|
Message
From: Robert Lluís<rlluis@p...>
Date: Mon Jul 3 07:52:28 CEST 2006
Subject: [pci] Issues with the PCI core
Hello,I used to experience the performance problem reported below by Andreas
Ehliar in this mailing-list.
I have fixed this issue by following Ehliar's suggestion -see below. If
someone is interested drop me an e-mail.
Regards,
Robert Lluis
> The asynchronuous FIFO should be modified to include a high watermark
> signal which is set to one if less than 32 words are available in the
> FIFO. The state machine for the wishbone slave should be modified to
> work in the following way:
>
> If a transaction is already under way, ignore the high watermark
> signal. If a transaction has not started, only allow it to start if
> the high watermark signal is zero.
>
> First, a performance problem:
>
> If we (via burst writes from the WB side to the PCI side) fill up the
> asynchronuous fifos in the core the performance will go from high to
> very low and never recover. The reason is that the asynchronuous
> fifos do not deal with anything less than a complete burst
> transaction. If the complete burst fits into the FIFO all is
> well. However, if the burst does not fit into the fifo, the logic
> around the fifo will end the burst transaction and start a new burst
> transaction as soon as there is new space in the fifo. This new burst
> will not be very long since there is very little space in the fifo at
> this point.
>
> This means that as soon as the fifo is full we will not be able to
> write a long burst into the fifo until it is relatively empty. Since
> we will continue writing short bursts into the fifo, the PCI side will
> only issue short bursts and therefore taking a very long time to drain
> the fifo.
>
> An illustration of what this problem looks like is located at
> http://www.da.isy.liu.se/~ehliar/opencores/pciburststuff.png
> Source code for the testbench is at:
> http://www.da.isy.liu.se/~ehliar/opencores/pciburststuff.tar.gz
>
> Perhaps this is the reason why some people claim that the core has
> poor performance? How much work would it be to modify the core to be
> able to merge burst transactions?
>
|
 |
Browse
