08-01-2019 07:49 AM
We find that the MIPI RX Subsystem is taking alot of the address space takes on the order of 0x1_0000 bytes.
This seems like alot consider the fact that the 3 components that make up the core each only take about 0x100 bytes.
is there a way to change the internal crossbar to have an address width of 0x1000 instead of 0x1_0000? (4kB instead of 64kB).
We can make an external I2C interface, but we cannot have both control of the RX controller (Hidden IP), and the D-PHY address space without using the internal crossbar.
Using a whole 128kB of address space is really wasteful when 8kB would more than suffice and not be too far in the "over optimization" realm.
Is this reasonable to expect in a near-term release?
08-07-2019 10:43 PM
08-02-2019 12:02 AM
Hello Mark @markramona
Thank you for your feedback. I understand your concern.
Unfortunately, This is not a bug and we do not have any milestone to update MIPI CSI-2 RX Subsystem register space.
If this un-optimized address space of MIPI CSI-2 IP halted down your project,
Please escalate this feature-enhacement request via your FAE.
08-02-2019 06:53 AM
There is no immediate need to have a more optimized space, it just seems like the MIPI subsystem is something you are actively working on and this is one of the (very small) pain points with have with it.
I know it requires coordination between writing the driver and the hardware, so it isn't exactly easy.
Thanks for taking the time to respond.
08-05-2019 04:08 AM
I do not think this is possible in near term.
The MIPI CSI2 RX IP is a subsystem, this means that it is consitued by multiple IP. As each IP need to have at least 4KB address space (minimum for an AXI4), the wrapper needs to be higher. This way the sub-IPs can still be addressed individually.
08-05-2019 06:17 AM
> As each IP need to have at least 4KB address space (minimum for an AXI4)
I'm asking only that each of the sub IPs take 4kB. Currently, each one is taking 64kB, a huge increase in the required address space.
08-07-2019 10:43 PM
08-08-2019 07:01 AM
08-13-2019 02:02 AM
If your question is answered or your issue is solved, please kindly mark the response which helped as solution (click on "Accept as solution" button below the reply)
If this is not solved/answered, please reply in the topic giving more information on your current status.
Thanks and Regards,
03-01-2020 10:40 AM
Second, I can't seem to find where in the AXI specification it requires 4kB address space by a single slave.
It seems to me that much of this address space could be compressed quite a bit, and saving quite a few resources, at least in terms of address space from the master.
03-01-2020 06:33 PM
1. I have passed this feature/enhancement request to our developer team. Unfortunately, there is no target milestone for this,
If this request if very critical for your project, would you please contact your FAE to escalate this request to our Marketing/Developer team.
2. Yes, I think your understanding of no minimum address requirement is correct.
Thanks & regards
03-02-2020 05:16 AM
Thanks for the Update Leo.
I can wait for 2020.1.
It would be good to eventually get rid of the hacks we added to compress the address space.