Hi, there
We are using a XUP board to implement an algorithm on FPGA right now. Our design is pretty big on the chip. The device utilization summary is listed as follows:
Logic Utilization: Number of Slice Flip Flops: 15,413 out of 27,392 56% Number of 4 input LUTs: 15,218 out of 27,392 55% Logic Distribution: Number of occupied Slices: 12,892 out of 13,696 94% Number of Slices containing only related logic: 12,892 out of
12,892 100% Number of Slices containing unrelated logic: 0 out of 12,892 0% *See NOTES below for an explanation of the effects of unrelated logicEver since our design grew big, some weird results appeared. And if we tried different Placer cost table in the fast_runtime.opt, we got different results. Sometimes good ones. If we optimized the design, and then shrinked it a little bit, then we could get good result without trying difference cost tables. We defined the "good result" by seeing when the maximum frequency reached 100MHz which is what we are running at.
But now, the frequency reached 101.843 MHz, the weird result still appeared. Every time I run the code, we got different result. THe simulation is just fine.
I don't really know what to do next now. I thought we shrinked the design as much as possible.
So anyone has any idea? Is it possible to use Synplify which is said to be better? Will it improve a lot?
Thanks a lot. Zhaoyi