The Liquorix Kernel Is Still Ticking, Currently Based On Linux 4.1

[phoronix]

Phoronix: The Liquorix Kernel Is Still Ticking, Currently Based On Linux 4.1

As part of a recent wave of requests for new benchmarks you'd like to see on Phoronix, a reader had reminded me of the Liquorix kernel...

http://www.phoronix.com/scan.php?page=news_item&px=Liquorix-Kernel-Linux-4.1

[Norway]

Yes please!

[Dick Palmer]

Yes please!

Likewise.

With the comparison including a plain unpatched 4.1 kernel and a plain unpatched 4.1 which has been built with all applicable configs ported from Liquorix if there's a trivial way to achieve that (oldconfig or similar?).

[euspectre]

Yes, benchmarking that kernel would be interesting. Esp. - disk-related benchmarks because of BFQ, etc. Graphics/Video Playback/Gaming is interesing too.

[RealNC]

Is this actually benchmarkable using Phoronix? I thought the point of kernels such as this is to sacrifice performance for better interactivity (less latency spikes and such.)

[AnAkIn]

Yes, benchmarking that kernel would be interesting. Esp. - disk-related benchmarks because of BFQ, etc. Graphics/Video Playback/Gaming is interesing too.

The problem is that every time Michael does a BFQ benchmark he measures throughput rather than latency when transferring files and opening programs, so the results of the BFQ benchmarks have always been meaningless.
These ones though show a real improvement http://algo.ing.unimo.it/people/paol...ed/results.php

[mk2soldier]

I'm also very interested in a direct comparison..

[speculatrix]

I would find it quite interesting

[coder111]

It would be nice if we actually have some benchmarks that measure LATENCY.

Bandwidth/Throughput measurements are OK as a regression test, but they miss most of the strengths of Liquorix kernel.

[halo9en]

The problem is that every time Michael does a BFQ benchmark he measures throughput rather than latency when transferring files and opening programs, so the results of the BFQ benchmarks have always been meaningless.
These ones though show a real improvement http://algo.ing.unimo.it/people/paol...ed/results.php

Exactly. Also see here for BFQ benchmarks on HDD: https://www.youtube.com/watch?v=ZeNbS0rzpoY and here for SSD: https://www.youtube.com/watch?v=1cjZeaCXIyM

As for zen-kernel, I have used it years ago but don't trust it anymore. BTW I'm curious about "Vegas TCP congestion control" and "smaller TX net queues" (tx_queue_len set to 50 against bufferbloat): are these actually useful on a laptop or just intended for network devices?