Problem 140

Sum of the first thirty golden nuggets for modified Fibonacci GF.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 140
# Sum of the first thirty golden nuggets for modified Fibonacci GF.

function main() -> i32 {
    # First four nuggets; then n_k = 7*n_{k-2} - n_{k-4} + 7
    let mut n0: i64 = 2
    let mut n1: i64 = 5
    let mut n2: i64 = 21
    let mut n3: i64 = 42
    let mut total: i64 = n0 + n1 + n2 + n3
    let mut k: i32 = 4
    while k < 30 {
        let nxt: i64 = 7 * n2 - n0 + 7
        total = total + nxt
        n0 = n1
        n1 = n2
        n2 = n3
        n3 = nxt
        k = k + 1
    }
    printf("%lld\n", total)
    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; }

int32_t main(void);

int32_t main(void) {
    int64_t n0 = 2;
    int64_t n1 = 5;
    int64_t n2 = 21;
    int64_t n3 = 42;
    int64_t total = (((n0 + n1) + n2) + n3);
    int32_t k = 4;
    while (k < 30) {
        int64_t nxt = (((7 * n2) - n0) + 7);
        total = (total + nxt);
        n0 = n1;
        n1 = n2;
        n2 = n3;
        n3 = nxt;
        k = (k + 1);
    }
    printf("%lld\n", total);
    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 @main() -> i32 {
    %0 = arith.constant 2 : 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.constant 5 : i32
    %5 = arith.extsi %4 : i32 to i64
    %6 = llvm.mlir.constant(1 : i64) : i64
    %7 = llvm.alloca %6 x i64 : (i64) -> !llvm.ptr
    llvm.store %5, %7 : i64, !llvm.ptr
    %8 = arith.constant 21 : i32
    %9 = arith.extsi %8 : i32 to i64
    %10 = llvm.mlir.constant(1 : i64) : i64
    %11 = llvm.alloca %10 x i64 : (i64) -> !llvm.ptr
    llvm.store %9, %11 : i64, !llvm.ptr
    %12 = arith.constant 42 : i32
    %13 = arith.extsi %12 : i32 to i64
    %14 = llvm.mlir.constant(1 : i64) : i64
    %15 = llvm.alloca %14 x i64 : (i64) -> !llvm.ptr
    llvm.store %13, %15 : i64, !llvm.ptr
    %16 = llvm.load %3 : !llvm.ptr -> i64
    %17 = llvm.load %7 : !llvm.ptr -> i64
    %18 = arith.addi %16, %17 : i64
    %19 = llvm.load %11 : !llvm.ptr -> i64
    %20 = arith.addi %18, %19 : i64
    %21 = llvm.load %15 : !llvm.ptr -> i64
    %22 = arith.addi %20, %21 : i64
    %23 = llvm.mlir.constant(1 : i64) : i64
    %24 = llvm.alloca %23 x i64 : (i64) -> !llvm.ptr
    llvm.store %22, %24 : i64, !llvm.ptr
    %25 = arith.constant 4 : i32
    %26 = llvm.mlir.constant(1 : i64) : i64
    %27 = llvm.alloca %26 x i32 : (i64) -> !llvm.ptr
    llvm.store %25, %27 : i32, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %28 = llvm.load %27 : !llvm.ptr -> i32
    %29 = arith.constant 30 : i32
    %30 = arith.cmpi slt, %28, %29 : i32
    cf.cond_br %30, ^bb1, ^bb2
    ^bb1:
      %31 = arith.constant 7 : i32
      %32 = llvm.load %11 : !llvm.ptr -> i64
      %34 = arith.extsi %31 : i32 to i64
      %33 = arith.muli %34, %32 : i64
      %35 = llvm.load %3 : !llvm.ptr -> i64
      %36 = arith.subi %33, %35 : i64
      %37 = arith.constant 7 : i32
      %39 = arith.extsi %37 : i32 to i64
      %38 = arith.addi %36, %39 : i64
      %40 = llvm.load %24 : !llvm.ptr -> i64
      %41 = arith.addi %40, %38 : i64
      llvm.store %41, %24 : i64, !llvm.ptr
      %42 = llvm.load %7 : !llvm.ptr -> i64
      llvm.store %42, %3 : i64, !llvm.ptr
      %43 = llvm.load %11 : !llvm.ptr -> i64
      llvm.store %43, %7 : i64, !llvm.ptr
      %44 = llvm.load %15 : !llvm.ptr -> i64
      llvm.store %44, %11 : i64, !llvm.ptr
      llvm.store %38, %15 : i64, !llvm.ptr
      %45 = llvm.load %27 : !llvm.ptr -> i32
      %46 = arith.constant 1 : i32
      %47 = arith.addi %45, %46 : i32
      llvm.store %47, %27 : i32, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %48 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %49 = llvm.load %24 : !llvm.ptr -> i64
    %50 = llvm.call @printf(%48, %49) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %51 = arith.constant 0 : i32
    func.return %51 : i32
  }
}