Another Potential CPU Optimization For Mesa: Quadratic Probing

[phoronix]

Phoronix: Another Potential CPU Optimization For Mesa: Quadratic Probing

Mesa developer Thomas Helland is looking at reviving an old set of Mesa patches that could help out in some CPU-bound scenarios...

http://www.phoronix.com/scan.php?page=news_item&px=Mesa-Quadratic-Probing

[oooverclocker]

Every option that will improve the driver code, especially when it reduces the code, is welcome of course. As long as it's predictable that the way this works will ensure a better performance for most titles without decreasing the performance of fewer titles significantly.

[CrystalGamma]

It's really sad that this is the kind of optimization that people have to write, just because every nontrivial C project still has to write their own data structure implementations (which can have their own bugs and all), mostly because C is so poorly suited to generic programming. To be clear, this would not be a problem if this were a specific hashmap used in a specific part of code with special requirements, but this is the hash table used in all of mesa.

[bachchain]

It's really sad that this is the kind of optimization that people have to write, just because every nontrivial C project still has to write their own data structure implementations (which can have their own bugs and all), mostly because C is so poorly suited to generic programming. To be clear, this would not be a problem if this were a specific hashmap used in a specific part of code with special requirements, but this is the hash table used in all of mesa.

so what's the name of your silver bullet language

[haplo602]

It's really sad that this is the kind of optimization that people have to write, just because every nontrivial C project still has to write their own data structure implementations (which can have their own bugs and all), mostly because C is so poorly suited to generic programming. To be clear, this would not be a problem if this were a specific hashmap used in a specific part of code with special requirements, but this is the hash table used in all of mesa.

With great power comes great responsibility ... C definitely gives you that power ... However not many coders are up to the responsibility ...

[quaz0r]

With great power comes great responsibility ... C definitely gives you that power ... However not many coders are up to the responsibility ...

Noob, having absolutely no clue you of course missed the point and instead got it totally backwards. It isn't that C is just so powerful it takes a great mystical guru to unleash its glory. To the contrary, the point here is that C is ancient and severely lacking, and that there are more modern and featureful alternatives.

[damg]

It's really sad that this is the kind of optimization that people have to write, just because every nontrivial C project still has to write their own data structure implementations (which can have their own bugs and all), mostly because C is so poorly suited to generic programming.

Well there's no lack of C libraries implementing various data structures (both of the "void *" type and "specialized structs via macro expansion" type) so it's not true that every C project has to write their own. But in this case, seeing as Mesa was first released in 1993, I'm guessing many of these options weren't around when this project got started. Also, having code that you can tweak for your own data/workload can be nice, especially when it's only a few hundred lines like this hash-table implementation.

[marek]

It's really sad that this is the kind of optimization that people have to write, just because every nontrivial C project still has to write their own data structure implementations (which can have their own bugs and all), mostly because C is so poorly suited to generic programming. To be clear, this would not be a problem if this were a specific hashmap used in a specific part of code with special requirements, but this is the hash table used in all of mesa.

This is an interesting point, but the C language is not to blame. Programmers are, because they love re-inventing stuff. Also, standard libraries are not always optimized or designed to be fast in the first place.

[funfunctor]

Rust ! By todays standards in language evolution, C is pretty darn shit tbh and quaz0r is very much correct above.

[SaucyJack]

so what's the name of your silver bullet language

..C++