Hi, John,
It's odd that just_built.jpeg doesn't show "ROACH" (or "ROACH2") in the XSG
block. I think missing that is what leads to the error dialog shown in
error.jpg. In after_ok.jpeg, the XSG block has "ROACH" in it.
Does the model build OK now? FWIW, I've used 14.2 with the new Shared BRAM
stuff and it's worked out OK for me.
The new Shared BRAM block can have a different latency from the old one
depending on which new options are checked ("Register Primitive Output" and
"Register Core Output"). The old Shared BRAM always had a latency of 1. If
that's important to your design, be sure to disable (i.e. UN-check) these two
new options in the mask dialog of the new Shared BRAM block. In the mlib_devel
version you have I believe these two new options default to enabled (i.e.
checked). Also in that version, the simulated latency of the new Shared BRAM
block does not properly account for the settings of these new options (it
always simulates with latency 1 even if these options are checked yet they each
add 1 to the block's overall latency).
Andrew Martens and I are working to get out a new mlib_devel version that will,
among other improvements, address (no pun intended!) these issues of the new
Shared BRAM block.
Hope this helps,
Dave
On Apr 25, 2013, at 2:35 PM, John Ford wrote:
> Hi all. Today I rebuilt a simple test model from scratch. It acts
> different from the one I had yesterday, but it's still fubar.
>
> The model is shown in just_built.jpeg
>
> When I try to modify the block and "apply" the change I get the error
> message in error.jpeg
>
> When I tried to simulate it I got:
>
> System Generator currently found installed into matlab default path
> Available System Generator installations:
> Current version of System Generator is 14.3.4477.
> Run << xlVersion >> at prompt to see installed versions of System Generator
> Warning: In instantiating linked block 'jford_sbram/Shared BRAM/mem/ram' :
> Xilinx Single Port Random Access Memory
> Block block (mask) does not have a parameter named 'use_rpm'
>
> Then matlab apologized profusely and crashed.
>
> I then started matlab again, loaded the model, and I was able to open the
> model and modify it, but now there's a warning and the munge_in and
> munge_out blocks under the mask have the link broken warning.
>
> A second screen shot is attached as after_ok.jpeg
>
> A third screen shot shows the innards of the yellow block as yellow.jpeg
>
> I attached the model for good measure.
>
>
> Here's info from git log which may shed light:
>
> commit e56543ee4997e04d2a84e17cab6b1c5a2a68ca1f
> Author: wnew <wes...@ska.ac.za>
> Date: Tue Mar 5 14:54:36 2013 +0200
>
> Updating the shared bram custom block
>
> Added the optimization option
> Added the option to register the outputs of the core and the primitives
>
> .../data/bram_block_custom_v2_1_0.mpd | 12 +-
> .../data/bram_block_custom_v2_1_0.tcl | 161 +++--
> .../bram_block_custom_v1_00_a/hdl/verilog/bram.v | 290 +++++----
> .../hdl/verilog/bram_block_custom.v | 9 +-
> .../data/bram_block_custom_v2_1_0.mpd | 4 +
> .../data/bram_block_custom_v2_1_0.tcl | 66 ++-
> .../bram_block_custom_v1_00_a/hdl/verilog/bram.v | 289 +++++----
> .../hdl/verilog/bram_block_custom.v | 3 +
> xps_library/@xps_bram/gen_mhs_ip.m | 20 +-
> xps_library/@xps_bram/xps_bram.m | 16 +-
> xps_library/xps_library.mdl | 670
> +++++++++++---------
> 11 files changed, 917 insertions(+), 623 deletions(-)
>
>
> Obviously something's out of sync...
>
> Do I need ISE 14.4 and 2012b?
>
> John
> <just_built.jpeg><error.jpeg><after_ok.jpeg><yellow.jpeg>
