Xilinx place and route cost table

What about Xilinx place and route cost table mean ? Or someone can give any information not just a number ? Thank you!

Generally there are 100 variations of the place and route algorithm otherwise known as cost tables. Which one is better for any given design varies with the design. Unless you have an issue meeting timing, or want to try down the speed grade of a design, then you don't need to bother with them and just use the default single run setting.

The ISE tools support multiple place and route (MPPR) which uses a number of these cost tables and tells you which it thinks did best. As far as I am aware the details of the tables are not public domain although Peter or Austin may say otherwise.

John Adair Enterpoint Ltd. - Home of Broaddown2. The Ultimate Spartan3 Development Board.

Think of the cost table as simply a random seed for the placement algorithm.

I am not sure if this is still the case, but as of 2-3 years ago, placement was not timing driven, and was a somewhat stochastic process. Routing has always been timing driven.

-- Regards, John Retta Owner and Designer Retta Technical Consulting Inc. Colorado Based Xilinx Consultant

email : snipped-for-privacy@rtc-inc.com web :

I have noticed over the past year or two with the current xilinx tools that when I run multi pass P&R, lots and lots of the passes will have different seed numbers, but it looks like they are the same route! By that I mean that all dozen or so of my timing constraints come out the same to 5 figures precision. If all the timespecs are identical, it must be the same route, right? In the past, with designs close to passing constraints I could run 30 MPPR passes and count on getting one that passes all constraints. But now when running 30 passes I only get 4 or 5 unique routes. Has anyone else noticed this phenomenon? I couldn't find mention of this in the answers database.

-Jeff