Hi All:
There are many papers about the 1D or 2D placement. However, the papers are almost for the algorithm discussion. The current method for dynamically partially reconfigurable architectures is EAPR flow, but the DPR blocks need to be defined at design-time by using the ucf file. All the area constraints are finally included in the static_full and partial bitstreams. So, I am very confused how 1D and 2D placement can be applied to a =93real=94 DPR system at run-time. Are there any research groups who can apply such placement methods in a =93real=94 DPR architecture? I know that some proposed methods, which contained a specific filter to re-modify the location information in the bitstream, can relocate a partial bitstream to a new location. However, the area sizes of DPR blocks need to be the same. However, for example, two DPR blocks are implemented at design-time. Is it possible that the two DPR areas can be merged for placing a partial bitstream that needs both the resources of two DPR blocks like real 1D or 2D placement at run-time? Is there any architecture that can freely be placed the partial bitstreams at run-time? Thanks very much!
Best regards, Huang