ludan
Newbie level 5
Hi buddies,
let me try to briefly summarize what I do before illustrating the problem.
I have a design with 16 sub-blocks.
1) the design is partitioned and 16 partitions are saved
2) I work block by block doing place and route
3) when all the blocks are placed&routed I reopen the main design and do "assembleDesign" to reimport all the previous 16 elaborated blocks
4) now all the cells in the 16 blocks are placed and routed, I should do just CTS and routing at top level, right?
there are 2 ways (in my mind at least) to do this:
1) committing the 16 partitions thus creating fences, then do CTS and route (this takes me to a funny result: not all the parasites are extracted and the wires connecting the 16 blocks to each other are straight (no buffers in between).
2) the other method is to avoid re-committing the partitions and do CTS by using the PreservePin statement (this avoids that CTS changes what you did previously in each of the 16 sub-block). This just works fine. The problem is with the routing because not having partitions, I am afraid that all the wires are re-routed (takes a lot of time) although they were already routed in the previous elaboration I did block by block.
Shall I use a set_dont_touch somewhere?
Concerning the problem of straight wires (no buffers) between each block, do you think it is a constraint problem or am I missing something?
Thanks for your time and thanks in advance for reading/answering
Daniele
let me try to briefly summarize what I do before illustrating the problem.
I have a design with 16 sub-blocks.
1) the design is partitioned and 16 partitions are saved
2) I work block by block doing place and route
3) when all the blocks are placed&routed I reopen the main design and do "assembleDesign" to reimport all the previous 16 elaborated blocks
4) now all the cells in the 16 blocks are placed and routed, I should do just CTS and routing at top level, right?
there are 2 ways (in my mind at least) to do this:
1) committing the 16 partitions thus creating fences, then do CTS and route (this takes me to a funny result: not all the parasites are extracted and the wires connecting the 16 blocks to each other are straight (no buffers in between).
2) the other method is to avoid re-committing the partitions and do CTS by using the PreservePin statement (this avoids that CTS changes what you did previously in each of the 16 sub-block). This just works fine. The problem is with the routing because not having partitions, I am afraid that all the wires are re-routed (takes a lot of time) although they were already routed in the previous elaboration I did block by block.
Shall I use a set_dont_touch somewhere?
Concerning the problem of straight wires (no buffers) between each block, do you think it is a constraint problem or am I missing something?
Thanks for your time and thanks in advance for reading/answering
Daniele
