Re: kern/42314: ipgphy support

[Frank Wille <frank%phoenix.owl.de@localhost>]

On Tue, 18 Aug 2015 15:13:17 +0200
Felix Deichmann <m4j0rd0m0%gmail.com@localhost> wrote:

> 2015-08-18 14:35 GMT+02:00 Robert Swindells <rjs%fdy2.co.uk@localhost>:
> > I have got it to build, starting from the latest version from OpenBSD,
> > are you able to test it ?
> 
> Almost, I will get hardware until next week (and also plan to look
> into the derelict stge(4) if nobody with more experience is willing
> to).

What's wrong with stge(4)?


> /sandpoint's Frank Wille has a DSM-G600 AFAIK, so he might be faster.

I compiled a recent generic sandpoint kernel with "patch-ipgphy-20100606"
applied, to test it on my G600. It works, but the result is diappointing.
The read-performance dropped dramatically and the write-performance doesn't
increase either.

FTP test results with 7.99.16:

	read	write	(MiB/sec)
ukphy	4.64	4.24
ipgphy	0.13	4.21

From my dmesg (ipgphy attached):

stge0 at pci0 dev 15 function 0: Sundance ST-1023 Gigabit Ethernet (rev. 0x41)
stge0: interrupting at irq 19
stge0: Ethernet address 00:13:46:da:xx:xx
ipgphy0 at stge0 phy 24: IP1000A 10/100/1000 PHY, rev. 0
ipgphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseT, 1000baseT-FDX, auto


It might be good when somebody repeats the test. It could also be caused
by strange effects in my network infrastructure. For example the read-
performance doubles when I do sftp instead of ftp...

-- 
Frank Wille