Skip to content

Instantly share code, notes, and snippets.

@mdcallag
Last active April 19, 2024 16:46
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save mdcallag/830c9c25c25b8e56572818472cd551b7 to your computer and use it in GitHub Desktop.
Save mdcallag/830c9c25c25b8e56572818472cd551b7 to your computer and use it in GitHub Desktop.
cmdline=""
old new1 new2
err: 0 1 764
wr-seq 4.9 4.9 4.9
- cpu 15% 15% 16% (us,sy): old(0.3,14.9) new1(1.4,13.7) new2(1.9,14.0)
wr-ran 108190 91090 93084
- cpu 41% 39% 37% (us,sy): old(6.9,34.1) new1(11.7,27.0) new2(13.5,24.0)
rd-seq 6.6 6.5 6.5
- cpu 31% 27% 27% (us,sy): old(0.5,30.7) new1(0.5,26.7) new2(0.5,26.5)
rd-ran 155498 117747 117726
- cpu 60% 49% 50% (us,sy): old(8.2,51.2) new1(22.8,26.5) new2(25.1,25.1)
new1:
[Thu Apr 18 20:20:48 2024] pcieport 0000:00:01.2: AER: Corrected error received: 0000:01:00.0
[Thu Apr 18 20:20:48 2024] nvme 0000:01:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[Thu Apr 18 20:20:48 2024] nvme 0000:01:00.0: device [144d:a80a] error status/mask=00000001/0000e000
[Thu Apr 18 20:20:48 2024] nvme 0000:01:00.0: [ 0] RxErr (First)
new2:
[Thu Apr 18 20:32:59 2024] pcieport 0000:00:01.2: AER: Corrected error received: 0000:01:00.0
[Thu Apr 18 20:32:59 2024] nvme 0000:01:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[Thu Apr 18 20:32:59 2024] nvme 0000:01:00.0: device [144d:a80a] error status/mask=00000001/0000e000
[Thu Apr 18 20:32:59 2024] nvme 0000:01:00.0: [ 0] RxErr (First)
cmdline="nosmt"
old new1 new2
err: 0 2 762
wr-seq 4.9 4.9 4.9
- cpu 15% 15% 22% (us,sy): old(1.9,13.3) new1(1.4,13.9) new2(1.8,20.2)
wr-ran 116275 98829 90440
- cpu 49% 51% 46% (us,sy): old(8.7,40.0) new1(13.4,37.7) new2(11.5,34.7)
rd-seq 6.9 6.6 6.5
- cpu 30% 32% 30% (us,sy): old(0.4,29.0) new1(0.3,31.3) new2(0.4,29.6)
rd-ran 155072 121896 124757
- cpu 60% 49% 49% (us,sy): old(8.2,51.2) new1(20.0,28.6) new2(22.0,27.2)
[Thu Apr 18 20:57:29 2024] nvme 0000:01:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[Thu Apr 18 20:57:29 2024] nvme 0000:01:00.0: device [144d:a80a] error status/mask=00000041/0000e000
[Thu Apr 18 20:57:29 2024] nvme 0000:01:00.0: [ 0] RxErr (First)
[Thu Apr 18 20:57:29 2024] nvme 0000:01:00.0: [ 6] BadTLP
[Thu Apr 18 21:01:10 2024] nvme 0000:01:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[Thu Apr 18 21:01:10 2024] nvme 0000:01:00.0: device [144d:a80a] error status/mask=00000001/0000e000
[Thu Apr 18 21:01:10 2024] nvme 0000:01:00.0: [ 0] RxErr (First)
cmdline="amd_pstate=disable"
old new1 new2
err: 0 1 761
wr-seq 4.9 5.0 5.0
- cpu 39% 36% 40% (us,sy): old(4.8,34.4) new1(5.4,30.4) new2(5.1,34.7)
wr-ran 62676 61764 62813
- cpu 41% 44% 38% (us,sy): old(6.3,35.0) new1(9.5,34.9) new2(9.0,28.7)
rd-seq 6.7 6.7 6.7
- cpu 42% 38% 41% (us,sy): old(0.3,41.6) new1(0.2,37.4) new2(0.8,39.8)
rd-ran 132860 107540 105937
- cpu 54% 54% 55% (us,sy): old(8.0,53.8) new1(22.6,31.2) new2(19.8,34.8)
cmdline="pcie_aspm=off"
In the results below, new1 mounts the 2nd m.2 at nvme1n1 while new2 mounts it at nvme0n1, not sure that matters
old new1 new2
err: 0 0 0
wr-seq 4.9 4.9 4.9
- cpu 15% 16% 16% (us,sy): old(1.9,12.8) new1(1.9,13.8) new2(1.0,15.2)
wr-ran 119158 103278 124441
- cpu 47% 43% 46% (us,sy): old(7.5,38.9) new1(14.4,28.4) new2(7.2,38.6)
rd-seq 6.5 6.6 6.5
- cpu 30% 33% 28% (us,sy): old(0.6,29.8) new1(0.4,32.3) new2(0.3,28.5)
rd-ran 155604 118090 157439
- cpu 60% 49% 60% (us,sy): old(8.7,51.5) new1(26.6,22.5) new2(7.9,52.2)
ser7-3 with non-HW, 5.15.0-102-generic, pcie_aspm=off
$ sudo dmesg -T | grep -i bios
[Thu Apr 18 23:31:27 2024] DMI: AZW SER/SER, BIOS SER7PRO_P5C8V38 01/10/2024
[Thu Apr 18 23:31:27 2024] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP5.RTL8], AE_NOT_FOUND (20210730/dswload2-162)
[Thu Apr 18 23:31:27 2024] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
[Thu Apr 18 23:31:27 2024] PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug
same on ser7-2
[Thu Apr 18 23:32:36 2024] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP5.RTL8], AE_NOT_FOUND (20210730/dswload2-162)
[Thu Apr 18 23:32:36 2024] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
[Thu Apr 18 23:32:36 2024] PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug
ugh, but same on ser7-1
cmdline="pcie_aspm=off", non-HWE kernl
old new1 new2
err: 0 0 0
wr-seq 4.8 4.7 4.8
- cpu 21% 18% 23% (us,sy): old(3.0,18.4) new1(3.6,14.0) new2(5.5,17.8)
wr-ran 201289 196998 218018
- cpu 43% 46% 46% (us,sy): old(8.5,34.8) new1(21.3,24.6) new2(9.3,36.7)
rd-seq 6.6 6.5 6.5
- cpu 34% 34% 19% (us,sy): old(1.1,33.3) new1(0.9,33.0) new2(0.6,18.2)
rd-ran 178665 159542 181128
- cpu 54% 46% 55% (us,sy): old(11.7,42.4) new1(11.7,34.0) new2(12.3,42.5)
ser7-3 with non-HW, 5.15.0-102-generic, pcie_aspm=off, read-only, 10 minutes
old new1 new2
rd-seq 6.4 6.5 6.4
- cpu 33% 29% 30% (us,sy): old(0.9,32.4) new1(0.8,27.5) new2(0.9,29.0)
rd-ran 80534 65163 82224
- cpu 45% 39% 47% (us:sy): old(11.7,32.9) new1(25.1,14.3) new2(11.7,33.6)
https://mega.nz/file/ejBlAQ5S#SjtEDdFnqgC1xKK7fac0NH8eRMUNe2-zBnWF1jViFOs
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment