.cmpi eq, %arg0, %17 : i64
    cf.cond_br %16, ^bb3, ^bb4
    ^bb3:
      %18 = arith.constant 1 : i32
      %19 = arith.extsi %18 : i32 to i64
      func.return %19 : i64
    ^bb4:
      cf.br ^bb5
    ^bb5:
    %20 = arith.constant 1 : i32
    %22 = arith.extsi %20 : i32 to i64
    %21 = arith.subi %arg0, %22 : i64
    %23 = arith.cmpi sge, %arg1, %21 : i64
    cf.cond_br %23, ^bb6, ^bb7
    ^bb6:
      %24 = arith.constant 1 : i32
      %26 = arith.extsi %24 : i32 to i64
      %25 = arith.shli %26, %arg0 : i64
      func.return %25 : i64
    ^bb7:
      cf.br ^bb8
    ^bb8:
    %27 = func.call @pack(%arg0, %arg1) : (i64, i64) -> i64
    %28 = arith.constant 11400714815028231189 : i32
    %30 = arith.extsi %28 : i32 to i64
    %29 = arith.muli %27, %30 : i64
    %31 = llvm.mlir.addressof @HSIZE : !llvm.ptr
    %32 = llvm.load %31 : !llvm.ptr -> i64
    %33 = arith.constant 1 : i32
    %35 = arith.extsi %33 : i32 to i64
    %34 = arith.subi %32, %35 : i64
    %36 = arith.andi %29, %34 : i64
    %37 = llvm.mlir.constant(1 : i64) : i64
    %38 = llvm.alloca %37 x i64 : (i64) -> !llvm.ptr
    llvm.store %36, %38 : i64, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %40 = llvm.mlir.addressof @HUSE : !llvm.ptr
    %41 = llvm.load %40 : !llvm.ptr -> !llvm.ptr
    %42 = llvm.load %38 : !llvm.ptr -> i64
    %43 = llvm.getelementptr %41[%42] : (!llvm.ptr, i64) -> !llvm.ptr, i8
    %39 = llvm.load %43 : !llvm.ptr -> i8
    %44 = arith.constant 0 : i32
    %46 = arith.extsi %39 : i8 to i32
    %45 = arith.cmpi ne, %46, %44 : i32
    cf.cond_br %45, ^bb10, ^bb11
    ^bb10:
      %48 = llvm.mlir.addressof @HKEY : !llvm.ptr
      %49 = llvm.load %48 : !llvm.ptr -> !llvm.ptr
      %50 = llvm.load %38 : !llvm.ptr -> i64
      %51 = llvm.getelementptr %49[%50] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %47 = llvm.load %51 : !llvm.ptr -> i64
      %52 = arith.cmpi eq, %47, %27 : i64
      cf.cond_br %52, ^bb12, ^bb13
      ^bb12:
        %54 = llvm.mlir.addressof @HVAL : !llvm.ptr
        %55 = llvm.load %54 : !llvm.ptr -> !llvm.ptr
        %56 = llvm.load %38 : !llvm.ptr -> i64
        %57 = llvm.getelementptr %55[%56] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %53 = llvm.load %57 : !llvm.ptr -> i64
        func.return %53 : i64
      ^bb13:
        cf.br ^bb14
      ^bb14:
      %58 = llvm.load %38 : !llvm.ptr -> i64
      %59 = arith.constant 1 : i32
      %61 = arith.extsi %59 : i32 to i64
      %60 = arith.addi %58, %61 : i64
      %62 = llvm.mlir.addressof @HSIZE : !llvm.ptr
      %63 = llvm.load %62 : !llvm.ptr -> i64
      %64 = arith.constant 1 : i32
      %66 = arith.extsi %64 : i32 to i64
      %65 = arith.subi %63, %66 : i64
      %67 = arith.andi %60, %65 : i64
      llvm.store %67, %38 : i64, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    %69 = arith.constant 1 : i32
    %71 = arith.extsi %69 : i32 to i64
    %70 = arith.subi %arg0, %71 : i64
    %72 = arith.constant 1 : i32
    %74 = arith.extsi %72 : i32 to i64
    %73 = arith.subi %arg1, %74 : i64
    %68 = func.call @max_n(%70, %73) : (i64, i64) -> i64
    %76 = arith.constant 1 : i32
    %78 = arith.extsi %76 : i32 to i64
    %77 = arith.subi %arg0, %78 : i64
    %79 = arith.addi %68, %arg1 : i64
    %80 = arith.constant 1 : i32
    %82 = arith.extsi %80 : i32 to i64
    %81 = arith.subi %79, %82 : i64
    %75 = func.call @max_n(%77, %81) : (i64, i64) -> i64
    %83 = arith.addi %68, %75 : i64
    %84 = arith.constant 1 : i32
    %85 = llvm.mlir.addressof @HUSE : !llvm.ptr
    %86 = llvm.load %85 : !llvm.ptr -> !llvm.ptr
    %87 = llvm.load %38 : !llvm.ptr -> i64
    %88 = arith.trunci %84 : i32 to i8
    %89 = llvm.getelementptr %86[%87] : (!llvm.ptr, i64) -> !llvm.ptr, i8
    llvm.store %88, %89 : i8, !llvm.ptr
    %90 = llvm.mlir.addressof @HKEY : !llvm.ptr
    %91 = llvm.load %90 : !llvm.ptr -> !llvm.ptr
    %92 = llvm.load %38 : !llvm.ptr -> i64
    %93 = llvm.getelementptr %91[%92] : (!llvm.ptr, i64) -> !llvm.ptr, i64
    llvm.store %27, %93 : i64, !llvm.ptr
    %94 = llvm.mlir.addressof @HVAL : !llvm.ptr
    %95 = llvm.load %94 : !llvm.ptr -> !llvm.ptr
    %96 = llvm.load %38 : !llvm.ptr -> i64
    %97 = llvm.getelementptr %95[%96] : (!llvm.ptr, i64) -> !llvm.ptr, i64
    llvm.store %83, %97 : i64, !llvm.ptr
    func.return %83 : i64
  }
  func.func @sum_Q_upto(%arg0: i64, %arg1: i64) -> i64 {
    %98 = arith.constant 0 : i32
    %100 = arith.extsi %98 : i32 to i64
    %99 = arith.cmpi sle, %arg0, %100 : i64
    cf.cond_br %99, ^bb15, ^bb16
    ^bb15:
      %101 = arith.constant 0 : i32
      %102 = arith.extsi %101 : i32 to i64
      func.return %102 : i64
    ^bb16:
      cf.br ^bb17
    ^bb17:
    %103 = arith.constant 0 : i32
    %105 = arith.extsi %103 : i32 to i64
    %104 = arith.cmpi eq, %arg1, %105 : i64
    cf.cond_br %104, ^bb18, ^bb19
    ^bb18:
      %106 = arith.constant 1 : i32
      %108 = arith.extsi %106 : i32 to i64
      %107 = arith.subi %arg0, %108 : i64
      %109 = arith.muli %arg0, %107 : i64
      %110 = arith.constant 2 : i32
      %112 = arith.extsi %110 : i32 to i64
      %111 = arith.divsi %109, %112 : i64
      func.return %111 : i64
    ^bb19:
      cf.br ^bb20
    ^bb20:
    %113 = arith.constant 0 : i32
    %114 = arith.extsi %113 : i32 to i64
    %115 = llvm.mlir.constant(1 : i64) : i64
    %116 = llvm.alloca %115 x i64 : (i64) -> !llvm.ptr
    llvm.store %114, %116 : i64, !llvm.ptr
    %117 = arith.constant 0 : i32
    %118 = arith.extsi %117 : i32 to i64
    %119 = llvm.mlir.constant(1 : i64) : i64
    %120 = llvm.alloca %119 x i64 : (i64) -> !llvm.ptr
    llvm.store %118, %120 : i64, !llvm.ptr
    %121 = arith.constant 0 : i32
    %122 = arith.extsi %121 : i32 to i64
    %123 = llvm.mlir.constant(1 : i64) : i64
    %124 = llvm.alloca %123 x i64 : (i64) -> !llvm.ptr
    llvm.store %122, %124 : i64, !llvm.ptr
    cf.br ^bb21
    ^bb21:
    %125 = llvm.load %120 : !llvm.ptr -> i64
    %126 = arith.cmpi slt, %125, %arg0 : i64
    cf.cond_br %126, ^bb22, ^bb23
    ^bb22:
      %128 = llvm.load %124 : !llvm.ptr -> i64
      %127 = func.call @max_n(%128, %arg1) : (i64, i64) -> i64
      %129 = llvm.mlir.constant(1 : i64) : i64
      %130 = llvm.alloca %129 x i64 : (i64) -> !llvm.ptr
      llvm.store %127, %130 : i64, !llvm.ptr
      %131 = llvm.load %130 : !llvm.ptr -> i64
      %132 = arith.cmpi sgt, %131, %arg0 : i64
      cf.cond_br %132, ^bb24, ^bb25
      ^bb24:
        llvm.store %arg0, %130 : i64, !llvm.ptr
        cf.br ^bb26
      ^bb25:
        cf.br ^bb26
      ^bb26:
      %133 = llvm.load %116 : !llvm.ptr -> i64
      %134 = llvm.load %124 : !llvm.ptr -> i64
      %135 = llvm.load %130 : !llvm.ptr -> i64
      %136 = llvm.load %120 : !llvm.ptr -> i64
      %137 = arith.subi %135, %136 : i64
      %138 = arith.muli %134, %137 : i64
      %139 = arith.addi %133, %138 : i64
      llvm.store %139, %116 : i64, !llvm.ptr
      %140 = llvm.load %130 : !llvm.ptr -> i64
      llvm.store %140, %120 : i64, !llvm.ptr
      %141 = llvm.load %124 : !llvm.ptr -> i64
      %142 = arith.constant 1 : i32
      %144 = arith.extsi %142 : i32 to i64
      %143 = arith.addi %141, %144 : i64
      llvm.store %143, %124 : i64, !llvm.ptr
      cf.br ^bb21
    ^bb23:
    %145 = llvm.load %116 : !llvm.ptr -> i64
    func.return %145 : i64
  }
  func.func @main() -> i32 {
    %147 = llvm.mlir.addressof @HSIZE : !llvm.ptr
    %148 = llvm.load %147 : !llvm.ptr -> i64
    %149 = arith.constant 8 : i32
    %150 = arith.extsi %149 : i32 to i64
    %146 = func.call @calloc(%148, %150) : (i64, i64) -> !llvm.ptr
    %151 = llvm.mlir.addressof @HKEY : !llvm.ptr
    llvm.store %146, %151 : !llvm.ptr, !llvm.ptr
    %153 = llvm.mlir.addressof @HSIZE : !llvm.ptr
    %154 = llvm.load %153 : !llvm.ptr -> i64
    %155 = arith.constant 8 : i32
    %156 = arith.extsi %155 : i32 to i64
    %152 = func.call @calloc(%154, %156) : (i64, i64) -> !llvm.ptr
    %157 = llvm.mlir.addressof @HVAL : !llvm.ptr
    llvm.store %152, %157 : !llvm.ptr, !llvm.ptr
    %159 = llvm.mlir.addressof @HSIZE : !llvm.ptr
    %160 = llvm.load %159 : !llvm.ptr -> i64
    %161 = arith.constant 1 : i32
    %162 = arith.extsi %161 : i32 to i64
    %158 = func.call @calloc(%160, %162) : (i64, i64) -> !llvm.ptr
    %163 = llvm.mlir.addressof @HUSE : !llvm.ptr
    llvm.store %158, %163 : !llvm.ptr, !llvm.ptr
    %164 = arith.constant 1 : i32
    %165 = arith.extsi %164 : i32 to i64
    %166 = llvm.mlir.constant(1 : i64) : i64
    %167 = llvm.alloca %166 x i64 : (i64) -> !llvm.ptr
    llvm.store %165, %167 : i64, !llvm.ptr
    %168 = arith.constant 0 : i32
    %169 = arith.extsi %168 : i32 to i64
    %170 = llvm.mlir.constant(1 : i64) : i64
    %171 = llvm.alloca %170 x i64 : (i64) -> !llvm.ptr
    llvm.store %169, %171 : i64, !llvm.ptr
    cf.br ^bb27
    ^bb27:
    %172 = llvm.load %171 : !llvm.ptr -> i64
    %173 = arith.constant 10 : i32
    %175 = arith.extsi %173 : i32 to i64
    %174 = arith.cmpi slt, %172, %175 : i64
    cf.cond_br %174, ^bb28, ^bb29
    ^bb28:
      %176 = llvm.load %167 : !llvm.ptr -> i64
      %177 = arith.constant 7 : i32
      %179 = arith.extsi %177 : i32 to i64
      %178 = arith.muli %176, %179 : i64
      llvm.store %178, %167 : i64, !llvm.ptr
      %180 = llvm.load %171 : !llvm.ptr -> i64
      %181 = arith.constant 1 : i32
      %183 = arith.extsi %181 : i32 to i64
      %182 = arith.addi %180, %183 : i64
      llvm.store %182, %171 : i64, !llvm.ptr
      cf.br ^bb27
    ^bb29:
    %184 = arith.constant 0 : i32
    %185 = arith.extsi %184 : i32 to i64
    %186 = llvm.mlir.constant(1 : i64) : i64
    %187 = llvm.alloca %186 x i64 : (i64) -> !llvm.ptr
    llvm.store %185, %187 : i64, !llvm.ptr
    %188 = arith.constant 0 : i32
    %189 = arith.extsi %188 : i32 to i64
    %190 = llvm.mlir.constant(1 : i64) : i64
    %191 = llvm.alloca %190 x i64 : (i64) -> !llvm.ptr
    llvm.store %189, %191 : i64, !llvm.ptr
    cf.br ^bb30
    ^bb30:
    %192 = llvm.load %191 : !llvm.ptr -> i64
    %193 = arith.constant 7 : i32
    %195 = arith.extsi %193 : i32 to i64
    %194 = arith.cmpi sle, %192, %195 : i64
    cf.cond_br %194, ^bb31, ^bb32
    ^bb31:
      %196 = llvm.load %187 : !llvm.ptr -> i64
      %198 = llvm.load %167 : !llvm.ptr -> i64
      %199 = llvm.load %191 : !llvm.ptr -> i64
      %197 = func.call @sum_Q_upto(%198, %199) : (i64, i64) -> i64
      %200 = arith.addi %196, %197 : i64
      llvm.store %200, %187 : i64, !llvm.ptr
      %201 = llvm.load %191 : !llvm.ptr -> i64
      %202 = arith.constant 1 : i32
      %204 = arith.extsi %202 : i32 to i64
      %203 = arith.addi %201, %204 : i64
      llvm.store %203, %191 : i64, !llvm.ptr
      cf.br ^bb30
    ^bb32:
    %205 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %206 = llvm.load %187 : !llvm.ptr -> i64
    %207 = llvm.call @printf(%205, %206) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %209 = llvm.mlir.addressof @HKEY : !llvm.ptr
    %210 = llvm.load %209 : !llvm.ptr -> !llvm.ptr
    func.call @free(%210) : (!llvm.ptr) -> ()
    %212 = llvm.mlir.addressof @HVAL : !llvm.ptr
    %213 = llvm.load %212 : !llvm.ptr -> !llvm.ptr
    func.call @free(%213) : (!llvm.ptr) -> ()
    %215 = llvm.mlir.addressof @HUSE : !llvm.ptr
    %216 = llvm.load %215 : !llvm.ptr -> !llvm.ptr
    func.call @free(%216) : (!llvm.ptr) -> ()
    %217 = arith.constant 0 : i32
    func.return %217 : i32
  }
}