cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
zhangjun_beijing
Observer
Observer
481 Views
Registered: ‎05-27-2016

Why the max rate of aurora ip is 6.6Gbps, and GT ip's is 8.5Gbps ?

Jump to solution

Why the max rate of aurora ip is 6.6Gbps, and GT ip's is 8.5Gbps ?

because aurora ip implement more funtion? like CRC?

 

Thanks

0 Kudos
1 Solution

Accepted Solutions
drjohnsmith
Teacher
Teacher
453 Views
Registered: ‎07-09-2009

Good question,

As you say, the two are very different, 

 

The GT SERDES are the raw transvers,

      as each chip comes out, the SERDES tend to get faster.

 

Aurora is as you say, a complete IP, including all the flow control, CRC, state machines for synchronisation etc.

    Most of the IP is actually done in the logic of the FPGA, which as you imply , is a lot slower than the SERDES,

        and works on the parallel data,

An aurora IP "technically" is speed limited by the speed of the silicon,

    In a -1 chip the aurora will run slower than it could in a -3 speed chip.

     

So aurora "technically" is not speed limited,

   BUT

IP tends to be updated on an erratic schedule.

 

Not knowing where you have the quotes from, remember that you could be looking at two different numbers.

Any / most serial links has an overhead due to line encoding, 

  8:10 , framing, clock slip or what ever.

So the raw rate on the serial link is always much higher than the raw data rate 

   

If you can give us the references to where your looking , we might be able to help more.

   but bottom line, I'd just accept it.

 

 

<== If this was helpful, please feel free to give Kudos, and close if it answers your question ==>

View solution in original post

4 Replies
joancab
Advisor
Advisor
468 Views
Registered: ‎05-11-2015

Why not like that? Why cars can go faster than the maximum legal speed?

0 Kudos
drjohnsmith
Teacher
Teacher
454 Views
Registered: ‎07-09-2009

Good question,

As you say, the two are very different, 

 

The GT SERDES are the raw transvers,

      as each chip comes out, the SERDES tend to get faster.

 

Aurora is as you say, a complete IP, including all the flow control, CRC, state machines for synchronisation etc.

    Most of the IP is actually done in the logic of the FPGA, which as you imply , is a lot slower than the SERDES,

        and works on the parallel data,

An aurora IP "technically" is speed limited by the speed of the silicon,

    In a -1 chip the aurora will run slower than it could in a -3 speed chip.

     

So aurora "technically" is not speed limited,

   BUT

IP tends to be updated on an erratic schedule.

 

Not knowing where you have the quotes from, remember that you could be looking at two different numbers.

Any / most serial links has an overhead due to line encoding, 

  8:10 , framing, clock slip or what ever.

So the raw rate on the serial link is always much higher than the raw data rate 

   

If you can give us the references to where your looking , we might be able to help more.

   but bottom line, I'd just accept it.

 

 

<== If this was helpful, please feel free to give Kudos, and close if it answers your question ==>

View solution in original post

shantmoses
Contributor
Contributor
417 Views
Registered: ‎07-01-2008

I think because the Aurora 8b/10b IP uses the MGT CPLLs and those were limited to 6.6 Gbps in the 7-series GTX transceivers.

zhangjun_beijing
Observer
Observer
329 Views
Registered: ‎05-27-2016

I want to implement FC protocol ( ADVB, ASM...) on 7 series FPGA. It's be more easy using Aurora ip than GT IP.   I guess,  I can't get  8 Gbps rate even if using GT IP, because the Aurora IP's max rate have showed.

0 Kudos