[Sbs] SRS update by Jessie Twigger

[Ole Hansen]

Hi Kondo,

thanks for the PDF. This is informative. So it takes a reimplementation
of the kernel UDP stack to (almost) make it work. Looks like these guys
really really do know what they are doing, and it really is a Big Deal.

Which makes me wonder, couldn't one simply put a network coprocessor
chip on the front end boards? I can see why implementing a TCP stack in
an FPGA is hard, but there are dedicated chips for just that.

Ole

On 07/02/2014 12:11 PM, Kondo Gnanvo wrote:
> Hi Ole,
> There was a talk by a TOTEM (LHC) experiment DAQ group at the last RD51
> coll. meeting at CERN
> addressing this issue. Please find attached pdf.
> I think Alex is already talking with the TOTEM guys about it
> 
> Best
> Kondo
> 
> On 7/2/2014 12:03 PM, Ole Hansen wrote:
>> On page 3: "data transfer via UDP"
>>
>> That worries me. UDP is by definition "unreliable", i.e. dropped packets
>> are not detected. Even if you allow packet loss, you wouldn't be able to
>> determine dead time, for example. Is there any sort of handshake for
>> data transfers, or do the front ends just send and forget?
>>
>> Just some thoughts.
>>
>> Ole
>>
>>
>>
>> 
-- 
Jens-Ole Hansen <ole at jlab.org> +1-757-269-7627
Staff Physicist
Jefferson Lab, Suite 4, Room 12/C107
12000 Jefferson Avenue, Newport News, VA 23606, USA

PGP: 7F74 730E 86F0 FA76 2588 F901 3A4D 3771 F121 ECCA

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
Url : https://mailman.jlab.org/pipermail/sbs/attachments/20140702/883884fb/attachment.bin