XPost: alt.computer.workshop

$ uname -a
Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025 x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out
real 404.33
user 19235.92
sys 3833.13

$ grep NATIVE linux-6.16/.config
CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
Keep me informed on the behaviour of this kernel.. As the "BugFree(tm)"
series didn't turn out too well, I'm starting a new series called the "ItWorksForMe(tm)" series, of which this new kernel is yet another
shining example.
-- Linus, in the announcement for 1.3.29

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Wed, 30 Jul 2025 18:45:43 -0700, % wrote:

vallor wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025
x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

stick it up your kernel

Close.

I've already upped my kernel. So up yours!

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"Never eat anything bigger than your head."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

vallor <[email protected]> news:[email protected] Thu, 31
Jul 2025 01:38:44 GMT in alt.computer.workshop, wrote:

$ uname -a
Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025 x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out
real 404.33
user 19235.92
sys 3833.13

Nice! Very nice.

$ grep NATIVE linux-6.16/.config
CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

Ahh. You've taken advantage of the hardware specific tweaks. Let me know
if it makes a noticeable difference in the way the system operates for you. With your permission and assistance, I will know of five of you (a very
small sample pool I admit) who are testing that feature with that kernel;
but also testing the same kernel without that feature enabled; looking for noticeable differences in performance or behaviour of the machine.

(Request: Let's keep the topic on Linux, please? :) )

I'll do my part, but, unfortunately we are still in the fumigation process
here on alt.computer.workshop. :)





--
I don't need no Dr. All I need...is my lawyer.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop, alt.atheism

% wrote:

Creon wrote:

On Wed, 30 Jul 2025 18:45:43 -0700, % wrote:

vallor wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025 >>>> x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :)  )

stick it up your kernel

Close.

I've already upped my kernel.  So up yours!

i know but you didn't have to go tongue first

Says squiggles the assworm aka blumpkin boy.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Wed, 30 Jul 2025 21:29:18 -0700, % wrote:

Creon wrote:

On Wed, 30 Jul 2025 18:45:43 -0700, % wrote:

vallor wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT
2025 x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

stick it up your kernel

Close.

I've already upped my kernel. So up yours!

i know but you didn't have to go tongue first

I didn't -- your head was in the way.

ObLinux:

$ HEAD https://cultnix.org
200 OK
[...]

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"Personally, I like my flying brains dark and evil."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On 2025-07-31, vallor <[email protected]> wrote:

$ uname -a
Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025 x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out
real 404.33
user 19235.92
sys 3833.13

$ grep NATIVE linux-6.16/.config
CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

What distribution are you running?
Also, are you obtaining the source from the official distribution
repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the
versions you are upgrading to.

pothead@LinuxMint:~$ uname -a
Linux LinuxMint 6.8.0-64-generic #67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux



--
pothead

"Our lives are fashioned by our choices. First we make our choices.
Then our choices make us."
-- Anne Frank

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On 31/07/2025 15:28, pothead wrote:

On 2025-07-31, vallor <[email protected]> wrote:

$ uname -a
Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025 x86_64 x86_64
x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out
real 404.33
user 19235.92
sys 3833.13

$ grep NATIVE linux-6.16/.config
CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

What distribution are you running?

I'm using Linux Mint 22.1 on my old 24 inch iMac

Also, are you obtaining the source from the official distribution repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the
versions you are upgrading to.

pothead@LinuxMint:~$ uname -a
Linux LinuxMint 6.8.0-64-generic #67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Thu, 31 Jul 2025 14:28:04 -0000 (UTC), pothead wrote:

On 2025-07-31, vallor <[email protected]> wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025
x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

What distribution are you running?

Linux Mint -- says so in the .sig. :)

Also, are you obtaining the source from the official distribution repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the versions
you are upgrading to.

I get the latest vanilla kernel from https://kernel.org and NVIDIA drivers
from https://nvidia.org -- will have to post instructions when I get a
chance. It's not difficult to do, but there is a "gotcha" with Mint:
don't try to use DKMS with NVIDIA drivers, doesn't seem to work right.

pothead@LinuxMint:~$ uname -a Linux LinuxMint 6.8.0-64-generic
#67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux

I've been criticized for taking the time to build and boot the latest
kernel, since the kernel that comes with Linux Mint works just fine.
Eh -- we all have our hobbies.

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"Give instruction to a wise man and he will be yet wiser."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Thu, 31 Jul 2025 14:28:04 -0000 (UTC), pothead wrote:

What distribution are you running?
Also, are you obtaining the source from the official distribution repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the versions
you are upgrading to.

pothead@LinuxMint:~$ uname -a Linux LinuxMint 6.8.0-64-generic
#67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux

Mint is based on Ubuntu and Ubuntu never had the latest, greatest kernels.
iirc 24.04 was 6.8. I'm running 25.04 and it has the 6.14.0-24-generic
kernel. Ubuntu plans to stay more current.

my Fedora box is 6.15.8. It gets new kernels frequently and is usually
only a step or two behind the latest release. I don't know how much
tweaking Mint does but if you like Cinnamon there is a Fedora spin. The packages will be updated more frequently but that may or may not make and difference to you, particularly kernels.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Thu, 31 Jul 2025 04:46:50 -0000 (UTC), Gremlin wrote:

looking for
noticeable differences in performance or behaviour of the machine.

What do you mean by "noticeable?"

Human sensory perception cannot detect differences in performance.
Such things must be rigorously measured.

The kernel is only an interface to hardware. The system calls provided
by the kernel would not seem to benefit much from a CONFIG_X86_NATIVE_CPU option:

https://www.chromium.org/chromium-os/developer-library/reference/linux-constants/syscalls/

But other things like context switching and scheduling could
benefit.

The main thing that CONFIG_X86_NATIVE_CPU would offer is accurate cash
sizes but I am not sure how the kernel utilizes caches in its operation.

What must be understood is that kernel optimization is only a
first step. All other software must be equally optimized if
a performance benefit is expected.

Moreover, there is much more to kernel optimization than
CONFIG_X86_NATIVE_CPU. Many other parameters can be tuned
either during build or during execution.

However, I have no doubt that optimization plays a decisive
role in performance, especially for High Performance Computing,
i.e. math/physics, image/video operations, file compression,
etc., etc.

I have made my own benchmarks demonstrating a 37% performance
improvement when using optimized code versus generic code --
and this was only for a single program. If all the dependencies
of this program were generic (as they are with distros) then
the results would have been even greater.

My advice:

Stop arguing. Get Gentoo. Optimize. Shut the fuck up.



--
Systemd: solving all the problems that you never knew you had.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On 2025-07-31, Creon <[email protected]> wrote:

On Thu, 31 Jul 2025 14:28:04 -0000 (UTC), pothead wrote:

On 2025-07-31, vallor <[email protected]> wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT 2025
x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

What distribution are you running?

Linux Mint -- says so in the .sig. :)

lol.
I have sigs turned off in slrn.
I guess I should turn them back on :)

Also, are you obtaining the source from the official distribution
repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the versions
you are upgrading to.

I get the latest vanilla kernel from https://kernel.org and NVIDIA drivers from https://nvidia.org -- will have to post instructions when I get a chance. It's not difficult to do, but there is a "gotcha" with Mint:
don't try to use DKMS with NVIDIA drivers, doesn't seem to work right.

pothead@LinuxMint:~$ uname -a Linux LinuxMint 6.8.0-64-generic
#67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux

I've been criticized for taking the time to build and boot the latest
kernel, since the kernel that comes with Linux Mint works just fine.
Eh -- we all have our hobbies.

As fate would have it I updated LM right after I posted the question and low and behold a new kernel was just released.

So now I am at:

pothead@LinuxMint:~$ uname -a
Linux LinuxMint 6.8.0-71-generic #71-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 22 16:52:38 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
pothead@LinuxMint:~$

Maybe someday I'll play around with compiling a custom kernel.



--
pothead

"Our lives are fashioned by our choices. First we make our choices.
Then our choices make us."
-- Anne Frank

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Thu, 31 Jul 2025 13:11:04 -0700, % wrote:

Creon wrote:

On Thu, 31 Jul 2025 14:28:04 -0000 (UTC), pothead wrote:

On 2025-07-31, vallor <[email protected]> wrote:

$ uname -a Linux lm 6.16.0 #1 SMP PREEMPT Wed Jul 30 18:03:42 PDT
2025 x86_64 x86_64 x86_64 GNU/Linux

$ tail -3 linux-6.16/nohup.out real 404.33 user 19235.92 sys 3833.13

$ grep NATIVE linux-6.16/.config CONFIG_CC_HAS_MARCH_NATIVE=y
CONFIG_X86_NATIVE_CPU=y

(Request: Let's keep the topic on Linux, please? :) )

What distribution are you running?

Linux Mint -- says so in the .sig. :)

Also, are you obtaining the source from the official distribution
repositories or do you roll your own?
I'm curious because LM which I run, is always way behind the versions
you are upgrading to.

I get the latest vanilla kernel from https://kernel.org and NVIDIA
drivers from https://nvidia.org -- will have to post instructions when
I get a chance. It's not difficult to do, but there is a "gotcha" with
Mint: don't try to use DKMS with NVIDIA drivers, doesn't seem to work
right.

pothead@LinuxMint:~$ uname -a Linux LinuxMint 6.8.0-64-generic
#67-Ubuntu SMP PREEMPT_DYNAMIC Sun Jun 15 20:23:31 UTC 2025 x86_64
x86_64 x86_64 GNU/Linux

I've been criticized

and rightly so

Are you done, or are you going to keep replying to the same
message?

Hey, if you don't get it -- there's always ACW in which
to ask about it!

tcpdrop.bt(8) System Manager's Manual tcpdrop.bt(8)

NAME
tcpdrop.bt - Trace kernel-based TCP packet drops with
details. Uses Linux bpftrace/eBPF

SYNOPSIS
tcpdrop.bt

DESCRIPTION
This tool traces TCP packets or segments that were
dropped by the kernel, and shows details from the IP and
TCP headers, the socket state, and the kernel stack
trace. This is useful for debugging cases of high kernel
drops, which can cause timer-based retransmits and per‐
formance issues.

This tool works using dynamic tracing of the tcp_drop()
kernel function, which requires a recent kernel version.

This tool is limited to ipv4, and cannot parse tcpflags
as bpftrace currently cannot parse socket buffers in the
way that bcc can.
_ _ _ _ _ _ _

Actually, I saw this catch a drop of an ipv6 TCP packet, so maybe
the man page is out of date...

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"How do I set my laser printer on stun?"

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

Joel <[email protected]> news:[email protected]
Thu, 31 Jul 2025 20:12:11 GMT in alt.computer.workshop, wrote:

Farley Flud <[email protected]> wrote:

My advice:

Stop arguing. Get Gentoo. Optimize. Shut the fuck up.

Totally crazy, mainstream distros are the norm.

The individual has a programming/coding background so their specific distro flavor may have something to do with their programming interest? I have no issues with it, personally. I'm not entirely sure what you mean by
mainstream distros, either.

Which ones would be considered mainstream to you? And why would they be such
vs the others? Is it just the reported userbase size that determines this
for you, or do you take other factors into consideration?





--
I don't need no Dr. All I need...is my lawyer.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 2 Aug 2025 05:30:49 -0000 (UTC), Gremlin wrote:

Farley Flud <[email protected]> news:pan$ab58a$bf644371$fa8ab850$[email protected] Thu, 31 Jul 2025 18:58:19 GMT in alt.computer.workshop, wrote:

On Thu, 31 Jul 2025 04:46:50 -0000 (UTC), Gremlin wrote:

looking for noticeable differences in performance or behaviour of the
machine.

What do you mean by "noticeable?"

Human sensory perception cannot detect differences in performance.
Such things must be rigorously measured.

Actually, depending on the hardware; human sensory is fast enough to
catch a performance difference.
In this specific request though, I was asking if they happened to notice
any difference in the systems performance; I assume they also understood
my question was asking how the system 'felt' between the differences.
Does that make better sense?

The rule of thumb is 500 milliseconds. For example if you click on a
button and it takes 500 msec for something to happen most users will feel
the system is sluggish.

For many things that might be considered more than sluggish. For context: (trigger warning: triggers get pulled)

https://www.youtube.com/watch?v=WzHG-ibZaKM

The bangs are about 125 msecs apart and there is no problem discerning the individual events. I don't know if it has been applied to computer systems
but the concept of JNDs has been around for a long time.

https://en.wikipedia.org/wiki/Just-noticeable_difference

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 2 Aug 2025 05:30:49 -0000 (UTC), Gremlin <[email protected]> wrote
in <[email protected]>:

Farley Flud <[email protected]> news:pan$ab58a$bf644371$fa8ab850$[email protected] Thu, 31 Jul 2025 18:58:19 GMT in alt.computer.workshop, wrote:

On Thu, 31 Jul 2025 04:46:50 -0000 (UTC), Gremlin wrote:

looking for noticeable differences in performance or behaviour of the
machine.

What do you mean by "noticeable?"

Human sensory perception cannot detect differences in performance. Such
things must be rigorously measured.

Actually, depending on the hardware; human sensory is fast enough to
catch a performance difference.
In this specific request though, I was asking if they happened to notice
any difference in the systems performance; I assume they also understood
my question was asking how the system 'felt' between the differences.
Does that make better sense?

I haven't noticed any difference.

Thought I'd try building Linux with it, and it took about the same amount
of time. Might try it with a "native" build of gcc, but the last
time I went through the gcc build dance was on HP-UX 10 back in
the 90's. In that case, I had to build gcc with HP's native
compiler, then use _that_ gcc to build gcc all over again.

"This is the way." But I probably wouldn't need to build the first-stage compiler, since I'd already be using gcc. Haven't looked into it yet.

Meanwhile, I haven't really been in my game for a couple of weeks. So
that might take priority. ;) Will be interesting to see if there's any noticeable difference in frame rates.

--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"Cats aren't clean, they're just covered with cat spit!"

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Sat, 02 Aug 2025 10:20:04 +0000, Farley Flud wrote:

On Sat, 2 Aug 2025 05:30:49 -0000 (UTC), Gremlin wrote:

The kernel is only an interface to hardware. The system calls
provided by the kernel would not seem to benefit much from a
CONFIG_X86_NATIVE_CPU option:

The intent behind it, (afaik), is to optimize specific routines from
the compilers perspective to better match the hardware it's being ran
on.

Again, the CONFIG_X86_NATIVE_CPU option is just a single option that in
and of itself is not likely to lead to a palpable performance boost.

(Example: CONFIG_X86_NATIVE_CPU does not enable Flexible Return and
Event Delivery (FRED) which is a new Intel CPU feature that greatly
improves kernel ring transitions.)

Regrettably, the default kernel happens to be riddled with "security" features that probably degrade performance much more than CONFIG_X86_NATIVE_CPU could override.

For example, the default kernel is built with strong stack protection
which means that a random "canary" is added to subroutines (functions). Calling a function in C is expensive enough but a "canary" only adds
more expense.

Optimizing the kernel to eliminate all that security crud and to select
more performance enhancing features, of which there are also many, is a
BIG job and requires a lot of fundamental understanding of both hardware
and OS design. Fortunately, however, because the kernel is so stable in
its design, if the job is done once it is done practically forever. The
same options can be used over over and over on future kernels.

But there is yet more. The kernel can also be optimized at run time, as opposed to build time. There are many parameters that deal with virtual memory and network throughput that can be tuned by the user. That is
the purpose of the /proc hierarchy as well as the "sysctl" utility.

My advice: Spend some time and effort now to optimize everything and
then you will have a reliable foundation for the future.

Yes, you read that right: "Farley Flud" wants you to turn off
security mitigations in your kernel. I do not recommend following his
advice.

He doesn't run a MAC subsystem either. I strongly recommend
running one to prevent security issues, which Linux Mint does by
default, using AppArmor.

(The alternative in Red Hat / Fedora land is SELinux, which was written
by the NSA. I've found AppArmor much easier to use -- YMMV.)

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"My cat likes to play Hide-And-Go-Puke."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Sat, 2 Aug 2025 05:30:49 -0000 (UTC), Gremlin wrote:

The kernel is only an interface to hardware. The system calls provided
by the kernel would not seem to benefit much from a
CONFIG_X86_NATIVE_CPU option:

The intent behind it, (afaik), is to optimize specific routines from the compilers perspective to better match the hardware it's being ran on.

Again, the CONFIG_X86_NATIVE_CPU option is just a single option that in
and of itself is not likely to lead to a palpable performance boost.

(Example: CONFIG_X86_NATIVE_CPU does not enable Flexible Return and Event Delivery (FRED) which is a new Intel CPU feature that greatly improves
kernel ring transitions.)

Regrettably, the default kernel happens to be riddled with "security"
features that probably degrade performance much more than CONFIG_X86_NATIVE_CPU could override.

For example, the default kernel is built with strong stack protection
which means that a random "canary" is added to subroutines (functions).
Calling a function in C is expensive enough but a "canary" only adds
more expense.

Optimizing the kernel to eliminate all that security crud and to
select more performance enhancing features, of which there are also
many, is a BIG job and requires a lot of fundamental understanding
of both hardware and OS design. Fortunately, however, because
the kernel is so stable in its design, if the job is done once it
is done practically forever. The same options can be used over
over and over on future kernels.

But there is yet more. The kernel can also be optimized at
run time, as opposed to build time. There are many parameters
that deal with virtual memory and network throughput that can
be tuned by the user. That is the purpose of the /proc hierarchy
as well as the "sysctl" utility.

My advice: Spend some time and effort now to optimize
everything and then you will have a reliable foundation for the
future.





--
Systemd: solving all the problems that you never knew you had.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On 2 Aug 2025 11:14:50 GMT, Creon wrote:

Yes, you read that right: "Farley Flud" wants you to turn off
security mitigations in your kernel. I do not recommend following his advice.

He doesn't run a MAC subsystem either. I strongly recommend
running one to prevent security issues, which Linux Mint does by
default, using AppArmor.

Yet another human parrot demonstrates his ability to recite
the standard dogma without a shred of understanding.

(The alternative in Red Hat / Fedora land is SELinux, which was written
by the NSA.

That must explain all those Russian spies that lurk outside
my window longing to discover my browsing history with a RowHammer
attack.


--
Systemd: solving all the problems that you never knew you had.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2 Aug 2025 08:15:32 GMT, vallor wrote:

Might try it with a "native" build of gcc,

That wouldn't make any difference.

Building GCC with "-march-native" would only affect GCC itself
and not any software compiled by it.

However, building a custom GCC is always advisable because the
default GCC supplied by the distros contains options that impart
hardening features to software such as "default-stack-clash-protection," "default-znow," and "sanitize." This security crud will degrade
the performance of all software built by GCC.

On my Gentoo system, I have all that useless garbage disabled.

But distro-supplied software is necessarily riddled with crippling
security junk.

ALL distros are seriously sub-optimal in terms of performance.





--
Gentoo: God's great gift to FOSS.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 02 Aug 2025 18:36:42 +0000, Farley Flud <[email protected]> wrote in <1858070be858ae99$39721$1044878$[email protected]>:

On 2 Aug 2025 08:15:32 GMT, vallor wrote:

Might try it with a "native" build of gcc,

That wouldn't make any difference.

You didn't read what I wrote. I was talking about the time to build
Linux, which probably _would_ depend on a native build of gcc.

--
-v System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"MIPS: Meaningless Indicator of Processor Speed."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On Sat, 02 Aug 2025 11:59:46 +0000, Farley Flud wrote:

On 2 Aug 2025 11:14:50 GMT, Creon wrote:

Yes, you read that right: "Farley Flud" wants you to turn off security
mitigations in your kernel. I do not recommend following his advice.

He doesn't run a MAC subsystem either. I strongly recommend running
one to prevent security issues, which Linux Mint does by default, using
AppArmor.

Yet another human parrot demonstrates his ability to recite the standard dogma without a shred of understanding.

(The alternative in Red Hat / Fedora land is SELinux, which was written
by the NSA.

That must explain all those Russian spies that lurk outside my window
longing to discover my browsing history with a RowHammer attack.

Knowing you, you probably run everything as "root" anyway -- leaving you
one browser exploit away from having your goose thoroughly cooked.

--
-c System76 Thelio Mega v1.1 x86_64 NVIDIA RTX 3090Ti 24G
OS: Linux 6.16.0 D: Mint 22.1 DE: Xfce 4.18
NVIDIA: 575.64.05 Mem: 258G
"George Orwell was an optimist."

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

XPost: alt.computer.workshop

On 2025-08-03, Creon <[email protected]> wrote:

On Sat, 02 Aug 2025 11:59:46 +0000, Farley Flud wrote:

On 2 Aug 2025 11:14:50 GMT, Creon wrote:

Yes, you read that right: "Farley Flud" wants you to turn off security
mitigations in your kernel. I do not recommend following his advice.

He doesn't run a MAC subsystem either. I strongly recommend running
one to prevent security issues, which Linux Mint does by default, using
AppArmor.

Yet another human parrot demonstrates his ability to recite the standard
dogma without a shred of understanding.

(The alternative in Red Hat / Fedora land is SELinux, which was written
by the NSA.

That must explain all those Russian spies that lurk outside my window
longing to discover my browsing history with a RowHammer attack.

Knowing you, you probably run everything as "root" anyway -- leaving you
one browser exploit away from having your goose thoroughly cooked.

ROTFLMAO!!!!

I saw that picture of some device he was building and just
the fact it was covered in hair, lint and dust was enough for me.

At least I think it was him?
The nyms keep changing.



--
pothead

"Our lives are fashioned by our choices. First we make our choices.
Then our choices make us."
-- Anne Frank

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)