Add packed array slicers

arbitrary_slicer_2d.sv

@@ -0,0 +1,64 @@
+//------------------------------------------------------------------------------
+// arbitrary_slicer_2d.sv
+// Konstantin Pavlov, pavlovconst@gmail.com
+//------------------------------------------------------------------------------
+
+// INFO -------------------------------------------------------------------------
+// Arbitrary slicer for 2D packed SystemVerilog arrays
+// Slices along any array edge, all array sides simultaneously
+//
+// You can also generalize this aproach to support as many dimentions as needed
+//
+// This module does NOT consume any FPGA resources though it is absolutely
+// synthesizable
+//
+
+
+/* --- INSTANTIATION TEMPLATE BEGIN ---
+
+arbitrary_slicer_2d #(
+  .I2_HI( 3 ),  .I2_LO( 0 ),
+  .I1_HI( 7 ),  .I1_LO( 0 ),
+
+  .O2_HI( 2 ),  .O2_LO( 1 ),
+  .O1_HI( 2 ),  .O1_LO( 1 )
+) S1 (
+  .in ( a[3:0][7:0] ),
+  .out( b[2:1][2:1] )
+);
+
+--- INSTANTIATION TEMPLATE END ---*/
+
+
+module arbitrary_slicer_3d #( parameter
+
+  // input array shape
+  I2_HI = 3,  // most significant size
+  I2_LO = 0,
+
+  I1_HI = 7,  // least significant size
+  I1_LO = 0,
+
+  // sliced array shape
+  O2_HI = 2,  // most significant  size
+  O2_LO = 1,
+
+  O1_HI = 2,  // least significant size
+  O1_LO = 1
+)(
+  input        [I2_HI:I2_LO][I1_HI:I1_LO] in,
+  output logic [O2_HI:O2_LO][O1_HI:O1_LO] out
+);
+
+
+  integer i;
+  always_comb begin
+
+      for ( i=O2_LO; i<=O2_HI; i++ ) begin
+        out[i][O1_HI:O1_LO] = in[i][O1_HI:O1_LO];
+      end
+
+  end
+
+endmodule
+

arbitrary_slicer_3d.sv

@@ -0,0 +1,74 @@
+//------------------------------------------------------------------------------
+// arbitrary_slicer_3d.sv
+// Konstantin Pavlov, pavlovconst@gmail.com
+//------------------------------------------------------------------------------
+
+// INFO -------------------------------------------------------------------------
+// Arbitrary slicer for 3D packed SystemVerilog arrays
+// Slices along any array edge, all array sides simultaneously
+//
+// You can also generalize this aproach to support as many dimentions as needed
+//
+// This module does NOT consume any FPGA resources though it is absolutely
+// synthesizable
+//
+
+
+/* --- INSTANTIATION TEMPLATE BEGIN ---
+
+arbitrary_slicer_3d #(
+  .I3_HI( 3 ),  .I3_LO( 0 ),
+  .I2_HI( 3 ),  .I2_LO( 0 ),
+  .I1_HI( 7 ),  .I1_LO( 0 ),
+
+  .O3_HI( 2 ),  .O3_LO( 1 ),
+  .O2_HI( 2 ),  .O2_LO( 1 ),
+  .O1_HI( 2 ),  .O1_LO( 1 )
+) S1 (
+  .in ( a[3:0][3:0][7:0] ),
+  .out( b[2:1][2:1][2:1] )
+);
+
+--- INSTANTIATION TEMPLATE END ---*/
+
+
+module arbitrary_slicer_3d #( parameter
+
+  // input array shape
+  I3_HI = 3,  // most significant size
+  I3_LO = 0,
+
+  I2_HI = 3,
+  I2_LO = 0,
+
+  I1_HI = 7,  // least significant size
+  I1_LO = 0,
+
+  // sliced array shape
+  O3_HI = 2,  // most significant  size
+  O3_LO = 1,
+
+  O2_HI = 2,
+  O2_LO = 1,
+
+  O1_HI = 2,  // least significant size
+  O1_LO = 1
+)(
+  input        [I3_HI:I3_LO][I2_HI:I2_LO][I1_HI:I1_LO] in,
+  output logic [O3_HI:O3_LO][O2_HI:O2_LO][O1_HI:O1_LO] out
+);
+
+
+  integer i,j;
+  always_comb begin
+
+    for ( i=O3_LO; i<=O3_HI; i++ ) begin
+      for ( j=O2_LO; j<=O2_HI; j++ ) begin
+        out[i][j][O1_HI:O1_LO] = in[i][j][O1_HI:O1_LO];
+      end //j
+    end //i
+
+  end
+
+endmodule
+