Linux 5.7 Netfilter To See AVX2 Optimizations For Big Performance Boost - Can Be Up To ~420%

[phoronix Administrator Find all posts View Profile Close phoronix]

Phoronix: Linux 5.7 Netfilter To See AVX2 Optimizations For Big Performance Boost - Can Be Up To ~420%

Linux 5.7's Netfilter framework is set to see better performance on modern Intel and AMD systems thanks to AVX2 optimizations...

http://www.phoronix.com/scan.php?pag...lter-AVX2-Perf

[Raka555 Junior Member Find all posts View Profile Close Raka555]

I assume it is hand optimized assembler ?

[boxie Senior Member Find all posts View Profile Close boxie]

Well, that's some... Blazing... Performance

[tiennou Junior Member Find all posts View Profile Close tiennou]

I assume it is hand optimized assembler ?

A look at the commit (https://git.kernel.org/pub/scm/linux...94d765c8eecbe1) points in that direction.

[cl333r Senior Member Find all posts View Profile Close cl333r]

To me as a desktop user this is just gibberish, as about all of the "news" on lxer and linuxtoday.

[cbxbiker61 Senior Member Find all posts View Profile Close cbxbiker61]

It's always welcome to see performance improvements. Optimizing arm would no doubt hit a larger user base with all of the arm based OpenWRT DdWRT routers (I've got 5 routers running OpenWRT and one of them would benefit from netfilter performance improvements).

[Setif Senior Member Find all posts View Profile Close Setif]

I hope It's not about some operations that occur once or twice a day, changed from taking 4.2 ms to 1.0 ms (420%).

[r08z Phoronix Member Find all posts View Profile Close r08z]

This is what ClearLinux does on a regular basis for all libraries and programs with a few simple avx2 instricts patches to help the program make better use of the -march=haswell compiler flag.

[milkylainen Senior Member Find all posts View Profile Close milkylainen]

This is what ClearLinux does on a regular basis for all libraries and programs with a few simple avx2 instricts patches to help the program make better use of the -march=haswell compiler flag.

Vectorization and architecture support is not the same as structured optimized assembly for complex data.
But if the code structure was different, maybe more standard vectorization would have helped.
Replacing a functions or calls with a hand optimized intrinsic is not the same either.

[ldesnogu Senior Member Find all posts View Profile Close ldesnogu]

It's always welcome to see performance improvements. Optimizing arm would no doubt hit a larger user base with all of the arm based OpenWRT DdWRT routers (I've got 5 routers running OpenWRT and one of them would benefit from netfilter performance improvements).

It seems the author agrees:
https://git.kernel.org/pub/scm/linux...94d765c8eecbe1

A similar strategy could be easily reused to implement specialised versions for other SIMD sets, and I plan to post at least a NEON version at a later time.