FPGA内部资源（一）DSP48E1
一、 实验过程中发现的问题
使用ISE进行项目的实现时出现以下错误。
意思很简单，就是使用DSP48E1的数量超出限制，因为没有接触过DSP48E1，所以尝试了很多错误的方法后，我找到项目下的.mrp文件，里面有一行显示 Number of DSP48E1s: 496 out of 480 103%，也就是说我使用的 XC6V1x130t 型号FPGA只有480个DSP48E1，但是我使用了496个。
我的项目中用到了4个fir滤波器，都是160阶的，而且其中两个系数对称，另外两个系数不对称，我在代码中尝试添加其中的某个滤波器，发现系数对称的滤波器会占用80个DSP48E1，系数不对称的fir滤波器会占用160个DSP48E1，并且结合查阅到的资料，DSP48E1主要是用来做乘法器使用的，这就非常合理了。
于是，我适当减少了fir滤波器的系数个数，最终可以成功实现。
二、学习DSP48E1
图中带*号的信号是级联专用信号，只在级联时使用，用户逻辑访问不了。我们只需要关注A、B、C、D、P、INMODE、OPMODE、ALUMODE这几个信号的使用即可。
INMODE
上图主要反映了INMODE与打拍寄存器的关系，可以用下面两个表概括，INMODE一般配置为7'b0000101，做乘法器使用。
OPMODE
X、Y、Z输出与OPMODE的关系如下表所示。
ALUMODE
最终P端口输出与ALUMODE的关系如下所示，其中的CIN就是48位输入信号C。
例程
`timescale 1ns / 1ps
//////////////////////////////////////////////////////////////////////////////////
// Company:
// Engineer:
//
// Create Date: 2024/06/19 16:58:05
// Design Name:
// Module Name: tb_DSP48E1
// Project Name:
// Target Devices:
// Tool Versions:
// Description:
//
// Dependencies:
//
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
//
//////////////////////////////////////////////////////////////////////////////////
module tb_DSP48E1();
reg R_clk;
reg R_reset;
reg[29:0] R_A;
reg[24:0] R_D;
reg[47:0] R_C;
reg[17:0] R_B;
wire[47:0] W_P;
wire[47:0] W_PCIN;
reg[4:0] INMODE;
reg[6:0] OPMODE;
reg[3:0] ALUMODE;
reg[3:0] R_num;
initial begin
R_clk = 1'b0;
R_reset = 1'b1;
R_A = 30'd100;
R_B = 18'd2;
R_C = 48'd2000;
R_D = 25'd200;
R_num = 4'd0;
OPMODE = 7'b0000101; // X-M Y-M Z-0
INMODE = 5'b00101; // (D+A1) * B2 这三个输出刚好对齐，第三拍得到运算结果
ALUMODE = 4'b0000; // X+Y+Z+C
#100 R_reset = 1'b0;
end
always#5 R_clk = ~R_clk;
always @(posedge R_clk) begin
if(R_reset)begin
R_A <= 30'd100;
R_B <= 18'd2;
R_C <= 48'd2000;
R_D <= 25'd200;
R_num <= 4'd0;
end
else if(R_num < 9)begin
R_num <= R_num + 1'b1;
R_A <= R_A + 1'b1;
R_B <= R_B + 1'b1;
R_C <= R_C + 1'b1;
R_D <= R_D + 1'b1;
end
else begin
OPMODE <= 7'b0110101; // X-M Y-M Z-C
R_A <= R_A + 1'b1;
R_B <= R_B + 1'b1;
R_C <= R_C + 1'b1;
R_D <= R_D + 1'b1;
end
end
DSP48E1 #(
// Feature Control Attributes: Data Path Selection
.A_INPUT("DIRECT"), // Selects A input source, "DIRECT" (A port) or "CASCADE" (ACIN port)
.B_INPUT("DIRECT"), // Selects B input source, "DIRECT" (B port) or "CASCADE" (BCIN port)
.USE_DPORT("TRUE"), // Select D port usage (TRUE or FALSE)
.USE_MULT("MULTIPLY"), // Select multiplier usage ("MULTIPLY", "DYNAMIC", or "NONE")
.USE_SIMD("ONE48"), // SIMD selection ("ONE48", "TWO24", "FOUR12")
// Pattern Detector Attributes: Pattern Detection Configuration
.AUTORESET_PATDET("NO_RESET"), // "NO_RESET", "RESET_MATCH", "RESET_NOT_MATCH"
.MASK(48'h3fffffffffff), // 48-bit mask value for pattern detect (1=ignore)
.PATTERN(48'h000000000000), // 48-bit pattern match for pattern detect
.SEL_MASK("MASK"), // "C", "MASK", "ROUNDING_MODE1", "ROUNDING_MODE2"
.SEL_PATTERN("PATTERN"), // Select pattern value ("PATTERN" or "C")
.USE_PATTERN_DETECT("NO_PATDET"), // Enable pattern detect ("PATDET" or "NO_PATDET")
// Register Control Attributes: Pipeline Register Configuration
.ACASCREG(1), // Number of pipeline stages between A/ACIN and ACOUT (0, 1 or 2)
.ADREG(1), // Number of pipeline stages for pre-adder (0 or 1)
.ALUMODEREG(1), // Number of pipeline stages for ALUMODE (0 or 1)
.AREG(1), // Number of pipeline stages for A (0, 1 or 2)
.BCASCREG(1), // Number of pipeline stages between B/BCIN and BCOUT (0, 1 or 2)
.BREG(2), // Number of pipeline stages for B (0, 1 or 2) 这里要用到B2，所以要置为2
.CARRYINREG(1), // Number of pipeline stages for CARRYIN (0 or 1)
.CARRYINSELREG(1), // Number of pipeline stages for CARRYINSEL (0 or 1)
.CREG(1), // Number of pipeline stages for C (0 or 1)
.DREG(1), // Number of pipeline stages for D (0 or 1)
.INMODEREG(1), // Number of pipeline stages for INMODE (0 or 1)
.MREG(1), // Number of multiplier pipeline stages (0 or 1)
.OPMODEREG(1), // Number of pipeline stages for OPMODE (0 or 1)
.PREG(1) // Number of pipeline stages for P (0 or 1)
)
DSP48E1_inst (
// Cascade: 30-bit (each) output: Cascade Ports
.ACOUT(), // 30-bit output: A port cascade output
.BCOUT(), // 18-bit output: B port cascade output
.CARRYCASCOUT(), // 1-bit output: Cascade carry output
.MULTSIGNOUT(), // 1-bit output: Multiplier sign cascade output
.PCOUT(), // 48-bit output: Cascade output
// Control: 1-bit (each) output: Control Inputs/Status Bits
.OVERFLOW(), // 1-bit output: Overflow in add/acc output
.PATTERNBDETECT(), // 1-bit output: Pattern bar detect output
.PATTERNDETECT(), // 1-bit output: Pattern detect output
.UNDERFLOW(), // 1-bit output: Underflow in add/acc output
// Data: 4-bit (each) output: Data Ports
.CARRYOUT(), // 4-bit output: Carry output
.P(W_P), // 48-bit output: Primary data output
// Cascade: 30-bit (each) input: Cascade Ports
.ACIN(1'b0), // 30-bit input: A cascade data input
.BCIN(1'b0), // 18-bit input: B cascade input
.CARRYCASCIN(1'b1), // 1-bit input: Cascade carry input
.MULTSIGNIN(1'b1), // 1-bit input: Multiplier sign input
.PCIN(W_PCIN), // 48-bit input: P cascade input
// Control: 4-bit (each) input: Control Inputs/Status Bits
.ALUMODE(ALUMODE), // 4-bit input: ALU control input
.CARRYINSEL(3'd0), // 3-bit input: Carry select input
.CLK(R_clk), // 1-bit input: Clock input
.INMODE(INMODE), // 5-bit input: INMODE control input
.OPMODE(OPMODE), // 7-bit input: Operation mode input
// Data: 30-bit (each) input: Data Ports
.A(R_A), // 30-bit input: A data input
.B(R_B), // 18-bit input: B data input
.C(R_C), // 48-bit input: C data input
.CARRYIN(1'b0), // 1-bit input: Carry input signal
.D(R_D), // 25-bit input: D data input
// Reset/Clock Enable: 1-bit (each) input: Reset/Clock Enable Inputs
.CEA1(1'b1), // 1-bit input: Clock enable input for 1st stage AREG
.CEA2(1'b1), // 1-bit input: Clock enable input for 2nd stage AREG
.CEAD(1'b1), // 1-bit input: Clock enable input for ADREG
.CEALUMODE(1'b1), // 1-bit input: Clock enable input for ALUMODE
.CEB1(1'b1), // 1-bit input: Clock enable input for 1st stage BREG
.CEB2(1'b1), // 1-bit input: Clock enable input for 2nd stage BREG
.CEC(1'b1), // 1-bit input: Clock enable input for CREG
.CECARRYIN(1'b1), // 1-bit input: Clock enable input for CARRYINREG
.CECTRL(1'b1), // 1-bit input: Clock enable input for OPMODEREG and CARRYINSELREG
.CED(1'b1), // 1-bit input: Clock enable input for DREG
.CEINMODE(1'b1), // 1-bit input: Clock enable input for INMODEREG
.CEM(1'b1), // 1-bit input: Clock enable input for MREG
.CEP(1'b1), // 1-bit input: Clock enable input for PREG
.RSTA(R_reset), // 1-bit input: Reset input for AREG
.RSTALLCARRYIN(R_reset), // 1-bit input: Reset input for CARRYINREG
.RSTALUMODE(R_reset), // 1-bit input: Reset input for ALUMODEREG
.RSTB(R_reset), // 1-bit input: Reset input for BREG
.RSTC(R_reset), // 1-bit input: Reset input for CREG
.RSTCTRL(R_reset), // 1-bit input: Reset input for OPMODEREG and CARRYINSELREG
.RSTD(R_reset), // 1-bit input: Reset input for DREG and ADREG
.RSTINMODE(R_reset), // 1-bit input: Reset input for INMODEREG
.RSTM(R_reset), // 1-bit input: Reset input for MREG
.RSTP(R_reset) // 1-bit input: Reset input for PREG
);
endmodule
运行结果如下图所示。
从DSP48E1结构图也能看出，最终输出结果P会在A、B、C、D输入的第四拍输出，并且C与A、B、D相差两拍，OPMODE也会延迟一拍配置进去，所以改变OPMODE之后，结果会是5170 = （108+208）* 10 + 2010，输出结果符合预期。
按照代码中的配置，X = M，Y=M，理论上加起来应该是2M，但实际就是M，这里应该是乘法器分两部分进行乘法操作，所以最后X+Y = M。
使用DSP48E1实现fir的卷积功能
主程序如下
`timescale 1ns / 1ps
//////////////////////////////////////////////////////////////////////////////////
// Company:
// Engineer:
//
// Create Date: 2024/06/22 11:15:16
// Design Name:
// Module Name: pck_fir
// Project Name:
// Target Devices:
// Tool Versions:
// Description:
//
// Dependencies:
//
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
//
//////////////////////////////////////////////////////////////////////////////////
// 使用dsp48e1实现fir滤波器
// 先串行输入滤波器系数，假设 x = [1,2...11]
// 再串行输入信号，假设 t = [31,32...1000]
// 最后得到结果为 y = conv(x,t)
module pck_fir # (
parameter C_COE_WIDTH = 16,
parameter C_INPUT_WIDTH = 16,
parameter C_OUTPUT_WIDTH = 48,
parameter C_COE_NUM = 11
)
(
input I_clk,
input I_rst,
input I_coe_param_vd,
input[C_COE_WIDTH-1:0] I_coe_param,
input I_data_vd,
input[C_INPUT_WIDTH-1:0] I_data,
output[C_OUTPUT_WIDTH-1:0] O_data
);
localparam C_A_WIDTH = 30;
localparam C_B_WIDTH = 18;
localparam C_C_WIDTH = 48;
localparam C_D_WIDTH = 25;
localparam C_P_WIDTH = 48;
reg[C_B_WIDTH-1:0] R_coe_params[C_COE_NUM-1:0];
reg[C_A_WIDTH-1:0] R_data;
wire[47:0] W_cin;
wire[24:0] W_din;
wire[47:0] W_pin;
wire[47:0] W_pout[C_COE_NUM-1:0];
assign W_cin = 48'd0;
assign W_din = 25'd0;
assign W_pin = 48'd0;
always @(posedge I_clk or negedge I_rst) begin
if(~I_rst)begin
R_data <= 'd0;
end
else if(I_data_vd)begin
R_data[C_INPUT_WIDTH-1:0] <= I_data;
end
else
R_data <= 0;
end
genvar i;
generate
for(i = 0;i < C_COE_NUM;i = i + 1)
begin
always @(posedge I_clk or negedge I_rst) begin
if(~I_rst)begin
R_coe_params[i] <= 'd0;
end
else begin
if(I_coe_param_vd)begin
if(i == 0)
R_coe_params[i][C_COE_WIDTH-1:0] <= I_coe_param;
else
R_coe_params[i] <= R_coe_params[i-1];
end
else
R_coe_params[i] <= R_coe_params[i];
end
end
end
endgenerate
pck_dsp48e1 pck_1(
.I_clk(I_clk),
.I_rst(~I_rst),
.A(R_data),
.B(R_coe_params[0]),
.C(W_cin),
.D(W_din),
.PCIN(W_pin),
.P(),
.PCOUT(W_pout[0])
);
genvar k;
generate
for(k = 1;k < C_COE_NUM;k = k + 1)
begin:dsp_gen
pck_dsp48e1 #(
.C_OPMODE(7'b0010101)
) pck_n(
.I_clk(I_clk),
.I_rst(~I_rst),
.A(R_data),
.B(R_coe_params[k]),
.C(W_cin),
.D(W_din),
.PCIN(W_pout[k-1]),
.P(),
.PCOUT(W_pout[k])
);
end
endgenerate
assign O_data = W_pout[C_COE_NUM-1];
endmodule
由于DSP48E1参数较多，可以进行适当封装，封装后如下。
`timescale 1ns / 1ps
//////////////////////////////////////////////////////////////////////////////////
// Company:
// Engineer:
//
// Create Date: 2024/06/22 10:45:31
// Design Name:
// Module Name: pck_dsp48e1
// Project Name:
// Target Devices:
// Tool Versions:
// Description:
//
// Dependencies:
//
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
//
//////////////////////////////////////////////////////////////////////////////////
module pck_dsp48e1 # (
parameter C_ALUMODE = 4'b0000,
parameter C_INMODE = 5'b00101,
parameter C_OPMODE = 7'b0000101
)
(
input I_clk,
input I_rst,
input[29:0] A,
input[17:0] B,
input[47:0] C,
input[24:0] D,
input[47:0] PCIN,
output[47:0] P,
output[47:0] PCOUT
);
DSP48E1 #(
// Feature Control Attributes: Data Path Selection
.A_INPUT("DIRECT"), // Selects A input source, "DIRECT" (A port) or "CASCADE" (ACIN port)
.B_INPUT("DIRECT"), // Selects B input source, "DIRECT" (B port) or "CASCADE" (BCIN port)
.USE_DPORT("TRUE"), // Select D port usage (TRUE or FALSE)
.USE_MULT("MULTIPLY"), // Select multiplier usage ("MULTIPLY", "DYNAMIC", or "NONE")
.USE_SIMD("ONE48"), // SIMD selection ("ONE48", "TWO24", "FOUR12")
// Pattern Detector Attributes: Pattern Detection Configuration
.AUTORESET_PATDET("NO_RESET"), // "NO_RESET", "RESET_MATCH", "RESET_NOT_MATCH"
.MASK(48'h3fffffffffff), // 48-bit mask value for pattern detect (1=ignore)
.PATTERN(48'h000000000000), // 48-bit pattern match for pattern detect
.SEL_MASK("MASK"), // "C", "MASK", "ROUNDING_MODE1", "ROUNDING_MODE2"
.SEL_PATTERN("PATTERN"), // Select pattern value ("PATTERN" or "C")
.USE_PATTERN_DETECT("NO_PATDET"), // Enable pattern detect ("PATDET" or "NO_PATDET")
// Register Control Attributes: Pipeline Register Configuration
.ACASCREG(1), // Number of pipeline stages between A/ACIN and ACOUT (0, 1 or 2)
.ADREG(1), // Number of pipeline stages for pre-adder (0 or 1)
.ALUMODEREG(1), // Number of pipeline stages for ALUMODE (0 or 1)
.AREG(1), // Number of pipeline stages for A (0, 1 or 2)
.BCASCREG(1), // Number of pipeline stages between B/BCIN and BCOUT (0, 1 or 2)
.BREG(2), // Number of pipeline stages for B (0, 1 or 2) 这里要用到B2，所以要置为2
.CARRYINREG(1), // Number of pipeline stages for CARRYIN (0 or 1)
.CARRYINSELREG(1), // Number of pipeline stages for CARRYINSEL (0 or 1)
.CREG(1), // Number of pipeline stages for C (0 or 1)
.DREG(1), // Number of pipeline stages for D (0 or 1)
.INMODEREG(1), // Number of pipeline stages for INMODE (0 or 1)
.MREG(1), // Number of multiplier pipeline stages (0 or 1)
.OPMODEREG(1), // Number of pipeline stages for OPMODE (0 or 1)
.PREG(1) // Number of pipeline stages for P (0 or 1)
)
DSP48E1_inst (
// Cascade: 30-bit (each) output: Cascade Ports
.ACOUT(), // 30-bit output: A port cascade output
.BCOUT(), // 18-bit output: B port cascade output
.CARRYCASCOUT(), // 1-bit output: Cascade carry output
.MULTSIGNOUT(), // 1-bit output: Multiplier sign cascade output
.PCOUT(PCOUT), // 48-bit output: Cascade output
// Control: 1-bit (each) output: Control Inputs/Status Bits
.OVERFLOW(), // 1-bit output: Overflow in add/acc output
.PATTERNBDETECT(), // 1-bit output: Pattern bar detect output
.PATTERNDETECT(), // 1-bit output: Pattern detect output
.UNDERFLOW(), // 1-bit output: Underflow in add/acc output
// Data: 4-bit (each) output: Data Ports
.CARRYOUT(), // 4-bit output: Carry output
.P(P), // 48-bit output: Primary data output
// Cascade: 30-bit (each) input: Cascade Ports
.ACIN(1'b0), // 30-bit input: A cascade data input
.BCIN(1'b0), // 18-bit input: B cascade input
.CARRYCASCIN(1'b1), // 1-bit input: Cascade carry input
.MULTSIGNIN(1'b1), // 1-bit input: Multiplier sign input
.PCIN(PCIN), // 48-bit input: P cascade input
// Control: 4-bit (each) input: Control Inputs/Status Bits
.ALUMODE(C_ALUMODE), // 4-bit input: ALU control input
.CARRYINSEL(3'd0), // 3-bit input: Carry select input
.CLK(I_clk), // 1-bit input: Clock input
.INMODE(C_INMODE), // 5-bit input: INMODE control input
.OPMODE(C_OPMODE), // 7-bit input: Operation mode input
// Data: 30-bit (each) input: Data Ports
.A(A), // 30-bit input: A data input
.B(B), // 18-bit input: B data input
.C(C), // 48-bit input: C data input
.CARRYIN(1'b0), // 1-bit input: Carry input signal
.D(D), // 25-bit input: D data input
// Reset/Clock Enable: 1-bit (each) input: Reset/Clock Enable Inputs
.CEA1(1'b1), // 1-bit input: Clock enable input for 1st stage AREG
.CEA2(1'b1), // 1-bit input: Clock enable input for 2nd stage AREG
.CEAD(1'b1), // 1-bit input: Clock enable input for ADREG
.CEALUMODE(1'b1), // 1-bit input: Clock enable input for ALUMODE
.CEB1(1'b1), // 1-bit input: Clock enable input for 1st stage BREG
.CEB2(1'b1), // 1-bit input: Clock enable input for 2nd stage BREG
.CEC(1'b1), // 1-bit input: Clock enable input for CREG
.CECARRYIN(1'b1), // 1-bit input: Clock enable input for CARRYINREG
.CECTRL(1'b1), // 1-bit input: Clock enable input for OPMODEREG and CARRYINSELREG
.CED(1'b1), // 1-bit input: Clock enable input for DREG
.CEINMODE(1'b1), // 1-bit input: Clock enable input for INMODEREG
.CEM(1'b1), // 1-bit input: Clock enable input for MREG
.CEP(1'b1), // 1-bit input: Clock enable input for PREG
.RSTA(I_rst), // 1-bit input: Reset input for AREG
.RSTALLCARRYIN(I_rst), // 1-bit input: Reset input for CARRYINREG
.RSTALUMODE(I_rst), // 1-bit input: Reset input for ALUMODEREG
.RSTB(I_rst), // 1-bit input: Reset input for BREG
.RSTC(I_rst), // 1-bit input: Reset input for CREG
.RSTCTRL(I_rst), // 1-bit input: Reset input for OPMODEREG and CARRYINSELREG
.RSTD(I_rst), // 1-bit input: Reset input for DREG and ADREG
.RSTINMODE(I_rst), // 1-bit input: Reset input for INMODEREG
.RSTM(I_rst), // 1-bit input: Reset input for MREG
.RSTP(I_rst) // 1-bit input: Reset input for PREG
);
endmodule
测试代码如下。
`timescale 1ns / 1ps
//////////////////////////////////////////////////////////////////////////////////
// Company:
// Engineer:
//
// Create Date: 2024/06/22 15:35:29
// Design Name:
// Module Name: tb_pck_fir
// Project Name:
// Target Devices:
// Tool Versions:
// Description:
//
// Dependencies:
//
// Revision:
// Revision 0.01 - File Created
// Additional Comments:
//
//////////////////////////////////////////////////////////////////////////////////
module tb_pck_fir();
reg R_clk;
reg R_data_vd;
reg R_coe_data_vd;
wire[15:0] W_data;
wire[15:0] W_coe_data;
reg R_rst;
reg[15:0] R_cnt;
wire[47:0] W_res;
initial begin
R_clk = 1'b0;
R_rst = 1'b0;
R_cnt = 5'd0;
R_data_vd = 1'b0;
R_coe_data_vd = 1'b0;
#2000 R_rst = 1'b1;
end
always#5 R_clk = ~R_clk;
always @(posedge R_clk) begin
if(~R_rst)begin
R_cnt <= 16'd0;
end
else if(R_cnt < 'd11)begin
R_cnt <= R_cnt + 1;
R_coe_data_vd <= 1'b1;
end
else if(R_cnt == 'd11)begin
R_coe_data_vd <= 1'b0;
R_cnt <= R_cnt + 1;
end
else if(R_cnt == 'd30)begin
R_cnt <= R_cnt + 1;
R_data_vd <= 1'b1;
end
else if(R_cnt > 'd1000)begin
R_data_vd <= 1'b0;
R_cnt <= R_cnt;
end
else
R_cnt <= R_cnt + 1;
end
assign W_coe_data = R_cnt;
assign W_data = R_cnt;
pck_fir pf(
.I_clk(R_clk),
.I_rst(R_rst),
.I_coe_param_vd(R_coe_data_vd),
.I_coe_param(W_coe_data),
.I_data_vd(R_data_vd),
.I_data(W_data),
.O_data(W_res)
);
endmodule
仿真结果如下