Mesa Looks At Switching To Jemalloc For Faster Performance

[pal666]

Lockless malloc is very simple and it works. It just doesn't need further development.

i'm sure it works, but does it work better than jemalloc and why mozilla didn't switch to it?

It's thread-safe without a need for locking by using inline assembler statements for the critical parts.

you don't get thread safety for free by switching to assembler. maybe you wanted to say it uses atomic operations in lock-free manner, but so are other mallocs

[pal666]

If pre-built mesa doesn't require jemalloc as a dependency

what is this 'pre-built' thing? your distro package? why would you care?

[pal666]

[QUOTE=schmidtbag;n901299If pre-built mesa doesn't require jemalloc as a dependency[/QUOTE]what is this 'pre-built' thing? your distro package? why would you care?

[sdack]

Lockless 1.2 was released over 4 years ago; jemalloc maybe a better option due to its continued development and use in other applications (i.e. more testing).

*lol*

So it was released a while back. What do you think it does and why do you expect it to be so much more complicated? It's a only few functions and the API is ancient and very simple.

Most allocators follow a complex implementation and therefore require continued development. Not just for the algorithms but for the configuration and build support, too. Lockless malloc is very simple and it works. It just doesn't need further development. It comes as a single C file with a few header files. It's thread-safe without a need for locking by using inline assembler statements for the critical parts. Sure, it doesn't come with a configure script of 10,000 lines, but why is this important when you can have something much simpler?

[Mystro256]

Did they test it with locklessmalloc, too? http://locklessinc.com/benchmarks_allocator.shtml

Lockless 1.2 was released over 4 years ago; jemalloc maybe a better option due to its continued development and use in other applications (i.e. more testing).

[Mystro256]

Whoa. the horror, a 10KB dependency used by a bunch of software.

Haha best response ever.

[sdack]

Did they test it with locklessmalloc, too? http://locklessinc.com/benchmarks_allocator.shtml

[cj.wijtmans]

From the mailing list post:

"TODO: The jemalloc dependency should be added to configure.ac before this."

I'm assuming it'll need to be linked, thus another dependency, but I'm just guessing.
Beyond that, if it's indeed used in a bunch of other projects such as Firefox, this is a non issue for most users.

Whoa. the horror, a 10KB dependency used by a bunch of software.

[Mystro256]

Out of curiosity, does this increase the amount of dependencies for users (with or without compiling mesa themselves)? If pre-built mesa doesn't require jemalloc as a dependency, then I'd say this performance gain is pretty awesome - always nice being able to get a substantial performance increase without sacrificing much.

From the mailing list post:

"TODO: The jemalloc dependency should be added to configure.ac before this."

I'm assuming it'll need to be linked, thus another dependency, but I'm just guessing.
Beyond that, if it's indeed used in a bunch of other projects such as Firefox, this is a non issue for most users.

[schmidtbag]

Out of curiosity, does this increase the amount of dependencies for users (with or without compiling mesa themselves)? If pre-built mesa doesn't require jemalloc as a dependency, then I'd say this performance gain is pretty awesome - always nice being able to get a substantial performance increase without sacrificing much.