Problem 228

Sides of Minkowski sum S_1864+...+S_1909.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 228
# Sides of Minkowski sum S_1864+...+S_1909.

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

function main() -> i32 {
    let L: i64 = 1864
    let R: i64 = 1909
    let phi: ptr<i32> = calloc(R + 1, 4)
    if phi == null { return 1 }
    let mut i: i64 = 0
    while i <= R {
        phi[i] = i as i32
        i = i + 1
    }
    i = 2
    while i <= R {
        if phi[i] == i as i32 {
            let mut j: i64 = i
            while j <= R {
                phi[j] = phi[j] - phi[j] / (i as i32)
                j = j + i
            }
        }
        i = i + 1
    }
    let mut total: i64 = 0
    let mut d: i64 = 1
    while d <= R {
        let first: i64 = ((L + d - 1) / d) * d
        if first <= R {
            total = total + (phi[d] as i64)
        }
        d = d + 1
    }
    printf("%lld\n", total)
    free(phi)
    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 L = 1864;
    int64_t R = 1909;
    int32_t* phi = (int32_t*)(calloc((R + 1), 4));
    if (phi == NULL) {
        return 1;
    }
    int64_t i = 0;
    while (i <= R) {
        phi[i] = ((int32_t)(i));
        i = (i + 1);
    }
    i = 2;
    while (i <= R) {
        if (phi[i] == ((int32_t)(i))) {
            int64_t j = i;
            while (j <= R) {
                phi[j] = (phi[j] - FLOW_CHECKED_DIV((phi[j]), (((int32_t)(i)))));
                j = (j + i);
            }
        }
        i = (i + 1);
    }
    int64_t total = 0;
    int64_t d = 1;
    while (d <= R) {
        int64_t first = (FLOW_CHECKED_DIV((((L + d) - 1)), (d)) * d);
        if (first <= R) {
            total = (total + ((int64_t)(phi[d])));
        }
        d = (d + 1);
    }
    printf("%lld\n", total);
    free(phi);
    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 @main() -> i32 {
    %0 = arith.constant 1864 : i32
    %1 = arith.extsi %0 : i32 to i64
    %2 = arith.constant 1909 : i32
    %3 = arith.extsi %2 : i32 to i64
    %5 = arith.constant 1 : i32
    %7 = arith.extsi %5 : i32 to i64
    %6 = arith.addi %3, %7 : i64
    %8 = arith.constant 4 : i32
    %9 = arith.extsi %8 : i32 to i64
    %4 = func.call @calloc(%6, %9) : (i64, i64) -> !llvm.ptr
    %10 = llvm.mlir.zero : !llvm.ptr
    %11 = llvm.icmp "eq" %4, %10 : !llvm.ptr
    cf.cond_br %11, ^bb0, ^bb1
    ^bb0:
      %12 = arith.constant 1 : i32
      func.return %12 : i32
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %13 = arith.constant 0 : i32
    %14 = arith.extsi %13 : i32 to i64
    %15 = llvm.mlir.constant(1 : i64) : i64
    %16 = llvm.alloca %15 x i64 : (i64) -> !llvm.ptr
    llvm.store %14, %16 : i64, !llvm.ptr
    cf.br ^bb3
    ^bb3:
    %17 = llvm.load %16 : !llvm.ptr -> i64
    %18 = arith.cmpi sle, %17, %3 : i64
    cf.cond_br %18, ^bb4, ^bb5
    ^bb4:
      %19 = llvm.load %16 : !llvm.ptr -> i64
      %20 = arith.trunci %19 : i64 to i32
      %21 = llvm.load %16 : !llvm.ptr -> i64
      %22 = llvm.getelementptr %4[%21] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %20, %22 : i32, !llvm.ptr
      %23 = llvm.load %16 : !llvm.ptr -> i64
      %24 = arith.constant 1 : i32
      %26 = arith.extsi %24 : i32 to i64
      %25 = arith.addi %23, %26 : i64
      llvm.store %25, %16 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %27 = arith.constant 2 : i32
    %28 = arith.extsi %27 : i32 to i64
    llvm.store %28, %16 : i64, !llvm.ptr
    cf.br ^bb6
    ^bb6:
    %29 = llvm.load %16 : !llvm.ptr -> i64
    %30 = arith.cmpi sle, %29, %3 : i64
    cf.cond_br %30, ^bb7, ^bb8
    ^bb7:
      %32 = llvm.load %16 : !llvm.ptr -> i64
      %33 = llvm.getelementptr %4[%32] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %31 = llvm.load %33 : !llvm.ptr -> i32
      %34 = llvm.load %16 : !llvm.ptr -> i64
      %35 = arith.trunci %34 : i64 to i32
      %36 = arith.cmpi eq, %31, %35 : i32
      cf.cond_br %36, ^bb9, ^bb10
      ^bb9:
        %37 = llvm.load %16 : !llvm.ptr -> i64
        %38 = llvm.mlir.constant(1 : i64) : i64
        %39 = llvm.alloca %38 x i64 : (i64) -> !llvm.ptr
        llvm.store %37, %39 : i64, !llvm.ptr
        cf.br ^bb12
        ^bb12:
        %40 = llvm.load %39 : !llvm.ptr -> i64
        %41 = arith.cmpi sle, %40, %3 : i64
        cf.cond_br %41, ^bb13, ^bb14
        ^bb13:
          %43 = llvm.load %39 : !llvm.ptr -> i64
          %44 = llvm.getelementptr %4[%43] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          %42 = llvm.load %44 : !llvm.ptr -> i32
          %46 = llvm.load %39 : !llvm.ptr -> i64
          %47 = llvm.getelementptr %4[%46] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          %45 = llvm.load %47 : !llvm.ptr -> i32
          %48 = llvm.load %16 : !llvm.ptr -> i64
          %49 = arith.trunci %48 : i64 to i32
          %50 = arith.divsi %45, %49 : i32
          %51 = arith.subi %42, %50 : i32
          %52 = llvm.load %39 : !llvm.ptr -> i64
          %53 = llvm.getelementptr %4[%52] : (!llvm.ptr, i64) -> !llvm.ptr, i32
          llvm.store %51, %53 : i32, !llvm.ptr
          %54 = llvm.load %39 : !llvm.ptr -> i64
          %55 = llvm.load %16 : !llvm.ptr -> i64
          %56 = arith.addi %54, %55 : i64
          llvm.store %56, %39 : i64, !llvm.ptr
          cf.br ^bb12
        ^bb14:
        cf.br ^bb11
      ^bb10:
        cf.br ^bb11
      ^bb11:
      %57 = llvm.load %16 : !llvm.ptr -> i64
      %58 = arith.constant 1 : i32
      %60 = arith.extsi %58 : i32 to i64
      %59 = arith.addi %57, %60 : i64
      llvm.store %59, %16 : i64, !llvm.ptr
      cf.br ^bb6
    ^bb8:
    %61 = arith.constant 0 : i32
    %62 = arith.extsi %61 : i32 to i64
    %63 = llvm.mlir.constant(1 : i64) : i64
    %64 = llvm.alloca %63 x i64 : (i64) -> !llvm.ptr
    llvm.store %62, %64 : i64, !llvm.ptr
    %65 = arith.constant 1 : i32
    %66 = arith.extsi %65 : i32 to i64
    %67 = llvm.mlir.constant(1 : i64) : i64
    %68 = llvm.alloca %67 x i64 : (i64) -> !llvm.ptr
    llvm.store %66, %68 : i64, !llvm.ptr
    cf.br ^bb15
    ^bb15:
    %69 = llvm.load %68 : !llvm.ptr -> i64
    %70 = arith.cmpi sle, %69, %3 : i64
    cf.cond_br %70, ^bb16, ^bb17
    ^bb16:
      %71 = llvm.load %68 : !llvm.ptr -> i64
      %72 = arith.addi %1, %71 : i64
      %73 = arith.constant 1 : i32
      %75 = arith.extsi %73 : i32 to i64
      %74 = arith.subi %72, %75 : i64
      %76 = llvm.load %68 : !llvm.ptr -> i64
      %77 = arith.divsi %74, %76 : i64
      %78 = llvm.load %68 : !llvm.ptr -> i64
      %79 = arith.muli %77, %78 : i64
      %80 = arith.cmpi sle, %79, %3 : i64
      cf.cond_br %80, ^bb18, ^bb19
      ^bb18:
        %81 = llvm.load %64 : !llvm.ptr -> i64
        %83 = llvm.load %68 : !llvm.ptr -> i64
        %84 = llvm.getelementptr %4[%83] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        %82 = llvm.load %84 : !llvm.ptr -> i32
        %85 = arith.extsi %82 : i32 to i64
        %86 = arith.addi %81, %85 : i64
        llvm.store %86, %64 : i64, !llvm.ptr
        cf.br ^bb20
      ^bb19:
        cf.br ^bb20
      ^bb20:
      %87 = llvm.load %68 : !llvm.ptr -> i64
      %88 = arith.constant 1 : i32
      %90 = arith.extsi %88 : i32 to i64
      %89 = arith.addi %87, %90 : i64
      llvm.store %89, %68 : i64, !llvm.ptr
      cf.br ^bb15
    ^bb17:
    %91 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %92 = llvm.load %64 : !llvm.ptr -> i64
    %93 = llvm.call @printf(%91, %92) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    func.call @free(%4) : (!llvm.ptr) -> ()
    %95 = arith.constant 0 : i32
    func.return %95 : i32
  }
}