[dev-servo] Planning to experiment with inline storage for DOMString / webidl binding codegen
One of the largest performance bottlenecks in Element.getAttribute (and I suspect a bunch of other webidl bindings as well) is copying (often very small) SpiderMonkey strings into heap allocated DOMStrings. It doesn't seem practical to always use JS strings as they don't use UTF-8 and it would require a lot more rooting and generally seems like it would require rewriting a large amount of code. But, it was pointed out that these (often very small) strings are always heap allocated. See #6906 and specifically these comments: https://github.com/servo/servo/issues/6906#issuecomment-159372789 https://github.com/servo/servo/issues/6906#issuecomment-159379936 https://github.com/servo/servo/issues/6906#issuecomment-159383632 I've created a crate[0] that provides a mostly drop-in replacement of std::string::String but with inline storage for small strings to avoid heap allocation. It also provides a StringExt trait to abstract string operations over both InlinableString and std::string::String. I'm planning on implementing this trait for DOMString as well, making the codegen emit bindings code that uses InlinableString for stack allocated strings, and make the webidl interface methods use `StringExt` instead of `DOMString` directy. My hope is that this will make a good chunk of that performance bottleneck go away. [0] https://github.com/fitzgen/inlinable_string If I bug you on irc with requests for help, please have patience with me :) If you see something architecturally wrong with this plan, please let me know sooner rather than later! Cheers, Nick ___ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
Re: [dev-servo] Planning to experiment with inline storage for DOMString / webidl binding codegen
> One of the largest performance bottlenecks in Element.getAttribute (and I > suspect a bunch of other webidl bindings as well) is copying (often very > small) SpiderMonkey strings into heap allocated DOMStrings. Alan Jeffrey has been investigating issues around string copying as well. Have you chatted with him to compare notes? jack. ___ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
Re: [dev-servo] Planning to experiment with inline storage for DOMString / webidl binding codegen
On Tue, Dec 1, 2015 at 5:48 PM, Jack Moffitt wrote: > > One of the largest performance bottlenecks in Element.getAttribute (and I > > suspect a bunch of other webidl bindings as well) is copying (often very > > small) SpiderMonkey strings into heap allocated DOMStrings. > > Alan Jeffrey has been investigating issues around string copying as > well. Have you chatted with him to compare notes? > I was not aware of that. I'm interested in Alan's initial findings and approach as well! (Pung for more info here: https://github.com/servo/servo/issues/6906#issuecomment-161153975) ___ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
Re: [dev-servo] Planning to experiment with inline storage for DOMString / webidl binding codegen
Yes, I've done some experiments with this. Inlining small strings didn't help as much as I'd hoped, I suspect because they're being converted to Strings or JSStrings a lot. I'm still in the middle of experimenting, I'm hoping to write up soon. The code is at https://github.com/asajeffrey/servo/tree/string_representation_experiments although it's a bit of a mess. A. On Tue, Dec 1, 2015 at 7:48 PM, Jack Moffitt wrote: > > One of the largest performance bottlenecks in Element.getAttribute (and I > > suspect a bunch of other webidl bindings as well) is copying (often very > > small) SpiderMonkey strings into heap allocated DOMStrings. > > Alan Jeffrey has been investigating issues around string copying as > well. Have you chatted with him to compare notes? > > jack. > ___ > dev-servo mailing list > dev-servo@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-servo > ___ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
Re: [dev-servo] Planning to experiment with inline storage for DOMString / webidl binding codegen
On 02/12/15 02:39, Nick Fitzgerald wrote: I've created a crate[0] that provides a mostly drop-in replacement of std::string::String but with inline storage for small strings to avoid heap allocation. [0]https://github.com/fitzgen/inlinable_string There’s also [1], already in used in Servo in some places, which behaves like Vec but stores items inline up to a certain length. I’ve also been playing with [2] to be like String but with a generic backing storage. (I’ve used it in [3] to return a single code point in UTF-8 as StringWrapper<[u8; 4]>.) Would it make sense to converge with some of these? [1] https://github.com/servo/rust-smallvec [2] https://github.com/SimonSapin/rust-std-candidates/blob/master/string-wrapper/lib.rs [3] https://github.com/SimonSapin/rust-utf8 -- Simon Sapin ___ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
