[opensuse] Why tumbleweed not faster (or fastest?)
https://www.phoronix.com/scan.php?page=article&item=mac-win-linux2018&num=6 Phoronix did benchmarks w/windows10 macOS 10.13.4, Ubuntu, Win10 WSL Ubuntu18.04 Fedora WS 28, & Clear Linux 22780. Clear linux was leader almost 60% of the time followed by a 21% showing for MacOS and Fedora 10%... Win10 was in last place 38% of the time, followed by Ubuntu @ 30% of the time in last place... They don't rank the other OS's in the synopsis, but the best Tumbleweed did was middle of the pack on many tests, one it got 2nd place, while several it was dead last almost 6x slower on some tests against Clear linux. How's that possible? I would have thought the linux OS's would have been grouped closer together, but apparently not. Tumbleweed didn't show in any of the graphicsMagick packages. Article didn't say why. Anyway not often I see tumbleweed mentioned, so ... at least that was a postive. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, 10 Jun 2018 07:40:26 -0700 L A Walsh <suse@tlinx.org> wrote:
several it was dead last almost 6x slower on some tests against Clear linux
It's not sensible to compare against Clear Linux, I seem to remember. Intel's version is heavily optimised for speed at the expense of other factors. I don't remember the details but I was never tempted to look at Clear because of those other factors. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On June 10, 2018 8:03:10 AM PDT, Dave Howorth <dave@howorth.org.uk> wrote:
On Sun, 10 Jun 2018 07:40:26 -0700 L A Walsh <suse@tlinx.org> wrote:
several it was dead last almost 6x slower on some tests against Clear linux
It's not sensible to compare against Clear Linux, I seem to remember. Intel's version is heavily optimised for speed at the expense of other factors. I don't remember the details but I was never tempted to look at Clear because of those other factors.
Well, if you can't even hint about those other factors, you provide no useful information at all. For most people, sitting at the keyboard, speed is very very important. It decides if we have to shell out for a new machine, or hang onto existing hardware another year. As long as all the machines and Linux version were doing the same benchmarks, you don't get to handwave those results away as if they don't count. The decisions made by each distro as to what they are going to run in the background, what filesystems they are going to use do make a difference. The end user pays that price every day. -- Sent from my Android phone with K-9 Mail. Please excuse my brevity. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, 10 Jun 2018 09:09:58 -0700 John Andersen <jsamyth@gmail.com> wrote:
On June 10, 2018 8:03:10 AM PDT, Dave Howorth <dave@howorth.org.uk> wrote:
On Sun, 10 Jun 2018 07:40:26 -0700 L A Walsh <suse@tlinx.org> wrote:
several it was dead last almost 6x slower on some tests against Clear linux
It's not sensible to compare against Clear Linux, I seem to remember. Intel's version is heavily optimised for speed at the expense of other factors. I don't remember the details but I was never tempted to look at Clear because of those other factors.
Well, if you can't even hint about those other factors, you provide no useful information at all.
Well, thanks for that useless comment, especially since Gertjan was able to remember some of the reasons just ten minutes later and well before you posted your response. I apologize for not having a perfect memory and for not being willing to do your searches for you after you were given my hint. Oh, and by the way, I didn't request a personal copy of your insult, so please don't send me any more. Get a mailer that does reply-to-list properly. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Op zondag 10 juni 2018 16:40:26 CEST schreef L A Walsh:
https://www.phoronix.com/scan.php?page=article&item=mac-win-linux2018&num=6
Phoronix did benchmarks w/windows10 macOS 10.13.4, Ubuntu, Win10 WSL Ubuntu18.04 Fedora WS 28, & Clear Linux 22780.
Clear linux was leader almost 60% of the time followed by a 21% showing for MacOS and Fedora 10%... Win10 was in last place 38% of the time, followed by Ubuntu @ 30% of the time in last place...
They don't rank the other OS's in the synopsis, but the best Tumbleweed did was middle of the pack on many tests, one it got 2nd place, while several it was dead last almost 6x slower on some tests against Clear linux.
How's that possible? I would have thought the linux OS's would have been grouped closer together, but apparently not.
Tumbleweed didn't show in any of the graphicsMagick packages. Article didn't say why.
Anyway not often I see tumbleweed mentioned, so ... at least that was a postive. These benchmarks IMO are useless without exact knowledge of what is installed by default. Like always on phoronix. The OS's al running the same linux standard base, same kernel/desktop etc. would come a bit closer. F.e. try to actually run Clear Linux in a VM and you'll see that it's only optimized for max. performance in benchmarks. Eh, some VW stuff comes to mind.
-- Gertjan Lettink a.k.a. Knurpht openSUSE Board Member openSUSE Forums Team -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Knurpht @ openSUSE wrote:
These benchmarks IMO are useless without exact knowledge of what is installed by default.
Wouldn't it be what was running and what the kernel factors were (optimization, scheduling, fs choices, background processes running by default, etc).
Like always on phoronix. The OS's al running the same linux standard base, same kernel/desktop etc.
But to choose exactly same kernel wipes out the differences that were chosen for each distro. Do they include only x686 optimizations, or do they have different libs for different processors?
would come a bit closer. F.e. try to actually run Clear Linux in a VM and you'll see that it's only optimized for max. performance in benchmarks. Eh, some VW stuff comes to mind.
VW? as in Volkswagon? From what I read in comments, Clear Linux was heavily worked on by Intel engineers who try to maximally tune it for new hardware. I don't know how true that is or if it complete doodoo, but if certainly would seem to explain slower speeds from distros that focus on new features over performance, for example, or wide hardware support (including other architectures) over a narrower focus. I forwarded the info here to see if anyone had more concrete knowledge or get a reactions like good/bad/ignore it/not our focus...etc. FWIW, I always compile my own kernel and some tools and and use mtune/mcpu=native in my own kernel in hopes of it squeezing a bit more out of the HW, but as to whether or not that helps? I noted on Clear linux, that they end to try to evaluate your use case to tailor the distro or a build to the customer. I used to be more performance oriented -- working in the compiler group @ Intel as a first job tends to nurture that. I still have an interest in making my machines run well, so I pay attention to such articles. Probably one of the few who do....who knows. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (4)
-
Dave Howorth -
John Andersen -
Knurpht ＠ openSUSE -
L A Walsh