Hi,
On Sat, Jul 4, 2015 at 7:27 AM, mo.ucina mo.ucina@gmail.com wrote:
Hello Guys,
@hdegoede
I was just wandering where we stand with having the u-boot including a specific definition for the R1 . I noticed that last month, Jun 15 there was a patch submitted by 'fabioca' to include the R1 into u-boot "New sunxi board: Lamobo Bananapi R1 http://u-boot.10912.n7.nabble.com/New-sunxi-board-Lamobo-Bananapi-R1-tp217120.html" . Is there a possibility that this could get into the next u-boot release v2015.07 ? The reason is that , the R1 needs more specific treatment of the CONFIG_GMAC_TX_DELAY then the rest of the bananapi family . So it would be good to have Fabio's patch in but perhaps with a different value for CONFIG_GMAC_TX_DELAY=4 , as suggested by Thomas Kaiser here :
http://comments.gmane.org/gmane.comp.hardware.netbook.arm.sunxi/17495
and Igor P. here :
https://github.com/igorpecovnik/lib/blob/next/patch/add-lamobo-r1-uboot.patc...
The layout for the R1 board is different to that of the other bananapi boards , so perhaps the setting of 4 is more appropriate . The best throughput testing that I have achieved with the R1 , using the u-boot v2015.04 (which has the generic bananapi setting of CONFIG_GMAC_TX_DELAY=3) is 290 Mbits/sec :
iperf -c 192.168.1.153 -t 30 -u -b 1g
Client connecting to 192.168.1.153, UDP port 5001 Sending 1470 byte datagrams, IPG target: 11.76 us (kalman adjust) UDP buffer size: 160 KByte (default)
[ 3] local 192.168.1.1 port 58279 connected with 192.168.1.153 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.0 sec 1.01 GBytes 290 Mbits/sec [ 3] Sent 738867 datagrams [ 3] Server Report: [ 3] 0.0-30.0 sec 1.01 GBytes 290 Mbits/sec 0.056 ms 7/738867 (0.00095%)
Perhaps with setting of 4 , we can improve this . On the wiki Thomas mentions "noone achieved more than 370/460 Mbits/sec TX/RX using iperf" with this setting . Which is quite a bit better than what I am getting now . It may not be the whole solution , but definitely a step in the right direction
There's discussion on this upstream [1] and once there's an agreement on this for both u-boot and the kernel DT we can look to get it included for F-23.
[1] http://lists.denx.de/pipermail/u-boot/2015-July/218249.html