[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #24 from rai...@ultra-secure.de --- I still have very slow IO in FreeBSD 13.0 amd64 on KVM. Even with this kvmclock driver (compile from ports as of today). Host: ii qemu-system-x86

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Dave Cottlehuber changed: What|Removed |Added Status|New |In Progress -- You are

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #23 from commit-h...@freebsd.org --- A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=563d5929d5f28631999a32d2aab4e72e1bf2c323 commit 563d5929d5f28631999a32d2aab4e72e1bf2c323 Author:

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Dimitry Andric changed: What|Removed |Added CC||d...@freebsd.org --- Comment #22

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Allan Jude changed: What|Removed |Added CC||allanj...@freebsd.org --- Comment

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 me+freebsd@mathieu.digital changed: What|Removed |Added CC|

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Dave Cottlehuber changed: What|Removed |Added CC||d...@freebsd.org

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Mark Linimon changed: What|Removed |Added Attachment #181048|file_216759.txt |x86_x86_tsc.patch

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 and...@azar-a.net changed: What|Removed |Added Attachment #181048||maintainer-approval+

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #17 from and...@azar-a.net --- (In reply to deJong from comment #16) It doesn't seem like the patchset is applied at all. > kern.timecounter.choice: TSC-low(-100) i8254(0) ACPI-fast(900) HPET(950) > dummy(-100) It should

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #16 from deJong --- I have tested the patch set and unfortunately it made no difference root@testbsd11:~ # sysctl kern.timecounter.choice kern.timecounter.choice: TSC-low(-100) i8254(0) ACPI-fast(900)

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 deJong changed: What|Removed |Added CC|

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #14 from and...@azar-a.net --- Bob, this issue is a major one. I had to delve into almost academical works around all available timers and their connection with virtualization. Seems the only "correct" work is done by VMWare

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #13 from bob.caut...@gmail.com --- As an interested party to this bug I have to raise an issue with this potential workaround. First though... thanks everybody who discovered and tested this... BUT

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #12 from and...@azar-a.net --- Ok I tried removing the "hypervisor" feature from CPU and it resulted in 1390MB/s on TSC-low by default in the VPS (same as kvm-clock I must notice)... So seems like FreeBSD does something as not

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #11 from and...@azar-a.net --- Here are the latest results supporting my previous post: root@debian8-test:~# sysbench --num-threads=1 --test=memory --memory-total-size=512M --memory-block-size=1K --debug run sysbench 0.4.12:

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #10 from and...@azar-a.net --- About linux, seems kvm-clock is broken somehow, I adjusted to TSC and got 3700 of the host... And it seems same is for FreeBSD. Only BSD forces HPET or ACPI on any virtualised platform. So there

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #9 from kvanbie...@gmail.com --- Strange. I installed centos 7 on my server and qemu and i had the same lousy performance as i had with proxmox and bsd -- You are receiving this mail because: You are the assignee for the bug.

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #8 from and...@azar-a.net --- In the last post the memory performance is fine (twice slower) because the VPS is CPU limited to 1.7GHz while host is not. -- You are receiving this mail because: You are the assignee for the bug.

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #7 from and...@azar-a.net --- Same config VPS on QEMU with CentOS 7 # sysbench --num-threads=1 --test=memory --memory-total-size=1G run sysbench 0.4.12: multi-threaded system evaluation benchmark Running the test with

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 --- Comment #6 from Roger Pau Monné --- This are the results of the tests on bare-metal using FreeBSD 12 (less than 1 month old): # sysbench --num-threads=1 --test=memory --memory-total-size=4G run sysbench 0.4.12:

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system (under QEMU emulation, but not only)

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Bartek Rutkowski changed: What|Removed |Added Summary|[kern] Memory speed with|[kern]

[Bug 216759] [kern] Memory speed with small blocks (1K) up to 35 times slower than host system under QEMU emulation

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216759 Bartek Rutkowski changed: What|Removed |Added CC|