简单双口RAM有两个端口Port A和port B,其中Port A用于写数据，Port B用于读数据，读写接口可以独立时钟工作。这一点和真双口RAM是有区别的，真双口RAM的A B两个Port都可以进行读写操作。        

         RAM是FPGA中重要的数据结构，可用于数据的缓存和跨时钟域信号处理。本文就xilink 的简单双口RAM 进行读写测试。

         本文对简单双口RAM做如下仿真：先对Port A进行连续写操作128次，每次数据和地址加1,结束后开始从Port B口读对应128个地址的数据，读写时钟独立，对比读写数据是否一致。

仿真结果如下：

数据写入仿真

数据读出仿真，数据在地址变化后延迟一个时钟输出。

ram ip的设置如下

注意不要勾选 Primitives Output Regster,不然Port B数据会延迟两个周期。

简单双口RAM的接口定义如下

直接上代码 dual_ram_top.v

`timescale 1ns / 1ps
//
// Company: 
// Engineer: 
// 
// Create Date: 2024/03/22 20:18:12
// Design Name: 
// Module Name: dual_ram_top
// Project Name: 
// Target Devices: 
// Tool Versions: 
// Description: 
// 
// Dependencies: 
// 
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
// 
//


module dual_ram_top(
	input clka,
	input clkb,
	input rst_n,
	output [7:0] doutb
    );

reg [7:0]dina;
reg [12:0] addra;

reg ena;
reg wea;
reg [3:0] wr_state;
always@(posedge clka or negedge rst_n) begin
	if(!rst_n) begin
		wr_state <= 4'd0;
		
		ena <= 1'b0;
		wea <= 1'b0;
		dina <= 8'd0;
		addra <= 13'd0;
	end
	else begin
		case(wr_state)
			4'd0:begin
				wr_state <= 4'd1;
				
				wea <= 1'b1; // 写使能
			end
			
			4'd1:begin
				wr_state <= 4'd2;
				
				addra <= 13'd0;
				dina <= 8'd0;
				ena <= 1'b1;
				
			end
			
			4'd2:begin
				if(addra == 13'd127) begin
					wr_state <= 4'd3;
					
					ena <= 1'b0;
					addra <= 13'd0;
					dina <= 8'd0;
				end
				else begin
					wr_state <= wr_state;
					
					addra <= addra + 13'd1;
					dina  <= dina + 8'd1;
					ena <= 1'b1;
				end
			end
			
			4'd3:begin // write end
				wr_state <= 4'd4;
				
				wea <= 1'b0; // 读使能
			end
			
			4'd4:begin
				wr_state <= wr_state;
			end
		endcase
	end
end

wire rd_start;
assign rd_start =(wr_state==4'd4)?1:0;

reg [12:0] addrb;
reg enb;

reg [3:0] rd_state;
always@(posedge clkb or negedge rst_n) begin
	if(!rst_n) begin
		rd_state <= 4'd0;
		
		enb <= 1'b0;
		addrb <= 13'd0;
	end
	else begin
		case(rd_state) 
			4'd0:begin
				if(rd_start) begin
					rd_state <= 4'd1;
					
					enb <= 1'd1;
					addrb <= 13'd0;
				end
				else begin
					rd_state <= rd_state;
					
					enb <= 1'd0;
					addrb <= 13'd0;
				end
			end
			
			4'd1:begin
				if(addrb == 13'd127) begin
					rd_state <= 4'd2;
					
					enb <= 1'd0;
					addrb <= 13'd0;
				end
				else begin
					rd_state <= rd_state;
					
					enb <= 1'd1;
					addrb <= addrb +13'd1;	
				end
			end
			
			4'd2:begin
				rd_state <= rd_state;
				
				enb <= 1'd0;
				addrb <= 13'd0;
			end
		endcase
	end
end

	
blk_mem_gen_0 blk_mem_gen_uut0 (
  .clka(clka),    // input wire clka
  .ena(ena),      // input wire ena
  .wea(wea),      // input wire [0 : 0] wea
  .addra(addra),  // input wire [12 : 0] addra
  .dina(dina),    // input wire [7 : 0] dina
  .clkb(clkb),    // input wire clkb
  .enb(enb),      // input wire enb
  .addrb(addrb),  // input wire [12 : 0] addrb
  .doutb(doutb)  // output wire [7 : 0] doutb
);
endmodule

仿真测试代码ram_wr_top.v

`timescale 1ns / 1ps
//
// Company: 
// Engineer: 
// 
// Create Date: 2024/03/22 22:27:59
// Design Name: 
// Module Name: ram_wr_top
// Project Name: 
// Target Devices: 
// Tool Versions: 
// Description: 
// 
// Dependencies: 
// 
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
// 
//


module ram_wr_top(
    );
	reg clka;
	reg clkb;
	reg rst_n;
	wire [7:0]doutb;
	
	localparam clk_prioda = 50;
	localparam clk_priodb = 20;
	
	initial begin
		clka =1'b0;
		clkb = 1'b0;
		rst_n = 1'b0;
		#(100)
			rst_n = ~rst_n;
	end
	
	always#(clk_prioda/2) clka = ~clka;
	always#(clk_priodb/2) clkb = ~clkb;
	
	dual_ram_top dual_ram_top_uut0(
	.clka(clka),
	.clkb(clkb),
	.rst_n(rst_n),
	.doutb(doutb)
	);
endmodule

通过连续读写对比 验证了IP读写的一致性。