Problem 039

Perimeter p ≤ 1000 maximizing number of integer right triangles {a,b,c}.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 039
# Perimeter p ≤ 1000 maximizing number of integer right triangles {a,b,c}.

function main() -> i32 {
    let mut best_p: i64 = 0
    let mut best_c: i64 = 0
    let mut p: i64 = 12
    while p <= 1000 {
        let mut count: i64 = 0
        let mut a: i64 = 1
        while a < p / 3 {
            let mut b: i64 = a
            while b < p / 2 {
                let c: i64 = p - a - b
                if b > c { break }
                if a * a + b * b == c * c {
                    count = count + 1
                }
                b = b + 1
            }
            a = a + 1
        }
        if count > best_c {
            best_c = count
            best_p = p
        }
        p = p + 1
    }
    printf("%lld\n", best_p)
    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 best_p = 0;
    int64_t best_c = 0;
    int64_t p = 12;
    while (p <= 1000) {
        int64_t count = 0;
        int64_t a = 1;
        while (a < FLOW_CHECKED_DIV((p), (3))) {
            int64_t b = a;
            while (b < FLOW_CHECKED_DIV((p), (2))) {
                int64_t c = ((p - a) - b);
                if (b > c) {
                    break;
                }
                if (((a * a) + (b * b)) == (c * c)) {
                    count = (count + 1);
                }
                b = (b + 1);
            }
            a = (a + 1);
        }
        if (count > best_c) {
            best_c = count;
            best_p = p;
        }
        p = (p + 1);
    }
    printf("%lld\n", best_p);
    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 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.constant 0 : 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 12 : 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
    cf.br ^bb0
    ^bb0:
    %12 = llvm.load %11 : !llvm.ptr -> i64
    %13 = arith.constant 1000 : i32
    %15 = arith.extsi %13 : i32 to i64
    %14 = arith.cmpi sle, %12, %15 : i64
    cf.cond_br %14, ^bb1, ^bb2
    ^bb1:
      %16 = arith.constant 0 : i32
      %17 = arith.extsi %16 : i32 to i64
      %18 = llvm.mlir.constant(1 : i64) : i64
      %19 = llvm.alloca %18 x i64 : (i64) -> !llvm.ptr
      llvm.store %17, %19 : i64, !llvm.ptr
      %20 = arith.constant 1 : i32
      %21 = arith.extsi %20 : i32 to i64
      %22 = llvm.mlir.constant(1 : i64) : i64
      %23 = llvm.alloca %22 x i64 : (i64) -> !llvm.ptr
      llvm.store %21, %23 : i64, !llvm.ptr
      cf.br ^bb3
      ^bb3:
      %24 = llvm.load %23 : !llvm.ptr -> i64
      %25 = llvm.load %11 : !llvm.ptr -> i64
      %26 = arith.constant 3 : i32
      %28 = arith.extsi %26 : i32 to i64
      %27 = arith.divsi %25, %28 : i64
      %29 = arith.cmpi slt, %24, %27 : i64
      cf.cond_br %29, ^bb4, ^bb5
      ^bb4:
        %30 = llvm.load %23 : !llvm.ptr -> i64
        %31 = llvm.mlir.constant(1 : i64) : i64
        %32 = llvm.alloca %31 x i64 : (i64) -> !llvm.ptr
        llvm.store %30, %32 : i64, !llvm.ptr
        cf.br ^bb6
        ^bb6:
        %33 = llvm.load %32 : !llvm.ptr -> i64
        %34 = llvm.load %11 : !llvm.ptr -> i64
        %35 = arith.constant 2 : i32
        %37 = arith.extsi %35 : i32 to i64
        %36 = arith.divsi %34, %37 : i64
        %38 = arith.cmpi slt, %33, %36 : i64
        cf.cond_br %38, ^bb7, ^bb8
        ^bb7:
          %39 = llvm.load %11 : !llvm.ptr -> i64
          %40 = llvm.load %23 : !llvm.ptr -> i64
          %41 = arith.subi %39, %40 : i64
          %42 = llvm.load %32 : !llvm.ptr -> i64
          %43 = arith.subi %41, %42 : i64
          %44 = llvm.load %32 : !llvm.ptr -> i64
          %45 = arith.cmpi sgt, %44, %43 : i64
          cf.cond_br %45, ^bb9, ^bb10
          ^bb9:
            cf.br ^bb8
          ^bb10:
            cf.br ^bb11
          ^bb11:
          %46 = llvm.load %23 : !llvm.ptr -> i64
          %47 = llvm.load %23 : !llvm.ptr -> i64
          %48 = arith.muli %46, %47 : i64
          %49 = llvm.load %32 : !llvm.ptr -> i64
          %50 = llvm.load %32 : !llvm.ptr -> i64
          %51 = arith.muli %49, %50 : i64
          %52 = arith.addi %48, %51 : i64
          %53 = arith.muli %43, %43 : i64
          %54 = arith.cmpi eq, %52, %53 : i64
          cf.cond_br %54, ^bb12, ^bb13
          ^bb12:
            %55 = llvm.load %19 : !llvm.ptr -> i64
            %56 = arith.constant 1 : i32
            %58 = arith.extsi %56 : i32 to i64
            %57 = arith.addi %55, %58 : i64
            llvm.store %57, %19 : i64, !llvm.ptr
            cf.br ^bb14
          ^bb13:
            cf.br ^bb14
          ^bb14:
          %59 = llvm.load %32 : !llvm.ptr -> i64
          %60 = arith.constant 1 : i32
          %62 = arith.extsi %60 : i32 to i64
          %61 = arith.addi %59, %62 : i64
          llvm.store %61, %32 : i64, !llvm.ptr
          cf.br ^bb6
        ^bb8:
        %63 = llvm.load %23 : !llvm.ptr -> i64
        %64 = arith.constant 1 : i32
        %66 = arith.extsi %64 : i32 to i64
        %65 = arith.addi %63, %66 : i64
        llvm.store %65, %23 : i64, !llvm.ptr
        cf.br ^bb3
      ^bb5:
      %67 = llvm.load %19 : !llvm.ptr -> i64
      %68 = llvm.load %7 : !llvm.ptr -> i64
      %69 = arith.cmpi sgt, %67, %68 : i64
      cf.cond_br %69, ^bb15, ^bb16
      ^bb15:
        %70 = llvm.load %19 : !llvm.ptr -> i64
        llvm.store %70, %7 : i64, !llvm.ptr
        %71 = llvm.load %11 : !llvm.ptr -> i64
        llvm.store %71, %3 : i64, !llvm.ptr
        cf.br ^bb17
      ^bb16:
        cf.br ^bb17
      ^bb17:
      %72 = llvm.load %11 : !llvm.ptr -> i64
      %73 = arith.constant 1 : i32
      %75 = arith.extsi %73 : i32 to i64
      %74 = arith.addi %72, %75 : i64
      llvm.store %74, %11 : i64, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %76 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %77 = llvm.load %3 : !llvm.ptr -> i64
    %78 = llvm.call @printf(%76, %77) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %79 = arith.constant 0 : i32
    func.return %79 : i32
  }
}