[Expired for linux (Ubuntu) because there has been no activity for 60
days.]
** Changed in: linux (Ubuntu)
Status: Incomplete => Expired
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
I don't see how you can conclude, that this is an AMD related problem
now. It's not impossible but is it likely ? I don't think so.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
No I don't have any AMD based system right now.
It should be rather quick since the it doesn't take a long time to
reproduce the issue.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
I have never done that before, how long would that take ? Were you able
to reproduce the issue BTW ?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel
Before biseciton, please also give latest mainline kernel a try:
https://kernel.ubuntu.com/~kernel-ppa/mainline/v6.2/amd64/
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Sounds like a kernel bisection is needed:
https://wiki.ubuntu.com/Kernel/KernelBisection
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel linux-image-5.19.0-32
To quote myself from the original bug report: "I ran the tests on ZFS
and ext4, it makes no difference."
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel
So is this ZFS specific? From the description it sounds like the slow
down happens to ext4 too?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel
10 days later, it would be interesting for me to learn, if anyone
attempted to reproduce this and what the results were ? You'd have to
find a reasonably large project to compile with cmake/ninja that takes
around a minute or more to build. And you'd time the build time on each
kernel version.
--
Then I did the same on 5.19. There is a huge difference.
I am fairly certaan that everything from "zio_execute" to "SHA256Tra..." is
time spent in ZFS. To the right right side of `SHA256Tra...` is the compiler, I
checked this when looking at it in more detail. The compiler graph is basically
I sampled 10s of the build in both kernels with `perf` and run hotspot
on it for a flamegraph.
** Attachment added: "Flamegraph of 10s action on 15"
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2007717/+attachment/5649865/+files/flamegraph-15.png
--
You received this bug
So I tried this again with 5.19.0-32 and "amd_pstate=passive". Checked
with `dmesg` that it was indeed the right kernel and the right option.
But it makes no difference. Still 45s build time on 5.15.0-60 vs 1m on
5.19.0-32.
--
You received this bug notification because you are a member of Kernel
Please try kernel parameter "amd_pstate=passive".
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel linux-image-5.19.0-32 regression, compile speed down by 33%
Maybe it shares the same root cause as LP: #2006942?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel linux-image-5.19.0-32 regression, compile speed down by 33%
Does mainline kernel have the issue too?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel linux-image-5.19.0-32 regression, compile speed down by 33%
Status in
bonnie++ runs apparently proportional with installed RAM size, which
means on my machine it takes forever and does considerable SSD wear (I
cancelled it after a few minutes). I could run something less punishing
like maybe fio, with some parameters of your choosing.
--
You received this bug
To further isolate the nature of the slowdown, we should figure out if
this is caused by disk I/O degradation, CPU frequency scaling, or
something else. Can you use the bonnie++ package to check the disk
throughput under the old and new kernels?
--
You received this bug notification because you
I should add that I am using cmake/ninja. Ninja is usually better at
maximizing the CPU load than make and my system has 32 true CPU cores
and 64 hyperthreading ones.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
apport information
** Tags added: apport-collected jammy wayland-session
** Description changed:
I just yesterday and the day before made some benchmark tests for a
comparison between ext4 and zfs. I ran repeated builds of a large
software package and it always came out to 45s. Now today
** Package changed: ubuntu => linux (Ubuntu)
** Tags added: regression-update
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/2007717
Title:
Kernel linux-image-5.19.0-32 regression,
20 matches
Mail list logo