Hey Glenn
FWIW, the 10GbE core is due to be overhauled soon. Both the TX and RX
sides will be changed internally, though we'll try to keep the
external interface the same. BRAM requirements will decrease and
timing should be improved.
There is a known problem with the RX fifo buffering whereby it can
lockup if it is allowed to overflow and you don't do a pre-emptive
acknowledge. However, I have never had problems with the TX.
Jan Wagner was having a similar problem on the IBOB last year. I'm not
sure what the outcome was.
Begin forwarded message:
From: Jan Wagner <[email protected]>
Date: 17 February 2008 16:11:03 GMT+02:00
To: John Ford <[email protected]>
Cc: [email protected]
Subject: Re: [casper] 10 GbE config software
Another thing, I'm not totally sure if it is the powerpc program or
something else, but our iBob 10G's don't work very reliably -- the
ARP/ping always works, but the Simulink/hardware level UDP
packetizer sometimes is "stuck" and blasts out 36-byte UDP packets
with frame errors for every frame. Sometimes OTOH after reset it
blasts out the proper 4 Gbits/ 8192-byte UDP stream. If anybody
happens to know a solution, please tell! Perhaps software + hardware
interfere somehow...?
On 27 Mar 2009, at 06:39, G Jones wrote:
Hello,
Every now and again the output of my BEE2 10 GbE port seems to be
garbled. That is, the UDP packets do not show up as UDP packets, or
even as IP packets, just as garbage on the ethernet MAC layer. The
ARP packets are still mostly fine, but usually when this happens,
there will be a few IPs that have garbled ARP packets as well. I
remember talking to someone about this before and it sounded like it
was a known problem. I am using the CASPER version of the core (the
check box indicating the lightweight core with no CRC check on
received UDP packets is checked). Is this indeed a known problem?
How do people deal with it?
Thanks,
Glenn
