cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Highlighted
Adventurer
Adventurer
8,686 Views
Registered: ‎04-02-2010

Independent PCS/PMA in the same Quad

Jump to solution

Device: xcku115-flvf1924-2-e-es2

Tool: Vivado 2015.1

IP: 10G Ethernet PCS/PMA version 6.0

 

I'd like to understand the dependency cause by using more than one PCS/PMA core in a GTH quad.  I've generated the core with the shared resources in the example design and found that the some of the shared logic takes reset, etc. from one GTH.  So it looks like there is a master GTH that interfaces with the shared logic and the other 3 GTHs in the quad are slaves that rely on the master.  Is this the case?

 

I'm concerned about data droppages on the slave GTHs caused by events on the master GTH such as removing an SFP.

 

I appreciate your help in clarifying this issue.

Tags (2)
0 Kudos
1 Solution

Accepted Solutions
Highlighted
Moderator
Moderator
16,210 Views
Registered: ‎02-16-2010
In 7 series GTH, there is one QPLL in one quad. So, this needs to be shared by all GT's in the quad. So your understanding about one master and 3 slaves is the way to implement multiple instances of same core in the same quad.

When events like removing SFP happens, the expectation is that only RX portion of the data path is reset which will not disturb the clocking structure of any other data path except for that particular RX path.
------------------------------------------------------------------------------
Don't forget to reply, give kudo and accept as solution
------------------------------------------------------------------------------

View solution in original post

2 Replies
Highlighted
Moderator
Moderator
16,211 Views
Registered: ‎02-16-2010
In 7 series GTH, there is one QPLL in one quad. So, this needs to be shared by all GT's in the quad. So your understanding about one master and 3 slaves is the way to implement multiple instances of same core in the same quad.

When events like removing SFP happens, the expectation is that only RX portion of the data path is reset which will not disturb the clocking structure of any other data path except for that particular RX path.
------------------------------------------------------------------------------
Don't forget to reply, give kudo and accept as solution
------------------------------------------------------------------------------

View solution in original post

Highlighted
Adventurer
Adventurer
8,631 Views
Registered: ‎04-02-2010
That sounds good. Thanks!
0 Kudos