Comments

255 Comments
karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:0m0.000s user:4m41.300s sys:5m6.656s [No idea why the zero real time!!! That was what was displayed on the terminal]

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:107m15.682s user:74m0.200s sys:34m28.435s [Again this laptop is more than 10 times slower than it has been in other performance tests! It is always either less than 10 seconds or more than a 100 -- for reasons unknown to myself!]

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon 5500 Graphics – rea1:14m27.737s user:6m46.867s sys:5m36.560s

What has this to do with the Linux kernel???

There seems to be a trend to put Python stuff on the same page as the Linux Kernels.

They should have their own pages

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:10m0.151s user:4m35.980s sys:4m50.094s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:106m53.822s user:73m40.371s sys:34m9.800s N.B. sedna is running more than 10 times slower than it has previously

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon 5500 Graphics – rea1:14m21.262s user:6m43.405s sys:5m31.973s

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:10m2.600s user:4m37.432s sys:4m48.368s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:107m15.550s user:73m40.593s sys:34m37.832s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon 5500 Graphics – rea1:14m21.545s user:6m45.166s sys:5m30.618s

Note that sedna is more than 10 times slower than it was was with kernel 5.15.4-200.fc35

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:9m59.356s user:4m37.163s sys:4m49.931s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:9m6.846s user:3m41.810s sys:4m30.957s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon 5500 Graphics – rea1:14m20.629s user:6m43.230s sys:5m32.528s

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:10m1.437s user:4m37.027s sys:4m54.102s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:9m46.494s user:4m4.803s sys:4m46.083s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon 5500 Graphics – rea1:14m26.302s user:6m45.430s sys:5m35.645s

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:10m4.244s user :4m33.639s sys:4m57.794s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:9m20.422s user:3m45.257s sys:4m34.040s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics – rea1:4m33.415s user:6m51.145s sys:5m39.879s

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU – real:10m1.554s user:4m35.989s sys:4m50.504s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 – real:0m11.607s user:0m0.688s sys:0m3.238s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics – real:14m20.253s user: 6m44.191s sys:5m31.945s

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU -- real:10m1.785s user:4m36.552s sys:4m48.732s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 -- real:9m32.130s user:3m48.535s sys:4m40.128s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics -- real:14m20.354s user:6m44.664s sys:5m32.164s

It is curious that sedna (an old slow CPU) runs the performances tests significantly faster than mercury (a modern faster CPU)! I have noticed that that mercury's fan seems to go into overdrive during the tests, so suspect that the CPU is over heating and throttling down -- not sure how verify this directly.

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU -- real:10m2.380s user:4m36.831s sys:4m48.276s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 -- real:9m4.268s user:3m45.730s sys:4m33.048s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics -- real:14m23.069s user:6m45.132s sys:5m34.944s

It is curious that sedna (an old slow CPU) runs the performances tests significantly faster than mercury (a modern faster CPU)!

karma

Works for me: 1 desktop & 2 laptops all using the Mate Desktop Environment.

sirius: desktop 16GB Intel i7-3770 CPU -- real:10m2.352s user:4m37.112s sys:4m52.832s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 -- real:9m36.819s user:4m0.837s sys:4m38.138s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics -- real:15m14.904s user:7m6.550s sys:5m53.871s

It is curious that sedna (an old slow CPU) runs the performances tests significantly faster than mercury (a modern faster CPU)!

karma

Works for me: 1 desktop & 2 laptop all using the Mate Desktop Environment. I find it curious that sedna (an old slow CPU) runs the performances tests significantly faster than mercury (a modern fastr CPU)!

sirius: desktop 16GB Intel i7-3770 CPU real:9m58.718s user:4m37.535s sys:4m48.128s

sedna: laptop 8GB Intel i5-2520M CPU Lenovo T420 real:9m8.155s user:3m47.186s sys:4m40.173s

mercury: laptop 32GB AMD Ryzen 7 5700U with Radeon Graphics real:14m40.592s user:6m56.018s sys:5m39.209s

@oprizal The laptop was extremely lightly loaded, It was only running the applications I mentioned above. My email client is on my main desktop. There are no applications that would have changed the elapsed time tenfold for some runs and not others. Also if you have a look at all the time i've recorded, they are all either very close to 10 minutes or 110 minutes. I never saw times between 15 and 105 minutes. So applications should not be blamed for the excessive elapsed time.

The differences in elapsed times are significant, greater than 5 sigma, to use the parlance of the LHC at CERN.

@jforbes For every kernel I have tested, on both the laptop and the desktop, I have run the tests in near identical conditions as far as is practicable. The machines have all the latest updates applied prior to testing. I have a newer more powerful AMD desktop that I do most of my work on.

The laptop and the desktop are rebooted after installing the new kernel. The System monitor is expanded. LibreOffice is started to display some notes. Palemoon is used to display the bodhi update details. The kernel-tests directory is installed afresh prior to the tests, and deleted afterwards. Both the regression tests are run as root in the order specified. The machines are never rebooted between the tests. No other work is done on the machines while the tests are running, nor between test runs.

Essentially the methodology is the same for all runs of the regression tests on both machines.

I reran just the performance test on the laptop and it ran for ms which is not significantly different to the previous run.

semanage boolean -m --on selinuxuser_execheap

time ./runtests.sh -t performance ; date

Test suite called with performance Using config in CONFIG.sedna Sat 16 Oct 08:52:46 NZDT 2021 Latency measurements Sat 16 Oct 09:02:21 NZDT 2021 Calculating file system latency Sat 16 Oct 09:06:15 NZDT 2021 Local networking Sat 16 Oct 09:10:58 NZDT 2021 Bandwidth measurements Sat 16 Oct 10:22:28 NZDT 2021 Calculating context switch overhead Sat 16 Oct 10:36:22 NZDT 2021 ./performance/lmbench3 PASS

Test suite complete PASS

Your log file is being submitted... Upload successful! The following information is not submitted with your log; it is for informational purposes only. Vulnerability status: /sys/devices/system/cpu/vulnerabilities/itlb_multihit:KVM: Mitigation: VMX unsupported /sys/devices/system/cpu/vulnerabilities/l1tf:Mitigation: PTE Inversion /sys/devices/system/cpu/vulnerabilities/mds:Mitigation: Clear CPU buffers; SMT vulnerable /sys/devices/system/cpu/vulnerabilities/meltdown:Mitigation: PTI /sys/devices/system/cpu/vulnerabilities/spec_store_bypass:Mitigation: Speculative Store Bypass disabled via prctl and seccomp /sys/devices/system/cpu/vulnerabilities/spectre_v1:Mitigation: usercopy/swapgs barriers and __user pointer sanitization /sys/devices/system/cpu/vulnerabilities/spectre_v2:Mitigation: Full generic retpoline, IBPB: conditional, IBRS_FW, STIBP: conditional, RSB filling /sys/devices/system/cpu/vulnerabilities/srbds:Not affected /sys/devices/system/cpu/vulnerabilities/tsx_async_abort:Not affected

real 103m42.556s user 68m42.265s sys 36m32.128s Sat 2021-10-16 10:36:27 NZDT

semanage boolean -m --off selinuxuser_execheap

karma

Desktop 16GB Intel i7-3770 CPU & laptop 8GB Intel i5-2520M CPU Lenovo T420 (the desktop ran the performance test in 10m03s -- but the laptop took 106m24s, about 10 times longer) - all using the Mate Desktop Environment. Other than the slowness, the regression tests both passed.

@jforbes The laptop should not be taking more than 10 times the elapsed time of the desktop to do the performance test, as on last kernel 5.14.11 version the laptop was actually slightly quicker by a few seconds!

karma

Works for me: desktop 16GB Intel i7-3770 CPU & laptop 8GB Intel i5-2520M CPU Lenovo T420 (the desktop ran the performance test in 10m03s -- but the laptop was slightly faster at 9m38s) - all using the Mate Desktop Environment.

@jforbes The laptop should not be taking more than 10 times the elapsed time of the desktop to do the performance test, as on last kernel version it is actually slightly quicker by a few seconds!

karma

Desktop 16GB Intel i7-3770 CPU & laptop 8GB Intel i5-2520M CPU Lenovo T420 (the desktop ran the performance test in 10m02s -- but the laptop was slightly faster at 9m40s) - all using the Mate Desktop Environment.

In the performance test, the laptop is more than 10 times slower than the desktop, whereas with 5.14.9 the laptop was slightly faster!

Works for me: desktop 16GB Intel i7-3770 CPU & laptop 8GB Intel i5-2520M CPU Lenovo T420 (the desktop ran the performance test in 10m02s -- but the laptop was slightly faster at 9m40s) - all using the Mate Desktop Environment.

??? Could anyone enlighten me as to why sometimes the laptop is slightly faster than the desktop, yet at other times it is more than 10 times slower? ???

I have been running this kernel on my Ryzen 5950X box for 5 days and 20 minutes according to top. I've had no hangups nor other such problems that I can attribute to the kernel. I use this box for over 10 hours most days for a wide variety of tasks from compiling to browsing the web -- Fedora 34.

And using Fedora 35, I have been running on an Intel i7-3770 CPU for 6 days and 20 hours non stop without problems, but the box doesn't get much use. My laptop with an Intel i5-2520M CPU (had to be rebooted for an unrelated reason) has been used on & off for about 7 days -- also Fedora 35.