Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Facebook's std::vector optimization

github.com

Post reply on HN
Print view View original

51–60 of 93 posts

Re: Facebook's std::vector optimization

#51

Post by chickenandrice »

otterley wrote: post #48
chickenandrice wrote: post #36

Greetings Facebook, several decades ago welcomes you. Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. Most of the time if performance and allocation is so critical, you're better off not using a vector anyway. A fixed sized array is much more cache friendl…

> Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. OTOH, game programmers don't share this kind of knowledge on public fora as freely as Internet companies do. It may be "old news" to you, but it was unpublished. Knowledge is great, but dissemination is eve…

Actually, game companies as well as various indie developers have widely distributed this knowledge. Some examples of places to look/things to search for implementations or similar/better:

-GDC Slide Decks -Game Programming Data Structure/Performance Books -Game Industry Publications - ex: Game Programming Gems -EA STL -Game Dev Sites -Any available source to various engines/games in the last many years should have more than a few inside

Moreover, this is also a common thing in a lot of embedded and high-performance development in general.

Just because someone didn't make a github project and post it on HN doesn't mean it doesn't exist. This seems to be a trend with web developers - reinventing what exists without properly checking what is there, and doing so usually poorly. I say this as someone who has done game development and web development at various points of my career (currently web by day, game by night) so I am not trying to be condescending, just pointing out an unfortunate trend.

Downvote if you want, but I see the first game dev who replied already confirmed what I said.

Re: Facebook's std::vector optimization

#52

Post by chickenandrice »

DonPellegrino wrote: post #41
dbaupp wrote: post #39

Earlier quoted context omitted.

> More to the point, who would use a c++ library from Facebook? Hopefully don't need to explain the reasons here. Could you explain them for those of us not in the loop? Does Facebook have a bad reputation for C++?

I would also like expanations, because Facebook actually has a good reputation when it comes to their compiled languages engineers. Their C++ and D engineers built HHVM/Hack, their OCaml engineers built some great analysis tools and much of the supporting code for the HHVM/Hack platform, etc., the list goes on, so I'd like to know why someone would want to avoid their C++ library based on the "Facebook" name only.

Because Facebook also has a reputation of not playing nice with people, the rules, intellectual property, and so on. This is hardly a company anyone should support or trust and if you can't figure that out, I can't help you.

As far as their work on HHVM, it was necessary due to failure by bad technology choices from the start. There's very little interesting about this work unless you somehow love PHP, want to make debugging your production applications more difficult, and refuse to address your real problems. I am 100% sure no one outside of the PHP community cares about anything Facebook has done in C++.

Simply having a large company with lots of developers who might have even had good reputations elsewhere or even be smart doesn't mean much. Having worked in many places with lots of smart developers, I can tell you stories about too many geniuses in the room. Calling Facebook developers engineers is also about as apt as calling janitors sanitation engineers. We're programmers, or developers, or perhaps software architects at best depending on the position. I happen to have an EE and CS degree but given I do programming for a living, I'd hardly call myself an engineer. But we're way off topic :)

Re: Facebook's std::vector optimization

#53

Post by to3m »

mraison wrote: post #47
chickenandrice wrote: post #36

Greetings Facebook, several decades ago welcomes you. Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. Most of the time if performance and allocation is so critical, you're better off not using a vector anyway. A fixed sized array is much more cache friendl…

> most of us had in-house stl libraries for decades now I've no doubt there has been many talented programmers for decades, but what about open-sourcing these "in house" libraries if they are so great, to avoid countless other programmers the need to reinvent the wheel? I think it would be more constructive than bashing Facebook for being late to the party.

I personally believe that "decades" would be an exaggeration.

As to why people don't open source their libraries - it's a big hassle and doesn't especially benefit them personally. Indeed if other people are going to have to spend time reinventing the wheel I suppose some might even see it as a competitive advantage. And as a general rule, video games companies don't seem to be much into open source anyway.

(Additionally, if the ones I've used are anything to go by, the libraries aren't so great. They attack specific problems with std::vector, and that's fine, but usually not in any general way. No need to do a full extension of the allocator mechanism if you know you can just call your company's allocator's special try_realloc function, for example - but now of course nobody else can use it. And there usually isn't any impetus to "fix" this, because it isn't actually a problem.)

Re: Facebook's std::vector optimization

#54

Post by chickenandrice »

chickenandrice wrote: post #51
otterley wrote: post #48

Earlier quoted context omitted.

> Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. OTOH, game programmers don't share this kind of knowledge on public fora as freely as Internet companies do. It may be "old news" to you, but it was unpublished. Knowledge is great, but dissemination is eve…

Actually, game companies as well as various indie developers have widely distributed this knowledge. Some examples of places to look/things to search for implementations or similar/better: -GDC Slide Decks -Game Programming Data Structure/Performance Books -Game Industry Publications - ex: Game Programming Gems -EA STL -Game Dev Sites -Any available source to various engines/games in the last many years should have m…

http://lmgtfy.com/?q=std%3A%3Avector+games

Re: Facebook's std::vector optimization

#55

Post by otterley »

chickenandrice wrote: post #51
otterley wrote: post #48

Earlier quoted context omitted.

> Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. OTOH, game programmers don't share this kind of knowledge on public fora as freely as Internet companies do. It may be "old news" to you, but it was unpublished. Knowledge is great, but dissemination is eve…

Actually, game companies as well as various indie developers have widely distributed this knowledge. Some examples of places to look/things to search for implementations or similar/better: -GDC Slide Decks -Game Programming Data Structure/Performance Books -Game Industry Publications - ex: Game Programming Gems -EA STL -Game Dev Sites -Any available source to various engines/games in the last many years should have m…

I agree that there are other public fora than company-sponsored blog posts, but I don't see your industry contributing its accumulated knowledge back to the library developers, either (assuming you're relying on open-source C++ libraries).

If this truly was a well-known issue long ago, it should also have been fixed long ago by game developers contributing back to the open-source developers who have made their work possible so that the rest of us didn't have to rediscover these performance issues.

If you want to avoid this kind of criticism, be a part of the solution outside your own industry walls; don't just brag post hoc about how you solved it in your castle for yourselves.

Re: Facebook's std::vector optimization

#56

Post by otterley »

chickenandrice wrote: post #52
DonPellegrino wrote: post #41

Earlier quoted context omitted.

I would also like expanations, because Facebook actually has a good reputation when it comes to their compiled languages engineers. Their C++ and D engineers built HHVM/Hack, their OCaml engineers built some great analysis tools and much of the supporting code for the HHVM/Hack platform, etc., the list goes on, so I'd like to know why someone would want to avoid their C++ library based on the "Facebook" name only.

Because Facebook also has a reputation of not playing nice with people, the rules, intellectual property, and so on. This is hardly a company anyone should support or trust and if you can't figure that out, I can't help you. As far as their work on HHVM, it was necessary due to failure by bad technology choices from the start. There's very little interesting about this work unless you somehow love PHP, want to make d…

> As far as their work on HHVM, it was necessary due to failure by bad technology choices from the start.

HHVM arose out of Facebook's desire to save on server purchasing and operating costs. Facebook could run perfectly well without it on Plain Old PHP, but they'd have to buy and power more servers.

I'd hardly call PHP a "bad technology choice" given the outstanding financial success of many companies that use it.

Re: Facebook's std::vector optimization

#57

Post by shin_lao »

I think the Folly small vector library is much more interesting and can yield better performance (if you hit the sweet spot).

https://github.com/facebook/folly/blob/master/folly/docs/sma...

From what I understand, using a "rvalue-reference ready" vector implementation with a good memory allocator must work at least as good as FBVector.

Re: Facebook's std::vector optimization

#58

Post by taejo »

merraksh wrote: post #40
kevin_thibedeau wrote: post #38

Earlier quoted context omitted.

It is still weird. * * is the other established convention used for exponentiation in languages like Python, Ada, m4, and others.

But * * has an entirely different meaning in C/C++.

Only as a prefix operator; * has a pointer-related meaning for prefix and an arithmetic meaning for infix, there's no reason * * shouldn't be the same.

Re: Facebook's std::vector optimization

#59

Post by micv »

otterley wrote: post #48
chickenandrice wrote: post #36

Greetings Facebook, several decades ago welcomes you. Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. Most of the time if performance and allocation is so critical, you're better off not using a vector anyway. A fixed sized array is much more cache friendl…

> Game programmers figured out the same and arguably better ways of doing this since each version of std::vector has been released. This is but a small reason most of us had in-house stl libraries for decades now. OTOH, game programmers don't share this kind of knowledge on public fora as freely as Internet companies do. It may be "old news" to you, but it was unpublished. Knowledge is great, but dissemination is eve…

EA open sourced parts of their in-house STL. A Github clone can be found at:

https://github.com/paulhodge/EASTL

I haven't looked at it myself but it might be an interesting start for those interested.

Post reply on HN