Problem 181
Ways to group 60 black and 40 white objects.
View problem on Project Euler
Performance comparison
| Metric | Our solution | Best known |
| Time complexity | O(n^5) | O(n * m) |
| Space complexity | O(n) | O(n) |
| Approach | Flow solution | Dynamic programming or generating function |
| Verdict | Unknown |
Flow source
# Project Euler 181
# Ways to group 60 black and 40 white objects.
extern {
function calloc(n: i64, size: i64) -> ptr<void>
function free(p: ptr<void>) -> void
function memcpy(dst: ptr<void>, src: ptr<void>, n: i64) -> ptr<void>
}
function main() -> i32 {
let MB: i32 = 60
let MW: i32 = 40
let rows: i64 = (MB + 1) as i64
let cols: i64 = (MW + 1) as i64
let prev: ptr<i64> = calloc(rows * cols, 8)
let cur: ptr<i64> = calloc(rows * cols, 8)
if prev == null || cur == null { return 1 }
prev[0] = 1
let mut ub: i32 = 0
while ub <= MB {
let mut uw: i32 = 0
while uw <= MW {
if !(ub == 0 && uw == 0) {
let mut i: i32 = 0
while i <= MB {
let mut j: i32 = 0
while j <= MW {
let mut total: i64 = 0
let mut k: i32 = 0
while i >= k * ub && j >= k * uw {
total = total + prev[((i - k * ub) as i64) * cols + ((j - k * uw) as i64)]
k = k + 1
}
cur[(i as i64) * cols + (j as i64)] = total
j = j + 1
}
i = i + 1
}
memcpy(prev as ptr<void>, cur as ptr<void>, rows * cols * 8)
}
uw = uw + 1
}
ub = ub + 1
}
printf("%lld\n", cur[(MB as i64) * cols + (MW as i64)])
free(prev)
free(cur)
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) {
int32_t MB = 60;
int32_t MW = 40;
int64_t rows = ((int64_t)((MB + 1)));
int64_t cols = ((int64_t)((MW + 1)));
int64_t* prev = (int64_t*)(calloc((rows * cols), 8));
int64_t* cur = (int64_t*)(calloc((rows * cols), 8));
if ((prev == NULL || cur == NULL)) {
return 1;
}
prev[0] = 1;
int32_t ub = 0;
while (ub <= MB) {
int32_t uw = 0;
while (uw <= MW) {
if ((!((ub == 0 && uw == 0)))) {
int32_t i = 0;
while (i <= MB) {
int32_t j = 0;
while (j <= MW) {
int64_t total = 0;
int32_t k = 0;
while ((i >= (k * ub) && j >= (k * uw))) {
total = (total + prev[((((int64_t)((i - (k * ub)))) * cols) + ((int64_t)((j - (k * uw)))))]);
k = (k + 1);
}
cur[((((int64_t)(i)) * cols) + ((int64_t)(j)))] = total;
j = (j + 1);
}
i = (i + 1);
}
memcpy(((void*)(prev)), ((void*)(cur)), ((rows * cols) * 8));
}
uw = (uw + 1);
}
ub = (ub + 1);
}
printf("%lld\n", cur[((((int64_t)(MB)) * cols) + ((int64_t)(MW)))]);
free(prev);
free(cur);
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 private @memcpy(!llvm.ptr, !llvm.ptr, i64) -> !llvm.ptr
func.func @main() -> i32 {
%0 = arith.constant 60 : i32
%1 = arith.constant 40 : i32
%2 = arith.constant 1 : i32
%3 = arith.addi %0, %2 : i32
%4 = arith.extsi %3 : i32 to i64
%5 = arith.constant 1 : i32
%6 = arith.addi %1, %5 : i32
%7 = arith.extsi %6 : i32 to i64
%9 = arith.muli %4, %7 : i64
%10 = arith.constant 8 : i32
%11 = arith.extsi %10 : i32 to i64
%8 = func.call @calloc(%9, %11) : (i64, i64) -> !llvm.ptr
%13 = arith.muli %4, %7 : i64
%14 = arith.constant 8 : i32
%15 = arith.extsi %14 : i32 to i64
%12 = func.call @calloc(%13, %15) : (i64, i64) -> !llvm.ptr
%16 = llvm.mlir.zero : !llvm.ptr
%17 = llvm.icmp "eq" %8, %16 : !llvm.ptr
%18 = scf.if %17 -> (i1) {
%19 = arith.constant true
scf.yield %19 : i1
} else {
%20 = llvm.mlir.zero : !llvm.ptr
%21 = llvm.icmp "eq" %12, %20 : !llvm.ptr
scf.yield %21 : i1
}
cf.cond_br %18, ^bb0, ^bb1
^bb0:
%22 = arith.constant 1 : i32
func.return %22 : i32
^bb1:
cf.br ^bb2
^bb2:
%23 = arith.constant 1 : i32
%24 = arith.constant 0 : i32
%25 = arith.extsi %23 : i32 to i64
%26 = arith.extsi %24 : i32 to i64
%27 = llvm.getelementptr %8[%26] : (!llvm.ptr, i64) -> !llvm.ptr, i64
llvm.store %25, %27 : i64, !llvm.ptr
%28 = arith.constant 0 : i32
%29 = llvm.mlir.constant(1 : i64) : i64
%30 = llvm.alloca %29 x i32 : (i64) -> !llvm.ptr
llvm.store %28, %30 : i32, !llvm.ptr
cf.br ^bb3
^bb3:
%31 = llvm.load %30 : !llvm.ptr -> i32
%32 = arith.cmpi sle, %31, %0 : i32
cf.cond_br %32, ^bb4, ^bb5
^bb4:
%33 = arith.constant 0 : i32
%34 = llvm.mlir.constant(1 : i64) : i64
%35 = llvm.alloca %34 x i32 : (i64) -> !llvm.ptr
llvm.store %33, %35 : i32, !llvm.ptr
cf.br ^bb6
^bb6:
%36 = llvm.load %35 : !llvm.ptr -> i32
%37 = arith.cmpi sle, %36, %1 : i32
cf.cond_br %37, ^bb7, ^bb8
^bb7:
%38 = llvm.load %30 : !llvm.ptr -> i32
%39 = arith.constant 0 : i32
%40 = arith.cmpi eq, %38, %39 : i32
%41 = scf.if %40 -> (i1) {
%42 = llvm.load %35 : !llvm.ptr -> i32
%43 = arith.constant 0 : i32
%44 = arith.cmpi eq, %42, %43 : i32
scf.yield %44 : i1
} else {
%45 = arith.constant false
scf.yield %45 : i1
}
%47 = arith.constant 1 : i1
%46 = arith.xori %41, %47 : i1
cf.cond_br %46, ^bb9, ^bb10
^bb9:
%49 = arith.constant 0 : i32
%50 = llvm.mlir.constant(1 : i64) : i64
%51 = llvm.alloca %50 x i32 : (i64) -> !llvm.ptr
llvm.store %49, %51 : i32, !llvm.ptr
cf.br ^bb12
^bb12:
%52 = llvm.load %51 : !llvm.ptr -> i32
%53 = arith.cmpi sle, %52, %0 : i32
cf.cond_br %53, ^bb13, ^bb14
^bb13:
%54 = arith.constant 0 : i32
%55 = llvm.mlir.constant(1 : i64) : i64
%56 = llvm.alloca %55 x i32 : (i64) -> !llvm.ptr
llvm.store %54, %56 : i32, !llvm.ptr
cf.br ^bb15
^bb15:
%57 = llvm.load %56 : !llvm.ptr -> i32
%58 = arith.cmpi sle, %57, %1 : i32
cf.cond_br %58, ^bb16, ^bb17
^bb16:
%59 = arith.constant 0 : i32
%60 = arith.extsi %59 : i32 to i64
%61 = llvm.mlir.constant(1 : i64) : i64
%62 = llvm.alloca %61 x i64 : (i64) -> !llvm.ptr
llvm.store %60, %62 : i64, !llvm.ptr
%63 = arith.constant 0 : i32
%64 = llvm.mlir.constant(1 : i64) : i64
%65 = llvm.alloca %64 x i32 : (i64) -> !llvm.ptr
llvm.store %63, %65 : i32, !llvm.ptr
cf.br ^bb18
^bb18:
%66 = llvm.load %51 : !llvm.ptr -> i32
%67 = llvm.load %65 : !llvm.ptr -> i32
%68 = llvm.load %30 : !llvm.ptr -> i32
%69 = arith.muli %67, %68 : i32
%70 = arith.cmpi sge, %66, %69 : i32
%71 = scf.if %70 -> (i1) {
%72 = llvm.load %56 : !llvm.ptr -> i32
%73 = llvm.load %65 : !llvm.ptr -> i32
%74 = llvm.load %35 : !llvm.ptr -> i32
%75 = arith.muli %73, %74 : i32
%76 = arith.cmpi sge, %72, %75 : i32
scf.yield %76 : i1
} else {
%77 = arith.constant false
scf.yield %77 : i1
}
cf.cond_br %71, ^bb19, ^bb20
^bb19:
%78 = llvm.load %62 : !llvm.ptr -> i64
%80 = llvm.load %51 : !llvm.ptr -> i32
%81 = llvm.load %65 : !llvm.ptr -> i32
%82 = llvm.load %30 : !llvm.ptr -> i32
%83 = arith.muli %81, %82 : i32
%84 = arith.subi %80, %83 : i32
%85 = arith.extsi %84 : i32 to i64
%86 = arith.muli %85, %7 : i64
%87 = llvm.load %56 : !llvm.ptr -> i32
%88 = llvm.load %65 : !llvm.ptr -> i32
%89 = llvm.load %35 : !llvm.ptr -> i32
%90 = arith.muli %88, %89 : i32
%91 = arith.subi %87, %90 : i32
%92 = arith.extsi %91 : i32 to i64
%93 = arith.addi %86, %92 : i64
%94 = llvm.getelementptr %8[%93] : (!llvm.ptr, i64) -> !llvm.ptr, i64
%79 = llvm.load %94 : !llvm.ptr -> i64
%95 = arith.addi %78, %79 : i64
llvm.store %95, %62 : i64, !llvm.ptr
%96 = llvm.load %65 : !llvm.ptr -> i32
%97 = arith.constant 1 : i32
%98 = arith.addi %96, %97 : i32
llvm.store %98, %65 : i32, !llvm.ptr
cf.br ^bb18
^bb20:
%99 = llvm.load %62 : !llvm.ptr -> i64
%100 = llvm.load %51 : !llvm.ptr -> i32
%101 = arith.extsi %100 : i32 to i64
%102 = arith.muli %101, %7 : i64
%103 = llvm.load %56 : !llvm.ptr -> i32
%104 = arith.extsi %103 : i32 to i64
%105 = arith.addi %102, %104 : i64
%106 = llvm.getelementptr %12[%105] : (!llvm.ptr, i64) -> !llvm.ptr, i64
llvm.store %99, %106 : i64, !llvm.ptr
%107 = llvm.load %56 : !llvm.ptr -> i32
%108 = arith.constant 1 : i32
%109 = arith.addi %107, %108 : i32
llvm.store %109, %56 : i32, !llvm.ptr
cf.br ^bb15
^bb17:
%110 = llvm.load %51 : !llvm.ptr -> i32
%111 = arith.constant 1 : i32
%112 = arith.addi %110, %111 : i32
llvm.store %112, %51 : i32, !llvm.ptr
cf.br ^bb12
^bb14:
%114 = arith.muli %4, %7 : i64
%115 = arith.constant 8 : i32
%117 = arith.extsi %115 : i32 to i64
%116 = arith.muli %114, %117 : i64
%113 = func.call @memcpy(%8, %12, %116) : (!llvm.ptr, !llvm.ptr, i64) -> !llvm.ptr
cf.br ^bb11
^bb10:
cf.br ^bb11
^bb11:
%118 = llvm.load %35 : !llvm.ptr -> i32
%119 = arith.constant 1 : i32
%120 = arith.addi %118, %119 : i32
llvm.store %120, %35 : i32, !llvm.ptr
cf.br ^bb6
^bb8:
%121 = llvm.load %30 : !llvm.ptr -> i32
%122 = arith.constant 1 : i32
%123 = arith.addi %121, %122 : i32
llvm.store %123, %30 : i32, !llvm.ptr
cf.br ^bb3
^bb5:
%124 = llvm.mlir.addressof @str_0 : !llvm.ptr
%126 = arith.extsi %0 : i32 to i64
%127 = arith.muli %126, %7 : i64
%128 = arith.extsi %1 : i32 to i64
%129 = arith.addi %127, %128 : i64
%130 = llvm.getelementptr %12[%129] : (!llvm.ptr, i64) -> !llvm.ptr, i64
%125 = llvm.load %130 : !llvm.ptr -> i64
%131 = llvm.call @printf(%124, %125) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
func.call @free(%8) : (!llvm.ptr) -> ()
func.call @free(%12) : (!llvm.ptr) -> ()
%134 = arith.constant 0 : i32
func.return %134 : i32
}
}