peltorator's blog

Ok, lets talk about cout.tie once and forever
By peltorator, 4 years ago, In English

When you use C++ and the input is really big you can't just use cin and cout. You need to speed up it with

ios::sync_with_stdio(0);
cin.tie(0);

Someone argues that the second line is unnecessary but it's not true. if the input and output alternate then adding the second line makes I/O more than twice faster. But then... Someone argues that we also need to use cout.tie(0).

I personally never use this and I don't know any case where it can help. So my question for today is the following: "Is there any case where we actually need cout.tie? Or is it completely useless?"

  • Vote: I like it
  • +276
  • Vote: I do not like it

Comments Comments (42)
| Write comment?
»
4 years ago, # |
  Vote: I like it +90 Vote: I do not like it

I have read several comments related to it by a user -is-this-fft- and I quote "cout.tie(NULL) does literally nothing because cout is already tied to NULL", "It sounds very strange when you consider what tie does. Every input or output stream in C++ is tied to an ostream or to null. Tyig cin to null means it's not tied to cout anymore which can speed things up. But cout is generally already tied to null."

Link to the comments — https://codeforces.me/blog/entry/87419?#comment-756655, https://codeforces.me/blog/entry/83672?#comment-709924

→ Reply
  • »
    »
    4 years ago, # ^ |
    ← Rev. 2   Vote: I like it +479 Vote: I do not like it

    Yes, you are correct.

    I think it would be good to reiterate what tie does because I see a lot of people copying these lines with only a vague understanding of what they do.

    Every stream in C++ is tied to an output stream, which can be null.

    What does this mean? First of all, it's important to understand that when you write std::cout << "asdf", it is not necessarily immediately printed on the screen. It turns out that it is much better (in terms of performance) to collect it into a buffer and then, at some point, flush the buffer — i.e. empty its contents to the screen (or file, or any other "device"), all at once.

    But now, consider the following. You are developing a console application and write something like:

    1 std::cout << "Please enter your age: ";
2 int age;
3 std::cin >> age;

    What would happen if std::cout didn't get flushed before line 3? The program would expect a reply from the user, but the user hasn't even had a chance to read the question. In an application like this, it would be a good idea if std::cout was somehow automatically flushed when we try to read from std::cin.

    And this is exactly the problem tie solves. If you have two streams fin and fout, and fin is tied to fout, then fout is automatically flushed when you try to read from fin. So what are the defaults? What is tied to what? Quoting the C++ reference:

    By default, the standard narrow streams cin and cerr are tied to cout, and their wide character counterparts (wcin and wcerr) to wcout. Library implementations may also tie clog and wclog.

    Now it becomes clear why people recommend using cin.tie(0) in competitive programming. Suppose you are solving a query problem like this.

    1 MyAwesomeDataStructure ds;
2 for (int i = 0; i < queryc; i++) {
3   Query q;
4   cin >> q;
5   cout << ds.solve(q) << '\n';
6 }

    If you didn't use cin.tie(0), we would flush cout every time we hit line 4; this is essentially as bad as writing endl on line 5. Concerns about the user not seeing the question are also irrelevant now because we are given the entire input at once.

    The reference doesn't explicitly say that cout isn't tied to anything, but on every C++ compiler I tried, cout << cout.tie() << endl; outputs 0. Also, even if you are using some strange compiler that ties cout to some other stream sout, you would only notice the performance hit if you wrote a lot to sout which doesn't really happen in competitive programming.

    So yes, cout.tie(0) doesn't do anything and in some sense it's a sign that the user doesn't understand cin.tie(0) very well either.

    → Reply
    • »
      »
      »
      4 years ago, # ^ |
        Vote: I like it -140 Vote: I do not like it

      While you are probably right in the sense that what you are saying is correct, I think it is a wrong way to understand cin.tie(0) in competitive programming. And the last line from your comment for me proves that you are wrong. The last line reads like the fact that "the user doesn't understand it" is bad somehow. I don't agree.

      cin.tie(0) is just a bullshit line to make cin faster. I don't give a fuck what it does inside or how it works or why it even exists. The same with using namespace std — it's just something that stupid computer needs to understand you, that's it.

      → Reply
      • »
        »
        »
        »
        4 years ago, # ^ |
          Vote: I like it +296 Vote: I do not like it

        Well, I don't think it is very bad to not know what tie does. Still, I prefer to know these things to prevent any unpleasant surprises.

        → Reply
        • »
          »
          »
          »
          »
          4 years ago, # ^ |
            Vote: I like it +446 Vote: I do not like it

          Sorry, you answered a question that Um_nik wasn't interested in knowing the answer to. Please apologize for wasting his time.

          → Reply
          • »
            »
            »
            »
            »
            »
            4 years ago, # ^ |
              Vote: I like it -205 Vote: I do not like it

            hi

            → Reply
          • »
            »
            »
            »
            »
            »
            4 years ago, # ^ |
              Vote: I like it +10 Vote: I do not like it

            for me Um_nik is a legend who doesn't gives a F about anything and when he does he F's that thing

            → Reply
          • »
            »
            »
            »
            »
            »
            4 years ago, # ^ |
              Vote: I like it +100 Vote: I do not like it

            No, it was an interesting read. Useless, but interesting.

            → Reply
            • »
              »
              »
              »
              »
              »
              »
              4 years ago, # ^ |
                Vote: I like it +10 Vote: I do not like it

              Classic um_nik way lol : useless but interesting

              → Reply
            • »
              »
              »
              »
              »
              »
              »
              4 years ago, # ^ |
                Vote: I like it +20 Vote: I do not like it

              Pretty much the same as CP. Right ?

              → Reply
              • »
                »
                »
                »
                »
                »
                »
                »
                4 years ago, # ^ |
                  Vote: I like it +43 Vote: I do not like it

                That depends on how you define usefulness.

                → Reply
        • »
          »
          »
          »
          »
          4 years ago, # ^ |
          ← Rev. 2   Vote: I like it +18 Vote: I do not like it

          Thanks. Now I completely understand why using tie would recieve Idleness Limit Exceeded for those interactive problems. :D

          → Reply
      • »
        »
        »
        »
        4 years ago, # ^ |
          Vote: I like it +21 Vote: I do not like it

        Um_nik, a guy with zero haters

        → Reply
    • »
      »
      »
      4 years ago, # ^ |
        Vote: I like it +66 Vote: I do not like it

      Why do we still need to flush the buffer for interactive problems if one has not added cin.tie(0)?

      → Reply
      • »
        »
        »
        »
        4 years ago, # ^ |
          Vote: I like it +44 Vote: I do not like it

        You know, I'm starting to think that you don't. I just went and resubmitted some solved interactive problems, removing any endl (those flush the output) and making sure there is no cin.tie(0). All of them got Accepted. Then I added cin.tie(0) and got (I think for the first time ever) the Idleness Limit Exceeded verdict.

        → Reply
    • »
      »
      »
      4 years ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      Your answer is very good but I mistakenly voted against, I'm sorry...

      → Reply
  • »
    »
    4 years ago, # ^ |
      Vote: I like it +26 Vote: I do not like it

    Thank you!

    → Reply
»
4 years ago, # |
  Vote: I like it -86 Vote: I do not like it
Do you agree?
→ Reply
  • »
    »
    4 years ago, # ^ |
      Vote: I like it +1 Vote: I do not like it

    I've seen your comments a lot lately on a lot of blogs. Now I think you're the one seeking +ve contribution.

    → Reply
»
4 years ago, # |
  Vote: I like it +33 Vote: I do not like it

cout.tie(0) is not necessary, because by default cin ties with cout, but cout ties with nullptr so it's basically does nothing.

You can check it by code:

Spoiler
→ Reply
»
4 years ago, # |
  Vote: I like it +8 Vote: I do not like it

If you do not properly understand, what cin.tie(smth) and cout.tie(smth) do, you might want to write them all together. It doesn't slow your code down (does it?) and it decreases the chance you mix them up and write only cout.tie(nullptr) instead of cin.tie(nullptr).

I have somehow memorized that cout.tie(nullptr) is useless and excluded it from my pre-written pattern. I also do not write it in competitions that doesn't allow participants to use pre-written code. I don't know a simple way to remember, which string is useless (is cin kind of tied to cout, or is it the opposite?). So, I have just realised, it is better to write both of them.

→ Reply
  • »
    »
    4 years ago, # ^ |
      Vote: I like it +18 Vote: I do not like it

    You're absolutely right. This question isn't about practice usage. I'm just interested in this.

    → Reply
»
4 years ago, # |
  Vote: I like it +1 Vote: I do not like it

Should we use cin.tie(0); in Interactive Problems where next input depends on what we print using cout?

→ Reply
  • »
    »
    4 years ago, # ^ |
      Vote: I like it +5 Vote: I do not like it

    I think there isn't actually any need for fast IO in interactive problems most of the time because we use flush and it's really slow. But at the same time, there is no reason not to use fast IO always.

    → Reply
»
4 years ago, # |
  Vote: I like it -49 Vote: I do not like it

"Once and forever".... What a joke)
There still exists people that don't know that comparator must return true if and only if lhs is strictly less than rhs.

cout.tie in competitive programming is useless AFAIK. But it is usefull when you need to guarantee that you write in tied output strictly before cout. But I don't know when there is need for it;)

→ Reply
»
4 years ago, # |
  Vote: I like it +1 Vote: I do not like it

Well, I also had a question related to this, that Is their any difference btw ios_base::sync_with_stdio(false); cin.tie(NULL); and cin.tie(0)->sync_with_stdio(0);(i saw this in Benq's template)

→ Reply
  • »
    »
    4 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Both do same
    But I'd say first is preferable since second way is confusing. sync_with_stdio is static method and it supposed to be called as static not as member of cout.

    → Reply
»
13 months ago, # |
  Vote: I like it 0 Vote: I do not like it

I am not sure of the logic behind it, but on a dp question on cses, my code gave runtime error without cout.tie(NULL) and got accepted when i added that line. https://cses.fi/paste/2f6f0fdf60a71e537a397d/ try submitting this same code without the cout.tie(NULL) line. it should give you a good idea

→ Reply
  • »
    »
    13 months ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    I have tried to submit the linked code with and without cout.tie(NULL) and it was accepted both times. No ideas where to find a good idea :)

    → Reply
    • »
      »
      »
      13 months ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      my bad, my code gave runtime error cause i used long long instead of int. don't know why it is, but you're right cout.tie(0) isnt the problem.

      → Reply
  • »
    »
    13 months ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Can you please provide the problem link?

    → Reply
»
4 months ago, # |
  Vote: I like it +43 Vote: I do not like it

With Problem 2005-E2 - Subtangle Game (Hard Version), Codeforces is now openly telling people to use cout.tie(NULL);. I urge its authors, BiNARyBeastt and Tsovak, to read this blog and avoid spreading misinformation.

TLDR: Do not use cout.tie(0) since it does nothing.

→ Reply
  • »
    »
    4 months ago, # ^ |
      Vote: I like it +43 Vote: I do not like it

    Quoting purplesyringa:

    // If you use std::cout.tie(nullptr);, you are wrong.
// This won't ever be supported.
// You don't need this line. Remove it.
// You have almost certainly added 12 useless bytes to every solution. They take space on disk.
// Disks are produced by young children on factories. Do you want children to work more because
// of you? Think of the children.
// And about yourself too. One of the pictures on this page is loaded from my domain. I know your IP.
    → Reply
  • »
    »
    4 months ago, # ^ |
      Vote: I like it +17 Vote: I do not like it

    Thanks for the information. I've been using that in my template all the time lol. I'll change it on polygon soon.

    → Reply
  • »
    »
    13 days ago, # ^ |
    ← Rev. 3   Vote: I like it +42 Vote: I do not like it

    Recently there was another problem statement (2053G - Naive String Splits) by Cocoly1990 and Caylex telling people to use cout.tie(0).

    Cocoly1990 and Caylex, please do not tell people to use cout.tie(0) in C++ as it does absolutely nothing! cout.tie(0) is like the homeopathy of competitive programming.

    → Reply
    • »
      »
      »
      13 days ago, # ^ |
        Vote: I like it -57 Vote: I do not like it

      TL; DR I think it's only a matter of personal taste. I'm sorry for not knowing it before the contest, but after I learned about it, I still don't see any reason that I must delete it. Can you elaborate on how much negative effect it has given the codeforces community? I think that equals to zero.

      First of all, you've notified the wrong people. Cocoly is in Grade 12 and studying hard for the College Entrance Exam (known as Gaokao), and Caylex has never had access to the contest on Polygon.

      Secondly, this is not to "tell people to use xxx". I've never said this is the only way to fasten your input or output speed (and, even, some solutions without this buff can also pass comfortably).

      Thirdly, its doing absolutely nothing is not the reason that I should definitely delete it. If you really have such mysophobia, please ping the authors of Atcoder rounds not to display main correct solutions with tedious templates, as most do absolutely nothing if you are only learning the solution to a single problem.

      By the way, the coordinators of the two rounds are the same and it was Scrasse who sent me the message to add a notice similar to the 2005E2 problem. I assume that he has also seen the issue raised by you, but he didn't think it very important to remove cout.tie either.

      → Reply
      • »
        »
        »
        »
        13 days ago, # ^ |
        ← Rev. 2   Vote: I like it +43 Vote: I do not like it

        I don't think "what about AtCoder" argument stands... Similar situations would often turn out not so similar on a closer look, and in this particular case, templates are not part of the statement, don't explicitly encourage people to use specific template, and more importantly, they are useful in some scenarios, even if not in the specific problem.

        But actually I'd also support removing the templates when they're clearly not needed, but that's beyond the point. It's just that templates have less perceived negative impact and, more importantly, removing them might be a non-trivial task if the solution uses only small useful part of the template and you'd like to get rid of useless parts. So, keeping them has at least some reasonable arguments, unlike keeping cout.tie(0).

        I think the main reason not to use cout.tie(0) is that using it promotes a cargo cult-like approach to coding, where you just treat some commands as magic without putting much consideration into what they actually do. Not like removing it immediately promotes better practices, but at least it doesn't promote the bad ones that openly.

        In any case, please treat it as a request to follow certain best practices, and not as a personal attack. Putting it there is not a stylistic preference like using tabs vs spaces or 0 vs NULL vs nullptr. I really see no viable argument to keep using cout.tie, other than the fact that pointing it out is a criticism and it in itself might trigger a defensive response.

        → Reply
      • »
        »
        »
        »
        13 days ago, # ^ |
        ← Rev. 2   Vote: I like it +27 Vote: I do not like it

        It seems that many people actually care, so I've just updated the statement of 2005E2 - Subtangle Game (Hard Version). For some reason, I see the new statement on Polygon but I still see the old statement on Codeforces.

        I cannot update the statement of 2053G - Naive String Splits because the problem is too "heavy" at the moment (high time limit, many solutions and tests) and Polygon does not allow me to create a package without exceeding my quota :D

        → Reply
        • »
          »
          »
          »
          »
          12 days ago, # ^ |
            Vote: I like it 0 Vote: I do not like it

          I think you can build without verification.

          → Reply
          • »
            »
            »
            »
            »
            »
            12 days ago, # ^ |
              Vote: I like it +11 Vote: I do not like it

            Oh, the problem does not pass verification right now. Can you take a look? _istil

            → Reply
            • »
              »
              »
              »
              »
              »
              »
              12 days ago, # ^ |
                Vote: I like it -42 Vote: I do not like it

              I don't want to take a look because it's clear that Codeforces community doesn't welcome my preparation.

              I'm not going to explain why I was sending the comment above more, as I don't want to open a blog writing about oh wow exposure of some stupid Chinese for gaining nonsense contributions on Codeforces and making people (i.e. me) feel disgusting for fun.

              Please downvote this comment as much as you can.

              → Reply