Re: Dom0 bad network performance

[Matthias Petermann <mp%petermann-it.de@localhost>]

Hello Greg,

Am 13.02.21 um 17:15 schrieb Greg Troxel:
> These are quite different!   The values are more or less in the same
> relationship as your tests using re0.  But they are also far more than
> the most-of-1Gb/s you might reasonably expect on an actual interface.
>
> You might also try dd from /dev/zero to /dev/null.  I'm seeing 16 GB/s
> on a sort of old box, 8 GB/s on older hardware, and maybe 3 on a PC
> Engines apu2.  That should test the user/kernel data copying without
> dragging TCP into it.

Using

	dd if=/dev/zero of=/dev/null bs=4m

it performs like:

	2,163,008,730 bytes/sec => ~ 2.2 GB/s (with Xen)
	2,402,700,957 bytes/sec => ~ 2.4 GB/s (without Xen)

The difference here is not significant, but compared to your example 
values on a lower level overall. This puts your question about the 
specific hardware in a very special light. It is about this pretty low 
end CPU built into a little NUC box:

[     1.000003] ioapic0 at mainbus0 apid 1: pa 0xfec00000, version 0x20, 
115 pins
[     1.000003] cpu0 at mainbus0 apid 0
[     1.000003] cpu0: Intel(R) Celeron(R) CPU  N3050  @ 1.60GHz, id 0x406c3
[     1.000003] cpu0: package 0, core 0, smt 0
[     1.000003] cpu1 at mainbus0 apid 4
[     1.000003] cpu1: Intel(R) Celeron(R) CPU  N3050  @ 1.60GHz, id 0x406c3
[     1.000003] cpu1: package 0, core 2, smt 0
[     1.000003] acpi0 at mainbus0: Intel ACPICA 20190405
[     1.000003] acpi0: X/RSDT: OemId <INTEL ,NUC5CPYB,0000004f>, AslId 
<AMI ,00010013>
[     1.000003] acpi0: autoconfiguration error: invalid PCI address for D026
[     1.000003] acpi0: MCFG: segment 0, bus 0-255, address 
0x00000000e0000000

I am running Xen 4.13 on 9-stable.

Anyway, I will continue to check the netstat output for my next test 
according your proposal.

Kind regards
Matthias