Brian K. White wrote:

On 2/6/2012 8:53 AM, Joachim Schrod wrote:

...

Felix Miata wrote:

...

On 2012/02/06 06:01 (GMT+0100) Anders Johansson composed:

...

On Sunday 05 February 2012 18:57:57 Felix Miata wrote:

...

...

...proprietary NVidia already installed; accordingly, xorg.conf was created consistent with NVidia custom high DPI operation...

...

kernel-desktop is a good choice for a desktop machine.

It used to be it was clear that was the case. Or maybe it was recently changed so that is now the case. I can't remember which, but I do remember there was a change discussed either here and/or on the factory list.

Executive Summary for Bob: -- Keep kernel-desktop and delete kernel-default. -- If you don't do so already, use the NVidia kmp packages from the community repository.

For Felix: Maybe you mean this discussion:

A few months ago, it was mentioned here that kernel-desktop is now the "new default" and should be used in most circumstances. kernel-default is primarly supplied for old hardware that doesn't have PAE capabilities.

That the name "kernel-default" stuck for the non-default kernel and that the two names were not switched was then attested as unfortunate. It is something that the devs contemplated to change for enhanced clarity, but then decided that it was too much hassle for the gain. AFAIR, there were too many problems with dependencies or something like that.

HTH, Joachim

PS: Ignore BC, put him in a kill-file. It's better for your health.

I don't really think kernel-default is for old hardware. I run it exclusively*, on brand new hardware, because I run servers that I need optimized for throughput instead of for latency,

Brian, please, take notion of the context of this thread. Please, try to imagine what the OP wanted and how we can help to achieve that. Pay attention to the posting history of the involved persons, if possible. If you start to contemplate differentiating kernels according to their optimization for throughput aganist their latency, you're in the 0.05 promille percentile who knows how to choose their kernels in specific ways. The other 99.9995% (rounded, of course) wants to have less hassle and don't want to care about these fine differences. And that's especially valid for Bob who should be well known to you as a person with eye-sight disabilities. I know both Bob and you from postings to this list, so, you will have realized Bob's context by yourself, too, surely, don't you? Concerning your contribution to this thread, I doubt that the difference between stock kernel-desktop and stock kernel-default gains a lot. If you're really down to this fine-grained differences, you need to analyze your workload and configure the kernel's scheduler and IO subsystems accordingly; individually, as your measurements will point out. Just choosing a kernel is cargo-cult sysadmin work and not professional. It lacks data from profiling and past change/problem/incident management, which you, as a professional sysadmin, surely took before engaging in your kernel configuratiuon selection; like we do at our customer's sites. But this is way beyond the discussion of this thread here.

From your posting history I judge that you and I are in the same situation as having to plan data center rollouts for large amounts of systems where such individual performance analysis brings large gains. But we should never forget that usage of Linux on the desktop has not quite the same amount of need for optimization. It runs good there as it is. Let's stop to tell the problems of kernel selections when you have 100,000 systems to roll out; for almost all basic installations it just works.

So, I would risk the notion of telling you: In the context of Bob's situtiation and his original question, your telling of kernel-choosing according to selection of throughput vs. latency is not helpful, not at all. Sorry, but that's my NSHO. Best, Joachim (Sorry that I'm not able to express this better at 3am, but I'm not a native English speaker. Reading it through, it's not as differentiated as I'd have liked it to be. But having a mother-in-law with only 45% sight on one eye and 5% sight on the other, I can sympathize with Bob and therefore contribute here more privately than I'd usually do.) -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org