Add mul2dsp multiplier splitting rule and ECP5 mapping

Signed-off-by: David Shah <dave@ds0.me>

1 files changed, 237 insertions, 0 deletions

diff --git a/techlibs/common/mul2dsp.v b/techlibs/common/mul2dsp.v
new file mode 100644
index 000000000..37ce2e485
--- /dev/null
+++ b/techlibs/common/mul2dsp.v
@@ -0,0 +1,237 @@
+// From Eddie Hung

+// extracted from: https://github.com/eddiehung/vtr-with-yosys/blob/vtr7-with-yosys/vtr_flow/misc/yosys_models.v#L220

+// revised by Andre DeHon

+// further revised by David Shah

+`ifndef DSP_A_MAXWIDTH

+`define DSP_A_MAXWIDTH 18

+`endif

+`ifndef DSP_A_MAXWIDTH

+`define DSP_B_MAXWIDTH 25

+`endif

+

+`ifndef ADDER_MINWIDTH

+`define ADDER_MINWIDTH AAA

+`endif

+

+`ifndef DSP_NAME

+`define DSP_NAME M18x25

+`endif

+

+`define MAX(a,b) (a > b ? a : b)

+`define MIN(a,b) (a < b ? a : b)

+

+(* techmap_celltype = "$mul" *)

+module \$mul (A, B, Y); 

+	parameter A_SIGNED = 0;

+	parameter B_SIGNED = 0;

+	parameter A_WIDTH = 1;

+	parameter B_WIDTH = 1;

+	parameter Y_WIDTH = 1;

+

+	input [A_WIDTH-1:0] A;

+	input [B_WIDTH-1:0] B;

+	output [Y_WIDTH-1:0] Y;

+

+	wire [1023:0] _TECHMAP_DO_ = "proc; clean";

+

+  generate

+    if (A_WIDTH<B_WIDTH) begin

+	generate

+		\$__mul_gen #(

+			.A_SIGNED(A_SIGNED),

+			.B_SIGNED(B_SIGNED),

+			.A_WIDTH(A_WIDTH),

+			.B_WIDTH(B_WIDTH),

+			.Y_WIDTH(Y_WIDTH)

+		) mul_slice (

+			.A(A),

+			.B(B),

+			.Y(Y[Y_WIDTH-1:0])

+		);

+	endgenerate

+	end

+    else begin

+	generate

+		\$__mul_gen #(

+			.A_SIGNED(B_SIGNED),

+			.B_SIGNED(A_SIGNED),

+			.A_WIDTH(B_WIDTH),

+			.B_WIDTH(A_WIDTH),

+			.Y_WIDTH(Y_WIDTH)

+		) mul_slice (

+			.A(B),

+			.B(A),

+			.Y(Y[Y_WIDTH-1:0])

+		);

+	endgenerate

+     end

+   endgenerate	

+endmodule

+

+module \$__mul_gen (A, B, Y);

+	parameter A_SIGNED = 0;

+	parameter B_SIGNED = 0;

+	parameter A_WIDTH = 1;

+	parameter B_WIDTH = 1;

+	parameter Y_WIDTH = 1;

+

+	input [A_WIDTH-1:0] A;

+	input [B_WIDTH-1:0] B;

+	output [Y_WIDTH-1:0] Y;

+

+	wire [1023:0] _TECHMAP_DO_ = "proc; clean";

+

+	generate

+	if (A_WIDTH > `DSP_A_MAXWIDTH) begin

+			localparam n_floored = A_WIDTH/`DSP_A_MAXWIDTH;

+			localparam n = n_floored + (n_floored*`DSP_A_MAXWIDTH < A_WIDTH ? 1 : 0);

+			wire [`DSP_A_MAXWIDTH+B_WIDTH-1:0] partial [n-1:1];

+			wire [Y_WIDTH-1:0] partial_sum [n-2:0];

+

+			\$__mul_gen #(

+				.A_SIGNED(A_SIGNED),

+				.B_SIGNED(B_SIGNED),

+				.A_WIDTH(`DSP_A_MAXWIDTH),

+				.B_WIDTH(B_WIDTH),

+				.Y_WIDTH(B_WIDTH+`DSP_A_MAXWIDTH)

+			) mul_slice_first (

+				.A(A[`DSP_A_MAXWIDTH-1:0]),

+				.B(B),

+				.Y(partial_sum[0][B_WIDTH+`DSP_A_MAXWIDTH-1:0])

+			);

+                        assign partial_sum[0][Y_WIDTH-1:B_WIDTH+`DSP_A_MAXWIDTH]=0;

+

+			genvar i;

+			generate

+			for (i = 1; i < n-1; i=i+1) begin:slice

+				\$__mul_gen #(

+					.A_SIGNED(A_SIGNED),

+					.B_SIGNED(B_SIGNED),

+					.A_WIDTH(`DSP_A_MAXWIDTH),

+					.B_WIDTH(B_WIDTH),

+					.Y_WIDTH(B_WIDTH+`DSP_A_MAXWIDTH)

+				) mul_slice (

+					.A(A[(i+1)*`DSP_A_MAXWIDTH-1:i*`DSP_A_MAXWIDTH]),

+					.B(B),

+					.Y(partial[i][B_WIDTH+`DSP_A_MAXWIDTH-1:0])

+				);

+				//assign partial_sum[i] = (partial[i] << i*`DSP_A_MAXWIDTH) + partial_sum[i-1];

+				assign partial_sum[i] = {

+					partial[i][B_WIDTH+`DSP_A_MAXWIDTH-1:0]

+					+ partial_sum[i-1][Y_WIDTH-1:(i*`DSP_A_MAXWIDTH)],

+					partial_sum[i-1][(i*`DSP_A_MAXWIDTH)-1:0]

+				};

+			end

+			endgenerate

+

+			\$__mul_gen #(

+				.A_SIGNED(A_SIGNED),

+				.B_SIGNED(B_SIGNED),

+				.A_WIDTH(A_WIDTH-(n-1)*`DSP_A_MAXWIDTH),

+				.B_WIDTH(B_WIDTH),

+				.Y_WIDTH(A_WIDTH-(n-1)*`DSP_A_MAXWIDTH+B_WIDTH),

+			) mul_slice_last (

+				.A(A[A_WIDTH-1:(n-1)*`DSP_A_MAXWIDTH]),

+				.B(B),

+				.Y(partial[n-1][A_WIDTH-(n-1)*`DSP_A_MAXWIDTH+B_WIDTH-1:0])

+			);

+			//assign Y = (partial[n-1] << (n-1)*`DSP_A_MAXWIDTH) + partial_sum[n-2];

+			assign Y = {

+				partial[n-1][A_WIDTH-(n-1)*`DSP_A_MAXWIDTH+B_WIDTH:0]

+				+ partial_sum[n-2][Y_WIDTH-1:((n-1)*`DSP_A_MAXWIDTH)],

+				partial_sum[n-2][((n-1)*`DSP_A_MAXWIDTH)-1:0]

+			};

+		end

+		else if (B_WIDTH > `DSP_B_MAXWIDTH) begin

+			localparam n_floored = B_WIDTH/`DSP_B_MAXWIDTH;

+			localparam n = n_floored + (n_floored*`DSP_B_MAXWIDTH < B_WIDTH ? 1 : 0);

+			wire [A_WIDTH+`DSP_B_MAXWIDTH-1:0] partial [n-1:1];

+			wire [Y_WIDTH-1:0] partial_sum [n-2:0];

+

+			\$__mul_gen #(

+				.A_SIGNED(A_SIGNED),

+				.B_SIGNED(B_SIGNED),

+				.A_WIDTH(A_WIDTH),

+				.B_WIDTH(`DSP_B_MAXWIDTH),

+				.Y_WIDTH(A_WIDTH+`DSP_B_MAXWIDTH)

+			) mul_first (

+				.A(A),

+				.B(B[`DSP_B_MAXWIDTH-1:0]),

+				.Y(partial_sum[0][A_WIDTH+`DSP_B_MAXWIDTH-1:0])

+			);

+                        assign partial_sum[0][Y_WIDTH-1:A_WIDTH+`DSP_B_MAXWIDTH]=0;

+

+			genvar i;

+			generate

+			for (i = 1; i < n-1; i=i+1) begin:slice

+				\$__mul_gen #(

+					.A_SIGNED(A_SIGNED),

+					.B_SIGNED(B_SIGNED),

+					.A_WIDTH(A_WIDTH),

+					.B_WIDTH(`DSP_B_MAXWIDTH),

+				        .Y_WIDTH(A_WIDTH+`DSP_B_MAXWIDTH)

+				) mul (

+					.A(A),

+					.B(B[(i+1)*`DSP_B_MAXWIDTH-1:i*`DSP_B_MAXWIDTH]),

+					.Y(partial[i][A_WIDTH+`DSP_B_MAXWIDTH-1:0])

+				);

+				//assign partial_sum[i] = (partial[i] << i*`DSP_B_MAXWIDTH) + partial_sum[i-1];

+                                // was:

+				//assign partial_sum[i] = {

+				//  partial[i][A_WIDTH+`DSP_B_MAXWIDTH-1:`DSP_B_MAXWIDTH], 

+				//	partial[i][`DSP_B_MAXWIDTH-1:0] + partial_sum[i-1][A_WIDTH+(i*`DSP_B_MAXWIDTH)-1:A_WIDTH+((i-1)*`DSP_B_MAXWIDTH)],

+				//	partial_sum[i-1][A_WIDTH+((i-1)*`DSP_B_MAXWIDTH):0]

+				assign partial_sum[i] = {

+					partial[i][A_WIDTH+`DSP_B_MAXWIDTH-1:0]

+					+ partial_sum[i-1][Y_WIDTH-1:(i*`DSP_B_MAXWIDTH)],

+					partial_sum[i-1][(i*`DSP_B_MAXWIDTH)-1:0] 

+				};

+			end

+			endgenerate

+

+			\$__mul_gen #(

+				.A_SIGNED(A_SIGNED),

+				.B_SIGNED(B_SIGNED),

+				.A_WIDTH(A_WIDTH),

+				.B_WIDTH(B_WIDTH-(n-1)*`DSP_B_MAXWIDTH),

+				.Y_WIDTH(A_WIDTH+B_WIDTH-(n-1)*`DSP_B_MAXWIDTH)

+			) mul_last (

+				.A(A),

+				.B(B[B_WIDTH-1:(n-1)*`DSP_B_MAXWIDTH]),

+				.Y(partial[n-1][A_WIDTH+B_WIDTH-(n-1)*`DSP_B_MAXWIDTH-1:0])

+			);

+                        // AMD: this came comment out -- looks closer to right answer

+			//assign Y = (partial[n-1] << (n-1)*`DSP_B_MAXWIDTH) + partial_sum[n-2];

+                        // was (looks broken)

+			//assign Y = {

+			//	partial[n-1][A_WIDTH+`DSP_B_MAXWIDTH-1:`DSP_B_MAXWIDTH],

+			//	partial[n-1][`DSP_B_MAXWIDTH-1:0] + partial_sum[n-2][A_WIDTH+((n-1)*`DSP_B_MAXWIDTH)-1:A_WIDTH+((n-2)*`DSP_B_MAXWIDTH)],

+			//	partial_sum[n-2][A_WIDTH+((n-2)*`DSP_B_MAXWIDTH):0]

+                       assign Y = {

+				partial[n-1][A_WIDTH+B_WIDTH-(n-1)*`DSP_B_MAXWIDTH-1:0]

+				+ partial_sum[n-2][Y_WIDTH-1:((n-1)*`DSP_B_MAXWIDTH)],

+				partial_sum[n-2][((n-1)*`DSP_B_MAXWIDTH)-1:0]

+			};

+		end

+		else begin 

+			wire [A_WIDTH+B_WIDTH-1:0] out;

+			wire [(`DSP_A_MAXWIDTH+`DSP_B_MAXWIDTH)-(A_WIDTH+B_WIDTH)-1:0] dummy;

+			wire Asign, Bsign;

+			assign Asign = (A_SIGNED ? A[A_WIDTH-1] : 1'b0);

+			assign Bsign = (B_SIGNED ? B[B_WIDTH-1] : 1'b0);

+			`DSP_NAME _TECHMAP_REPLACE_ (

+				.A({ {{`DSP_A_MAXWIDTH-A_WIDTH}{Asign}}, A }),

+				.B({ {{`DSP_B_MAXWIDTH-B_WIDTH}{Bsign}}, B }),

+				.OUT({dummy, out})

+			);

+			if (Y_WIDTH < A_WIDTH+B_WIDTH)

+				assign Y = out[Y_WIDTH-1:0];

+			else begin

+				wire Ysign = (A_SIGNED || B_SIGNED ? out[A_WIDTH+BWIDTH-1] : 1'b0);

+				assign Y = { {{Y_WIDTH-(A_WIDTH+B_WIDTH)}{Ysign}}, out[A_WIDTH+B_WIDTH-1:0] };

+			end

+		end

+	endgenerate

+endmodule

+

+