UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Observer wjun
Observer
10,115 Views
Registered: ‎02-11-2013

What exactly causes SOFT_ERRORs in GTP?

Hello,

I'm trying to connect a GTX and GTP over aurora, and having some problems.

The main symptom seems to be that rx_resetdone is asserted for a while, and then a long SOFT_RESET comes on,

after which rx_resetdone is dropped low.

 

However, I could not find much information online on what I could be doing wrong that could be causing this...

 

Both sides are using the same data rate and the same refclock.

I'm using vivado 2014.1

 

Any help is greatly appreciated.

 

Thank you!

0 Kudos
8 Replies
Observer wjun
Observer
10,110 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

Quick add-on:

After rx_resetdone_i is asserted,

there is a flurry of data on the rx_d_i and tx_d_i.

It seems the same data on the rx_d_i shows up on th tx_d_i the following cycle.

This continues until "bc4a" shows up on rx_d_i, after which both locations hold "bc4a",

until SOFT_ERRORs kill rx_resetdone_i

 

Thank you!

0 Kudos
Observer wjun
Observer
10,105 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

I seem to have missed some crucial information in my previous posts

 

It is only on the GTP side that this happens.

On the GTX side, both rx_resetdone and tx_resetdone are asserted stably.

 

Loopback works fine for both GTX and GTPs

0 Kudos
Xilinx Employee
Xilinx Employee
10,095 Views
Registered: ‎07-23-2012

Re: What exactly causes SOFT_ERRORs in GTP?

Hi,

Did you see this issue in simulation?

When you see soft errors, it means that you have notintable errors and disparity errors on GT side.

Please trace the data on both TX and RX sides to narrow down the issue.

Since you are using the aurora core, we can rule out the possible cases like mismatch in the comma settings/encoding etc.

Which version of core are using targeting? Try to search for known issues on those core versions.

Please update us on the devices as well.

Regards,
Krishna
-----------------------------------------------------------------------------------------------
Please mark the post as "Accept as solution" if the information provided answers your query/resolves your issue.

Give Kudos to a post which you think is helpful.
0 Kudos
Observer wjun
Observer
10,092 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

Hello,

Thank you for your reply!

 

I am using a VC707, with a custom Artix 7 daughter we have developed.

However, I am also seeing the same problem with a VC707+XM104 and AC701.

 

I am not yet sending any data over aurora. I just wanted to verify that lane_up and channel_up is correctly asserted.

However, that is not the case.

0 Kudos
Xilinx Employee
Xilinx Employee
10,089 Views
Registered: ‎07-23-2012

Re: What exactly causes SOFT_ERRORs in GTP?

Hi Wjun,

Did you check the quality of the link over IBERT? If yes, do you see a continuous link on the lanes of interest?

Can you try to use the example design and see if this works?

Which version of core are you using?

Regards,
Krishna
-----------------------------------------------------------------------------------------------
Please mark the post as "Accept as solution" if the information provided answers your query/resolves your issue.

Give Kudos to a post which you think is helpful.
0 Kudos
Observer wjun
Observer
10,086 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

I am using 10.2 cores on both sides.

 

I could not try ibert because the GTX ibert only allow datapaths of 30 or 40 bits, while the GTP ibert only allows datapath of 16 bits.

 

I am currently running an almost identical version of the code to the example design.

The only change I made was to wrap the entire example design in another module,

and get rid of the frame gen/check modules.

0 Kudos
Observer wjun
Observer
10,080 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

Interestingly, when I checked the VC707 side with a deeper ILA,

It seems that the GTX also has short periods of rx_resetdone_i being deasserted.

This happens at a mich higher frequency than the tx_resetdone being asserted on the GTP side,

and there are no SOFT_ERRORs associated with it.

 

This is very confusing...

0 Kudos
Observer wjun
Observer
10,077 Views
Registered: ‎02-11-2013

Re: What exactly causes SOFT_ERRORs in GTP?

Could this possibly be caused by a clock configuration that is not very amenable with each other?

 

When I try to test various configurations of the GTX/GTP using vio,

(resetting the gt, or configuring loopbac, for example)

ILA sometimes complain with the following message:

 

Unable to arm ILA 'hw_ila_1'. The core clock is slow or no core clock connected for this ILA or the ILA core may not meet timing.

 

And during compilation, I get the following warning, which I don't know if it is important or not:

 

WARNING: [Timing 38-127] No common period was found between clocks user_clk_i  {0 6.0605 12.121} and init_clk_i  {0 10 20} in 1000 cycles.

 

I hope this adds insight into my problem

 

Thank you!

0 Kudos