Use c_sdp_bdo_reorder_nof_blocks_max = 4 to save block RAM. Ensure...

Use c_sdp_bdo_reorder_nof_blocks_max = 4 to save block RAM. Ensure c_sdp_bdo_reorder_nof_blocks_max <= nof_bdo_destinations_max in revision. Choose nof_bdo_destinations_max = 8.

applications/lofar2/libraries/sdp/src/vhdl/sdp_bdo_pkg.vhd

@@ -59,11 +59,21 @@ package sdp_bdo_pkg is
   -- nof_destinations > 1 and how many block RAM resources are available for
   -- the reordering. Therefore c_sdp_bdo_reorder_nof_blocks_max = 16 is a
   -- suitable compromise value.
-  constant c_sdp_bdo_reorder_nof_blocks_max : natural := 16;
+  -- Conclusion:
+  -- . Choose c_sdp_bdo_reorder_nof_blocks_max = 4 to save block RAM resources,
+  --   and because it is acceptable to have smaller packets in case of more
+  --   than 4 destinations. Note that the reorder buffer is dual page.
+  -- . Choose nof_bdo_destinations_max = 8 in design revision, because it is
+  --   enough for CEP and to save logic resources and to ease timing closure.
+  -- Remark:
+  -- . The c_sdp_bdo_reorder_nof_blocks_max must be <=
+  --   nof_bdo_destinations_max, because otherwise this probably caused
+  --   Quartus synthesis to yield an segmentation error.
+  constant c_sdp_bdo_reorder_nof_blocks_max : natural := 4;
   constant c_sdp_bdo_reorder_nof_blocks_w   : natural := ceil_log2(c_sdp_bdo_reorder_nof_blocks_max + 1);
   constant c_sdp_bdo_reorder_nof_ch_max     : natural := c_sdp_bdo_reorder_nof_blocks_max *
                                                          c_sdp_nof_beamlets_per_block *
-                                                         c_sdp_nof_words_per_beamlet;  -- = 7808
+                                                         c_sdp_nof_words_per_beamlet;  -- = 4 * 488 * 1 = 1952 32b
 
   -- 32 * 3 + 4 = 100 fields
   constant c_sdp_bdo_destinations_info_nof_hdr_fields : natural := c_sdp_bdo_mm_nof_destinations_max * 3 + 4;