#!/usr/bin/env python3

(n, k), w = map(int, input().split()), list(map(int, input().split())),
ans, cnt = 0, [0] * k

# modulo
for x in w:
    if x % k == 0: ans += 1
    else: cnt[x % k] += 1

# pairs
for i in range(1, k // 2):
    x = min(cnt[i], cnt[k - i])
    cnt[i] -= x
    cnt[k - i] -= x
    ans += x
if k % 2 == 0:
    x = cnt[k // 2] // 2
    cnt[k // 2] -= 2 * x
    ans += x

# Left with at most 3 non-empty values, and possibly k/2.
ans = {tuple([0] * k): ans}


def calc(cnts):
    if tuple(cnts) in ans: return ans[tuple(cnts)]
    # Take the number in the set with the lowest count, and add something to it
    m = min((100, -1), *((cnts[m], m) for m in range(1, k) if cnts[m] > 0))[1]

    new_cnts, best = cnts[:], 0
    new_cnts[m] -= 1
    for i in range(1, k):
        if new_cnts[i] > 0:
            new_cnts[i] -= 1
            v = 0
            if m + i != k: new_cnts[(m + i) % k] += 1
            else: v += 1
            v += calc(new_cnts)
            if m + i != k: new_cnts[(m + i) % k] -= 1
            best = max(best, v)

            new_cnts[i] += 1

    ans[tuple(cnts)] = best
    return best


print(sum(w) // k - calc(cnt))