Re: What kernel do you use
Posted: 2017-12-17 02:45
I've been playing arround with Liquorix Kernel, and you guys, what have been using now days? Stock? Own Kernel?
https://liquorix.net/
https://liquorix.net/
Do you play games or make multimedia edition at this machine? For me it made difference on that.steve_v wrote:The one that comes from the Debian repos. Whatever version that is at the moment. Works for me.
I used to compile my own kernels, back in 1999 or so. Now that you're unlikely to find a machine with so little RAM that a custom kernel is needed, why bother with the added effort?
I haven't had any problems whatsoever with the stock kernel, and the supposed "optimisations" in e.g. the Liquorix build don't make any noticeable difference, at least as far as I can tell.
If by "make multimedia edition" you mean video editing, then no. But I play plenty of games.Capitain_Jack wrote: Do you play games or make multimedia edition at this machine? For me it made difference on that.
Multimedia means all media types, video, image and audio.steve_v wrote: If by "make multimedia edition" you mean video editing, then no. But I play plenty of games.
Okay then, I play videos, occasionally edit audio (less of the playing, as I have an MPD server elsewhere), and edit images every now and then.Capitain_Jack wrote:Multimedia means all media types, video, image and audio.
Code: Select all
empty@Puffy:~ $ uname -a
OpenBSD Puffy.lan 6.2 GENERIC.MP#298 amd64
empty@Puffy:~ $
Do a video render test, it is where I see most difference. Also when creating music with Reason 5 (trough wine) or any other hard professional audio recording/editing software.steve_v wrote:Okay then, I play videos, occasionally edit audio (less of the playing, as I have an MPD server elsewhere), and edit images every now and then.Capitain_Jack wrote:Multimedia means all media types, video, image and audio.
I even installed the Liquorix kernel yesterday to see if anything had changed... And I still see no difference.
No subjective "feels" that seem to be all the rage here, and no measurable performance difference in 3 games + Blender (Blenchmark). Identical framerates.
Dunno man, The Debian kernel seems just fine to me.
Liquorix is not a real time kernel, so what does this have to do with anything?Capitain_Jack wrote:This is a more comprehensive text talking about the differences:
https://help.ubuntu.com/community/Ubunt ... TimeKernel
Also here for detail into real time computing:
"...Real-time computing is sometimes misunderstood to be high-performance computing, but this is not an accurate classification....Therefore, the most important requirement of a real-time system is predictability and not performance..."
https://en.wikipedia.org/wiki/Real-time_computing
Ok, I see you need to see "the insides" to understand fully, take a read here:steve_v wrote: Liquorix is not a real time kernel, so what does this have to do with anything?
Nope, still no mention of realtime patches there. Yes, Liquorix tweaks the scheduler, but it's not a realtime kernel.Capitain_Jack wrote:Ok, I see you need to see "the insides" to understand fully, take a read here
I never said it is a real time kernel, but here the part of the text where he mention it's aiming a real time sytem:steve_v wrote:Nope, still no mention of realtime patches there.
https://techpatterns.com/forums/about2499.htmlEnables NOCB for all CPUs, increasing RCU overhead
Per the options in the kernel configuration, this increases overhead but reduces OS jitter and is more appropriate for a real-time system (what we're aiming for here, a better experience).
That's why I use it. This is a good discussion, we should and could open a new post, what do you say? have more or less the same conversation? Would you start this one?steve_v wrote: If I was doing pro-audio work where predictable latency is critical, I might use the realtime patches.
If I could detect a performance gain from the Liquorix patches with any workload I care about, I'd use that too.
Code: Select all
uname -a
Linux antix1 4.14.8-antix.1-amd64-smp #1 SMP PREEMPT Wed Dec 20 15:19:11 EET 2017 x86_64 GNU/Linux
That's a good one for old hardware, keeps it updated and very very light one.anticapitalista wrote:Code: Select all
uname -a Linux antix1 4.14.8-antix.1-amd64-smp #1 SMP PREEMPT Wed Dec 20 15:19:11 EET 2017 x86_64 GNU/Linux
Good adviceBulkley wrote:I always have at least two kernels installed and that they have proven to work
That's for sure, no matter witch kernel you use, it should be a rule. I think more than two of the same is exaggerated, to clarify that you can have more than one kernel installed with no major problem other than more update time.Bulkley wrote:always have at least two kernels installed and that they have proven to work... With two kernels in situ it makes for an easy test when troubleshooting.
Code: Select all
uname -a
Linux antix1 4.14.9-antix.1-amd64-smp #1 SMP PREEMPT Mon Dec 25 20:53:07 EET 2017 x86_64 GNU/Linux
Code: Select all
~ $ uname -a
Linux Xanadu 4.14.9-1-zen #1 ZEN SMP PREEMPT Tue Dec 26 00:19:11 UTC 2017 x86_64 GNU/Linux
~ $