Problem 097

Last ten digits of 28433 × 2^7830457 + 1.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 097
# Last ten digits of 28433 × 2^7830457 + 1.

function mod_mul(a: i64, b: i64, mod: i64) -> i64 {
    # overflow-safe multiply via addition
    let mut res: i64 = 0
    let mut x: i64 = a % mod
    let mut y: i64 = b
    if x < 0 { x = x + mod }
    while y > 0 {
        if y % 2 == 1 {
            res = res + x
            if res >= mod { res = res - mod }
        }
        x = x + x
        if x >= mod { x = x - mod }
        y = y / 2
    }
    return res
}

function mod_pow(base: i64, exp: i64, mod: i64) -> i64 {
    let mut r: i64 = 1
    let mut b: i64 = base % mod
    let mut e: i64 = exp
    while e > 0 {
        if e % 2 == 1 {
            r = mod_mul(r, b, mod)
        }
        b = mod_mul(b, b, mod)
        e = e / 2
    }
    return r
}

function main() -> i32 {
    let mod: i64 = 10000000000
    let p: i64 = mod_pow(2, 7830457, mod)
    let ans: i64 = (mod_mul(28433, p, mod) + 1) % mod
    printf("%lld\n", ans)
    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 mod_mul_i64_i64_i64(int64_t a, int64_t b, int64_t mod);
int64_t mod_pow_i64_i64_i64(int64_t base, int64_t exp, int64_t mod);
int32_t main(void);

int64_t mod_mul_i64_i64_i64(int64_t a, int64_t b, int64_t mod) {
    int64_t res = 0;
    int64_t x = FLOW_CHECKED_MOD((a), (mod));
    int64_t y = b;
    if (x < 0) {
        x = (x + mod);
    }
    while (y > 0) {
        if (FLOW_CHECKED_MOD((y), (2)) == 1) {
            res = (res + x);
            if (res >= mod) {
                res = (res - mod);
            }
        }
        x = (x + x);
        if (x >= mod) {
            x = (x - mod);
        }
        y = FLOW_CHECKED_DIV((y), (2));
    }
    return res;
}

int64_t mod_pow_i64_i64_i64(int64_t base, int64_t exp, int64_t mod) {
    int64_t r = 1;
    int64_t b = FLOW_CHECKED_MOD((base), (mod));
    int64_t e = exp;
    while (e > 0) {
        if (FLOW_CHECKED_MOD((e), (2)) == 1) {
            r = mod_mul_i64_i64_i64(r, b, mod);
        }
        b = mod_mul_i64_i64_i64(b, b, mod);
        e = FLOW_CHECKED_DIV((e), (2));
    }
    return r;
}

int32_t main(void) {
    int64_t mod = 10000000000;
    int64_t p = mod_pow_i64_i64_i64(2, 7830457, mod);
    int64_t ans = FLOW_CHECKED_MOD(((mod_mul_i64_i64_i64(28433, p, mod) + 1)), (mod));
    printf("%lld\n", ans);
    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 @mod_mul(%arg0: i64, %arg1: i64, %arg2: 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 = arith.remsi %arg0, %arg2 : i64
    %5 = llvm.mlir.constant(1 : i64) : i64
    %6 = llvm.alloca %5 x i64 : (i64) -> !llvm.ptr
    llvm.store %4, %6 : i64, !llvm.ptr
    %7 = llvm.mlir.constant(1 : i64) : i64
    %8 = llvm.alloca %7 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg1, %8 : i64, !llvm.ptr
    %9 = llvm.load %6 : !llvm.ptr -> i64
    %10 = arith.constant 0 : i32
    %12 = arith.extsi %10 : i32 to i64
    %11 = arith.cmpi slt, %9, %12 : i64
    cf.cond_br %11, ^bb0, ^bb1
    ^bb0:
      %13 = llvm.load %6 : !llvm.ptr -> i64
      %14 = arith.addi %13, %arg2 : i64
      llvm.store %14, %6 : i64, !llvm.ptr
      cf.br ^bb2
    ^bb1:
      cf.br ^bb2
    ^bb2:
    cf.br ^bb3
    ^bb3:
    %15 = llvm.load %8 : !llvm.ptr -> i64
    %16 = arith.constant 0 : i32
    %18 = arith.extsi %16 : i32 to i64
    %17 = arith.cmpi sgt, %15, %18 : i64
    cf.cond_br %17, ^bb4, ^bb5
    ^bb4:
      %19 = llvm.load %8 : !llvm.ptr -> i64
      %20 = arith.constant 2 : i32
      %22 = arith.extsi %20 : i32 to i64
      %21 = arith.remsi %19, %22 : i64
      %23 = arith.constant 1 : i32
      %25 = arith.extsi %23 : i32 to i64
      %24 = arith.cmpi eq, %21, %25 : i64
      cf.cond_br %24, ^bb6, ^bb7
      ^bb6:
        %26 = llvm.load %3 : !llvm.ptr -> i64
        %27 = llvm.load %6 : !llvm.ptr -> i64
        %28 = arith.addi %26, %27 : i64
        llvm.store %28, %3 : i64, !llvm.ptr
        %29 = llvm.load %3 : !llvm.ptr -> i64
        %30 = arith.cmpi sge, %29, %arg2 : i64
        cf.cond_br %30, ^bb9, ^bb10
        ^bb9:
          %31 = llvm.load %3 : !llvm.ptr -> i64
          %32 = arith.subi %31, %arg2 : i64
          llvm.store %32, %3 : i64, !llvm.ptr
          cf.br ^bb11
        ^bb10:
          cf.br ^bb11
        ^bb11:
        cf.br ^bb8
      ^bb7:
        cf.br ^bb8
      ^bb8:
      %33 = llvm.load %6 : !llvm.ptr -> i64
      %34 = llvm.load %6 : !llvm.ptr -> i64
      %35 = arith.addi %33, %34 : i64
      llvm.store %35, %6 : i64, !llvm.ptr
      %36 = llvm.load %6 : !llvm.ptr -> i64
      %37 = arith.cmpi sge, %36, %arg2 : i64
      cf.cond_br %37, ^bb12, ^bb13
      ^bb12:
        %38 = llvm.load %6 : !llvm.ptr -> i64
        %39 = arith.subi %38, %arg2 : i64
        llvm.store %39, %6 : i64, !llvm.ptr
        cf.br ^bb14
      ^bb13:
        cf.br ^bb14
      ^bb14:
      %40 = llvm.load %8 : !llvm.ptr -> i64
      %41 = arith.constant 2 : i32
      %43 = arith.extsi %41 : i32 to i64
      %42 = arith.divsi %40, %43 : i64
      llvm.store %42, %8 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %44 = llvm.load %3 : !llvm.ptr -> i64
    func.return %44 : i64
  }
  func.func @mod_pow(%arg0: i64, %arg1: i64, %arg2: i64) -> i64 {
    %45 = arith.constant 1 : i32
    %46 = arith.extsi %45 : i32 to i64
    %47 = llvm.mlir.constant(1 : i64) : i64
    %48 = llvm.alloca %47 x i64 : (i64) -> !llvm.ptr
    llvm.store %46, %48 : i64, !llvm.ptr
    %49 = arith.remsi %arg0, %arg2 : i64
    %50 = llvm.mlir.constant(1 : i64) : i64
    %51 = llvm.alloca %50 x i64 : (i64) -> !llvm.ptr
    llvm.store %49, %51 : i64, !llvm.ptr
    %52 = llvm.mlir.constant(1 : i64) : i64
    %53 = llvm.alloca %52 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg1, %53 : i64, !llvm.ptr
    cf.br ^bb15
    ^bb15:
    %54 = llvm.load %53 : !llvm.ptr -> i64
    %55 = arith.constant 0 : i32
    %57 = arith.extsi %55 : i32 to i64
    %56 = arith.cmpi sgt, %54, %57 : i64
    cf.cond_br %56, ^bb16, ^bb17
    ^bb16:
      %58 = llvm.load %53 : !llvm.ptr -> i64
      %59 = arith.constant 2 : i32
      %61 = arith.extsi %59 : i32 to i64
      %60 = arith.remsi %58, %61 : i64
      %62 = arith.constant 1 : i32
      %64 = arith.extsi %62 : i32 to i64
      %63 = arith.cmpi eq, %60, %64 : i64
      cf.cond_br %63, ^bb18, ^bb19
      ^bb18:
        %66 = llvm.load %48 : !llvm.ptr -> i64
        %67 = llvm.load %51 : !llvm.ptr -> i64
        %65 = func.call @mod_mul(%66, %67, %arg2) : (i64, i64, i64) -> i64
        llvm.store %65, %48 : i64, !llvm.ptr
        cf.br ^bb20
      ^bb19:
        cf.br ^bb20
      ^bb20:
      %69 = llvm.load %51 : !llvm.ptr -> i64
      %70 = llvm.load %51 : !llvm.ptr -> i64
      %68 = func.call @mod_mul(%69, %70, %arg2) : (i64, i64, i64) -> i64
      llvm.store %68, %51 : i64, !llvm.ptr
      %71 = llvm.load %53 : !llvm.ptr -> i64
      %72 = arith.constant 2 : i32
      %74 = arith.extsi %72 : i32 to i64
      %73 = arith.divsi %71, %74 : i64
      llvm.store %73, %53 : i64, !llvm.ptr
      cf.br ^bb15
    ^bb17:
    %75 = llvm.load %48 : !llvm.ptr -> i64
    func.return %75 : i64
  }
  func.func @main() -> i32 {
    %76 = arith.constant 5705032704 : i32
    %77 = arith.extsi %76 : i32 to i64
    %79 = arith.constant 2 : i32
    %80 = arith.constant 7830457 : i32
    %81 = arith.extsi %79 : i32 to i64
    %82 = arith.extsi %80 : i32 to i64
    %78 = func.call @mod_pow(%81, %82, %77) : (i64, i64, i64) -> i64
    %84 = arith.constant 28433 : i32
    %85 = arith.extsi %84 : i32 to i64
    %83 = func.call @mod_mul(%85, %78, %77) : (i64, i64, i64) -> i64
    %86 = arith.constant 1 : i32
    %88 = arith.extsi %86 : i32 to i64
    %87 = arith.addi %83, %88 : i64
    %89 = arith.remsi %87, %77 : i64
    %90 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %91 = llvm.call @printf(%90, %89) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %92 = arith.constant 0 : i32
    func.return %92 : i32
  }
}