Re: Call for re(4) TSO/VLAN testers

On Mon, Nov 13, 2006 at 04:37:57PM +0100, Pieter de Goeje wrote:
On Saturday 11 November 2006 02:10, Pyun YongHyeon wrote:
With TSO you can see reduced system load while large file transfers
is in progress. Please test the patch and report any unusual things
you've found.

Idle time has increased from 0% to 25% with this patch while sustaining the
transferrate (520Mbit/s) previously reached with iperf.
With the patch the number of interrupts/sec. was halved from 32K to 16K.

The test was performed on a fresh current without any tuning but with debug
support removed.

System details:
CPU: AMD Sempron(tm) 2400+ (1659.95-MHz 686-class CPU)
re0@pci0:9:0: class=0x020000 card=0x816910ec chip=0x816910ec rev=0x10
hdr=0x00
vendor = 'Realtek Semiconductor'
device = 'RTL8169 Gigabit Ethernet Adapter'

The receiving end was probably limiting the transferrate.


Thanks for reporting.

ATM re(4) uses still small number of Tx descriptors(i.e. 64 entries)
due to the hardware limitation of 8139C+. 8169 family can have
up to 1024 Tx descriptors. If we want to use more Tx descriptors
on 8169 the internal structure should be modified to support both
8139C+ and 8169. I guess it would require major Tx path overhaul.
Because I have just plain PCI 8169 hardware I can't sure increasing
number of Tx descriptors on re(4) help Tx performance of the driver.
I guess 64 Tx descriptors are not sufficient to saturate giga bit
link.
Would please try attached patch and report the performance of patched
re(4) driver?

--
Regards,
Pyun YongHyeon
_______________________________________________
freebsd-current@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: igb driver RX (was TX) hangs when out of mbuf clusters
    ... Now outside of the driver, at some point of time, clusters are freed. ... On Feb 9, 2011, at 8:32 AM, Jack Vogel wrote: ... The original PR and patch was kern/150516 from Beezar Liu, ... However I think now that this very report might be behind what we are seeing today. ...
    (freebsd-net)
  • [PATCH v3] hid: egalax: Convert to MT slots
    ... The firmware in the joojoo reports touches sequentially, ... report, which confuses the current driver. ... This is revision three of the patch, with one line changed to accomodate ...
    (Linux-Kernel)
  • [PATCH 0/2] Provide a driver for the Apple Magic Mouse
    ... I think this patch is ready for real review. ... Input Report that isn't published in the input Report descriptor that ... Because the mouse has only one switch and no scroll wheel, the driver ... the reported events. ...
    (Linux-Kernel)
  • Re: [PATCH] power_supply: Add driver for TWL4030/TPS65950 BCI charger
    ... How about this patch + changing the driver to return -ENODATA? ... There are cases when some device can not report any meaningful value, ... kernel log with error messages. ...
    (Linux-Kernel)
  • Re: Major qla2xxx regression on sparc64
    ... you would have seen the warning messages: ... report to driver author so this can be fixed!" ... That verbiage sounds fine -- so would you consider the previous patch ...
    (Linux-Kernel)