赞
踩
CPU是计算机的核心,因为它是计算机指令的处理单元。
计算机体系结构包含两个方面,一个方面是指令集,一个方面是硬件实现。指令集是计算机被定义拥有的执行指令,计算机通过支持指令集的运行,来完成计算工作并为程序员编程服务。硬件实现则是具体的硬件去实现指令集,这个硬件实现的核心就是CPU的设计。
这里写的CPU的设计是32位机器的CPU,指令和数据均为32位。支持指令为简化mips指令集。
CPU的设计包含数据通路的设计和控制器的设计。数据通路是执行指令必须的硬件(ALU、IM、DM、GRF等),控制器则是根据指令产生相应控制信号,来控制相应硬件以支持多条指令。
数据通路设计
CPU的功能是支持指令集,因此硬件设计是为了执行指令。
设计CPU的结构的方法:先选择一条需要经过最多硬件的指令,来为它构建数据通路。再依据其他指令在已有数据通路上添加硬件或线路,直到数据通路支持所有指令。
控制器设计
在已有的数据通路基础上,针对每一条指令,列出其所需要的控制信号,每一组控制信号对应一种指令的全部执行。将指令相应字段和部分计算结果作为控制器的输入,控制信号作为输出,依据上述映射关系(真值表)设计控制器。
CPU架构的设计是没有很多约束的,基本要求就是能够支持指令集,基于不同的考量可以有不同的设计。举例来说:对于beq指令是否跳转的判断,可以借用ALU的减法计算,也可以直接增设CMP比较器得出,两种方式都可以,因为功能正确。为了提高吞吐量,或者为了节省成本,会选择一些特别的设计,这一点在流水线CPU 的设计上可以明显地看出。
CPU具体设计的方法是我下面进行的几步:列出所需指令,写出功能模块,连接模块,构造控制器,全部连接起来。这些表格对最终代码实现十分重要,因为代码量较大,先从表格检查起,再依据表格写码可以减少bug。
列出支持指令并将其分类:
str | ld | cal_r | cal_i | lui | b_type | j | jr | jal | jalr | shamt |
---|---|---|---|---|---|---|---|---|---|---|
sw | lw | addu | ori | beq | sll | |||||
subu | slti | sra | ||||||||
slt | addiu | srl | ||||||||
sllv | ||||||||||
srav | ||||||||||
srlv |
先按照lw指令列出所需功能模块(lw经过模块最多),再依次检查现有模块是否支持其余指令,若不能支持,则添加相应模块。
module | input | output | 功能描述 |
---|---|---|---|
PC | D(端口名,下同) | Q | 指令计数器 |
ADD4 | PC | PC4 | 加4运算 |
IM | IA | IR | 指令存储器 |
RF | A1 | RD1 | 寄存器堆的读出部分 |
A2 | RD2 | ||
EXT | I16 | EXTD | 选择输出SIMM,LIMM,UIMM |
EXTOP | |||
CMP | D1 | RES | 比较两个数据的大小,以决定是否分支 |
D2 | |||
NPC | PC4 | NEXTPC | 计算BPC,JPC |
I26 | |||
NPCOP | |||
ALU | A | AO | 执行不同计算 |
B | |||
SHAMT | |||
ALUOP | |||
DM | DA | RD | 数据存储器,支持写入字和读出字 |
WD | |||
WM | |||
RM | |||
RF | A3 | 寄存器堆的写入部分 | |
WD3 | |||
WR |
注:上表的SIMM表示将16位立即数作符号扩展成32位;UIMM则是无符号扩展;LIMM则是专门为lui指令设计的将低16位移至高16位,并扩展至32位。BPC则是将指令地址(PC)按照分支指令地址变换方式计算分支地址,JPC则是将指令地址(PC)按照跳转指令地址变换方式计算跳转地址。
在上一步列出的所有功能模块基础上,考虑如何连接这些模块。
需要分别列出所有指令的数据通路(即在功能模块之间连线),并将其整合起来,得到最终的数据通路(完成功能模块的连接)。
module | input | LD | STR | CAL_R | CAL_I | LUI | B_TYPE | J | JR | JAL | JALR | SHAMT |
---|---|---|---|---|---|---|---|---|---|---|---|---|
PC | ||||||||||||
ADD4 | PC | Q | Q | Q | Q | Q | Q | Q | Q | Q | Q | Q |
IM | IA | Q | Q | Q | Q | Q | Q | Q | Q | Q | Q | Q |
PC | D | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 | PC4 |
RF | A1 | IR[RS] | IR[RS] | IR[RS] | IR[RS] | IR[RS] | IR[RS] | IR[RS] | ||||
A2 | IR[RT] | IR[RT] | IR[RT] | |||||||||
EXT | I16 | IR[I16] | IR[I16] | IR[I16] | IR[I16] | |||||||
CMP | D1 | RD1 | ||||||||||
D2 | RD2 | |||||||||||
NPC | PC4 | PC4 | PC4 | PC4 | ||||||||
I26 | IR[I16] | IR[I26] | IR[I26] | |||||||||
PC | D | NEXTPC | NEXTPC | RD1 | NEXTPC | RD1 | ||||||
ALU | A | RD1 | RD1 | RD1 | RD1 | |||||||
B | EXTD | EXTD | RD2 | EXTD | RD2 | |||||||
SHAMT | IR[SH] | |||||||||||
DM | DA | AO | AO | |||||||||
WD | RD2 | |||||||||||
RF | A3 | IR[RT] | IR[RD] | IR[RD] | IR[RT] | $31 | IR[RD] | IR[RD] | ||||
WD3 | RD | AO | AO | EXTD | PC4 | PC4 | AO |
注:IR[RS]是说,IR端口数据(指令)的RS字段。
INPUT | INPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | OUTPUT | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
INSTRUCTION | OP | FUNCT | PCSEL | EXTOP | NPCOP | ALUCTRL | ALUOP | BSEL | WM | RM | A3SEL | WD3SEL | WR |
LW | 100011 | X | 0 | 0 | X | 0 | 0(+) | 0 | 0(不写) | 1(读字) | 0 | 0 | 1 |
SW | 101011 | X | 0 | 0 | X | 0 | 0(+) | 0 | 1(写字) | 0 | X | X | 0 |
ADD | 000000 | 100000 | 0 | X | X | 15 | FUNCT | 1 | 0 | 0 | 1 | 1 | 1 |
ADDU | 000000 | 100001 | 0 | X | X | 15 | FUNCT | 1 | 0 | 0 | 1 | 1 | 1 |
SUB | 000000 | 100010 | 0 | X | X | 15 | FUNCT | 1 | 0 | 0 | 1 | 1 | 1 |
SUBU | 000000 | 100011 | 0 | X | X | 15 | FUNCT | 1 | 0 | 0 | 1 | 1 | 1 |
SHAMT | 000000 | FUNCT | 0 | X | X | 15 | FUNCT | 1 | 0 | 0 | 1 | 1 | 1 |
ORI | 001101 | X | 0 | 2 | X | 2 | 2 | 0 | 0 | 0 | 0 | 1 | 1 |
SLTI | 001010 | X | 0 | 0 | X | 3 | 3 | 0 | 0 | 0 | 0 | 1 | 1 |
ADDIU | 001001 | X | 0 | 0 | X | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
LUI | 001111 | X | 0 | 1 | X | X | X | X | 0 | 0 | 0 | 2 | 1 |
BEQ | 000100 | X | RES | X | 0 | X | X | X | 0 | 0 | X | X | 0 |
J | 000010 | X | 1 | X | 1 | X | X | X | 0 | 0 | X | X | 0 |
JR | 000000 | 001000 | 2 | X | X | X | X | X | 0 | 0 | X | X | 0 |
JAL | 000011 | X | 1 | X | 1 | X | X | X | 0 | 0 | 2 | 3 | 1 |
JALR | 000000 | 001001 | 2 | X | X | X | X | X | 0 | 0 | 1 | 3 | 1 |
NOP | 000000 | 000000 | 0 | X | X | X | X | X | 0 | 0 | X | X | 0 |
控制信号说明
PCSEL:用于选择下一个指令的地址是PC4,还是NPC,或者是[RS]寄存器里的地址。NPC是分支地址(BPC)或者跳转地址(JPC),简单来说就是非PC4的下一条指令可能地址。
EXTOP:EXT模块内部有多个输出,选择适当扩展立即数作为最终输出的信号。
NPCOP:选择是BPC还是JPC。(其实这里可以把PC4一起作为NPC模块的输入,直接选出下一条指令地址)
ALUCTRL&ALUOP:这里ALU 的控制信号是二次译码,ALUCTRL用来区分是否是R型计算指令,若是,则通过FUNCT字段进行二次译码决定ALU的输出结果(之所以二次译码只因为R型指令的OP字段全为0);若不是,则可通过指令的OP字段决定ALU的输出结果。
BSEL:ALU的第二个操作数的选择信号,对于addu指令,B端口输入是寄存器堆的RT输出,对于addi指令,B端口输入是立即数的符号扩展结果,B端口的输入来源不同,因此需要信号选择。
WM:数据存储器的写入控制信号。
RM:数据存储器的读出控制信号。
A3SEL:寄存器堆的写入寄存器编号的选择信号,因为addu指令的写入寄存器的编号由指令RD字段给出,lw指令和jalr指令的写入寄存器编号由指令RT字段给出,而jr指令的写入寄存器固定是$ra(31号寄存器),该端口数据的来源不同,需要信号来选择。
WD3SEL:寄存器堆写入数据的选择信号,因为写入的数据可以是DM中读出的数据(lw),也可以是ALU计算得到的数据(addu),也可以是经移位扩展得到的数据(lui),还可以是指令的返回地址(jal,jalr),需要信号来选择。
WR:寄存器堆的写入控制信号。
再次说明
控制信号的设计的基本要求是能满足所有指令在同一CPU 上正确执行,因此十分依靠CPU的硬件结构,不同的结构就有其特别的控制信号设计。控制信号的设计还需要尽可能的简洁,否则控制电路会更加复杂,上面ALU的控制信号所采用的二次译码就是简化的控制信号的电路设计。
Verilog语言编写的单周期CPU
module MFPC( //PC的多选器 input [31:0] pc4, input [31:0] nextpc, input [31:0] rd1, input [1:0] pcsel, output [31:0] d ); assign d = (pcsel == 2'd0)? pc4: (pcsel == 2'd1)? nextpc: (pcsel == 2'd2)? rd1: 32'h00003000; endmodule
module PC( //PC input clk, input reset, input enable, input [31:0] d, output [31:0] q ); reg [31:0] pc; initial begin pc = 32'h00003000; end always@(posedge clk) begin if(enable) begin if(reset) begin pc <= 32'h00003000; end else begin pc <= d; end end end assign q = pc; endmodule
module ADD4( //自增4模块 input [31:0] pc, output [31:0] pc4 ); assign pc4 = pc + 4; endmodule
module IM( //指令存储器模块 input [31:0] ia, input reset, input clk, output [31:0] ir ); parameter [31:0] num_instr = 1024; reg [31:0] im [num_instr-1:0]; integer i; initial begin $readmemh("code.txt",im); end assign ir = im[ia[11:2]]; endmodule
module MFA3( //A3端口数据选择器 input [4:0] ir_rt, input [4:0] ir_rd, input [1:0] a3sel, output [4:0] a3 ); assign a3 = (a3sel == 0)? ir_rt: (a3sel == 1)? ir_rd: (a3sel == 2)? 5'd31: 5'd0; endmodule
module MFWD3( //WD3端口数据选择器 input [31:0] rd, input [31:0] ao, input [31:0] extd, input [31:0] pc4, input [2:0] wd3sel, output [31:0] wd3 ); assign wd3 = (wd3sel == 0)? rd: (wd3sel == 1)? ao: (wd3sel == 2)? extd: (wd3sel == 3)? pc4: 0; endmodule
module EXT( //扩展模块 input [15:0] i16, input [1:0] extop, output [31:0] extd ); assign extd = (extop == 0)? {{16{i16[15]}},i16}: (extop == 1)? {i16,16'b0}: (extop == 2)? {16'b0,i16}: 0; endmodule
module CMP( //比较器 input [31:0] d1, input [31:0] d2, output res ); assign res = (d1 == d2); endmodule
module NPC( //下一条指令生成模块 input [31:0] pc4, input [25:0] i26, input npcop, output [31:0] nextpc ); assign nextpc = (npcop == 0)? $signed(pc4) + $signed({{14{i26[15]}},i26[15:0],2'b0}): {pc4[31:28],i26,2'b0}; endmodule
module ALU( //ALU模块 input [31:0] a, input [31:0] b, input [4:0] shamt, input [3:0] aluop, output [31:0] ao ); assign ao = (aluop == 0)? $signed(a) + $signed(b): (aluop == 1)? $signed(a) - $signed(b): (aluop == 2)? a | b: (aluop == 3)? ($signed(a) < $signed(b)): (aluop == 4)? (b << a[4:0]): (aluop == 5)? $signed($signed(b) >>> a[4:0]): (aluop == 6)? (b >> a[4:0]): (aluop == 7)? (b << shamt): (aluop == 8)? $signed($signed(b) >>> shamt): (aluop == 9)? (b >> shamt): 32'b0; endmodule
module MFB( //B端口选择器 input [31:0] extd, input [31:0] rd2, input bsel, output [31:0] b ); assign b = (bsel == 0)? extd: rd2; endmodule
module RF( //寄存器堆模块 input [4:0] a1, input [4:0] a2, input [4:0] a3, input [31:0] wd3, output [31:0] rd1, output [31:0] rd2, input reset, input clk, input wr ); reg [31:0] rf [31:1]; integer i; initial begin for(i = 1 ;i < 32; i = i + 1) begin rf[i] = 32'b0; end end always@(posedge clk) begin if(reset) begin for(i = 1;i < 32;i = i + 1) begin rf[i] = 32'b0; end end else if(wr && a3 != 5'b0) begin rf[a3] <= wd3; end end assign rd1 = (a1 == 5'd0)? 32'b0: rf[a1]; assign rd2 = (a2 == 5'd0)? 32'b0: rf[a2]; endmodule
module DM( //数据存储器模块 input [31:0] da, input [31:0] wd, input [1:0] wm, input [2:0] rm, input clk, input reset, output [31:0] rd ); parameter [31:0] num_data = 4096; // 1024words reg [7:0] dm [num_data-1:0]; integer i; wire [7:0] temp3,temp2,temp1,temp0; initial begin for(i = 0; i < num_data; i = i + 1) dm[i] = 8'b0; end always@(posedge clk) begin if(reset) begin for(i = 0; i < num_data; i = i + 1) dm[i] = 8'b0; end else if(wm != 0) begin case(wm) 1: begin //sw dm[da[11:0]] <= wd[7:0]; dm[da[11:0] + 12'd1] <= wd[15:8]; dm[da[11:0] + 12'd2] <= wd[23:16]; dm[da[11:0] + 12'd3] <= wd[31:24]; end 2: begin //sh dm[da[11:0]] <= wd[7:0]; dm[da[11:0] + 12'd1] <= wd[15:8]; end 3: begin //sb dm[da[11:0]] <= wd[7:0]; end endcase end end assign temp3 = dm[da[11:0] + 12'd3]; assign temp2 = dm[da[11:0] + 12'd2]; assign temp1 = dm[da[11:0] + 12'd1]; assign temp0 = dm[da[11:0]]; assign rd = (rm == 1)? {temp3,temp2,temp1,temp0}: //lw (rm == 2)? {16'b0,temp1,temp0}: //lhu (rm == 3)? {{16{temp1[7]}},temp1,temp0}: //lh (rm == 4)? {24'b0,temp0}: //lbu (rm == 5)? {{24{temp0[7]}},temp0}: //lb 0; endmodule
module MAINDECODER( //主译码器模块 input [5:0] op, input [5:0] funct, input [4:0] ir_rd, input res, output reg [1:0] pcsel, output reg [1:0] extop, output reg npcop, output reg [3:0] aluctrl, output reg bsel, output reg [1:0] wm, output reg [2:0] rm, output reg [1:0] a3sel, output reg [2:0] wd3sel, output reg wr ); always@(*) begin case(op) 6'b0: begin if(funct == 6'b001000) begin //jr pcsel = 2'd2; wm = 2'd0; rm = 3'd0; wr = 1'd0; end else if(funct == 6'b001001) begin //jalr pcsel = 2'd2; wm = 2'd0; rm = 3'd0; a3sel = 2'd1; wd3sel = 3'd3; wr = 1'd1; end else if(funct == 6'b0 && ir_rd == 5'b0) begin //nop pcsel = 2'd0; wm = 2'd0; rm = 3'd0; wr = 1'd0; end else begin //cal_r,shamt pcsel = 2'd0; aluctrl = 4'd15; bsel = 1'd1; wm = 2'd0; rm = 3'd0; a3sel = 2'd1; wd3sel = 3'd1; wr = 1'd1; end end 6'b100011: begin //lw pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd1; a3sel = 2'd0; wd3sel = 3'd0; wr = 1'd1; end 6'b100101: begin //lhu pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd2; a3sel = 2'd0; wd3sel = 3'd0; wr = 1'd1; end 6'b100001: begin //lh pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd3; a3sel = 2'd0; wd3sel = 3'd0; wr = 1'd1; end 6'b100100: begin //lbu pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd4; a3sel = 2'd0; wd3sel = 3'd0; wr = 1'd1; end 6'b100000: begin //lb pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd5; a3sel = 2'd0; wd3sel = 3'd0; wr = 1'd1; end 6'b101011: begin //sw pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd1; rm = 3'd0; wr = 1'd0; end 6'b101001: begin //sh pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd2; rm = 3'd0; wr = 1'd0; end 6'b101000: begin //sb pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd3; rm = 3'd0; wr = 1'd0; end 6'b001111: begin //lui pcsel = 2'd0; extop = 2'd1; wm = 2'd0; rm = 3'd0; a3sel = 2'd0; wd3sel = 3'd2; wr = 1'd1; end 6'b000100: begin //beq pcsel = (res == 0)? 2'd0: 2'd1; npcop = 1'd0; wm = 2'd0; rm = 3'd0; wr = 1'd0; end 6'b000010: begin //j pcsel = 2'd1; npcop = 1'd1; wm = 2'd0; rm = 3'd0; wr = 1'd0; end 6'b000011: begin //jal pcsel = 2'd1; npcop = 1'd1; wm = 2'd0; rm = 3'd0; a3sel = 2'd2; wd3sel = 3'd3; wr = 1'd1; end 6'b001101: begin //ori pcsel = 2'd0; extop = 2'd2; aluctrl = 4'd2; bsel = 1'd0; wm = 2'd0; rm = 3'd0; a3sel = 2'd0; wd3sel = 3'd1; wr = 1'd1; end 6'b001010: begin //slti pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd3; bsel = 1'd0; wm = 2'd0; rm = 3'd0; a3sel = 2'd0; wd3sel = 3'd1; wr = 1'd1; end 6'b001001: begin //addiu pcsel = 2'd0; extop = 2'd0; aluctrl = 4'd0; bsel = 1'd0; wm = 2'd0; rm = 3'd0; a3sel = 2'd0; wd3sel = 3'd1; wr = 1'd1; end endcase end endmodule
module ALUDECODER( //ALU译码器模块 input [3:0] aluctrl, input [5:0] funct, output reg [3:0] aluop ); initial aluop = 0; always@(*) begin if(aluctrl != 4'd15) begin aluop = aluctrl; end else begin case(funct) 6'b100001: aluop = 0;//addu 6'b100011: aluop = 1;//subu 6'b101010: aluop = 3;//slt 6'b000100: aluop = 4;//sllv 6'b000111: aluop = 5;//srav 6'b000110: aluop = 6;//srlv 6'b000000: aluop = 7;//sll 6'b000011: aluop = 8;//sra 6'b000010: aluop = 9;//srl endcase end end endmodule
module DECODER( //译码器模块 input [5:0] op, input [5:0] funct, input [4:0] ir_rd, input res, output [1:0] pcsel, output [1:0] extop, output npcop, output [3:0] aluop, output bsel, output [1:0] wm, output [2:0] rm, output [1:0] a3sel, output [2:0] wd3sel, output wr ); wire [3:0] aluctrl; MAINDECODER maindecoder ( .op(op), .funct(funct), .res(res), .ir_rd(ir_rd), .pcsel(pcsel), .extop(extop), .npcop(npcop), .aluctrl(aluctrl), .bsel(bsel), .wm(wm), .rm(rm), .a3sel(a3sel), .wd3sel(wd3sel), .wr(wr) ); ALUDECODER aludecoder ( .aluctrl(aluctrl), .funct(funct), .aluop(aluop) ); endmodule
`define rs 25:21 `define rt 20:16 `define rd 15:11 `define op 31:26 `define i16 15:0 `define sh 10:6 `define i26 25:0 `define funct 5:0 module mips( //CPU模块 input clk, input reset ); parameter enable = 1'd1; wire [31:0] d,q,pc4,ir,wd3,rd1,rd2,extd,nextpc,ao,b,rd; wire [4:0] a3; wire res; wire [1:0] pcsel; wire [1:0] extop; wire npcop; wire [3:0] aluop; wire bsel; wire [1:0] wm; wire [2:0] rm; wire [1:0] a3sel; wire [2:0] wd3sel; wire wr; MFPC mfpc ( .pc4(pc4), .nextpc(nextpc), .rd1(rd1), .pcsel(pcsel), .d(d) ); PC pc ( .clk(clk), .reset(reset), .enable(enable), .d(d), .q(q) ); ADD4 add4 ( .pc(q), .pc4(pc4) ); IM im ( .ia(q), .reset(reset), .clk(clk), .ir(ir) ); MFA3 mfa3 ( .ir_rt(ir[`rt]), .ir_rd(ir[`rd]), .a3sel(a3sel), .a3(a3) ); MFWD3 mfwd3 ( .rd(rd), .ao(ao), .extd(extd), .pc4(pc4), .wd3sel(wd3sel), .wd3(wd3) ); EXT ext ( .i16(ir[`i16]), .extop(extop), .extd(extd) ); CMP cmp ( .d1(rd1), .d2(rd2), .res(res) ); NPC npc ( .pc4(pc4), .i26(ir[`i26]), .npcop(npcop), .nextpc(nextpc) ); ALU alu ( .a(rd1), .b(b), .shamt(ir[`sh]), .aluop(aluop), .ao(ao) ); MFB mfb ( .extd(extd), .rd2(rd2), .bsel(bsel), .b(b) ); RF rf ( .a1(ir[`rs]), .a2(ir[`rt]), .a3(a3), .wd3(wd3), .rd1(rd1), .rd2(rd2), .reset(reset), .clk(clk), .wr(wr) ); DM dm ( .da(ao), .wd(rd2), .wm(wm), .rm(rm), .clk(clk), .reset(reset), .rd(rd) ); DECODER decoder ( .op(ir[`op]), .funct(ir[`funct]), .ir_rd(ir[`rd]), .res(res), .pcsel(pcsel), .extop(extop), .npcop(npcop), .aluop(aluop), .bsel(bsel), .wm(wm), .rm(rm), .a3sel(a3sel), .wd3sel(wd3sel), .wr(wr) ); always@(posedge clk) begin if(reset == 0) begin if(wr == 1'd1) $display("@%h: $%d <= %h", q, a3, wd3); if(wm != 2'd0) $display("@%h: *%h <= %h", q, ao, rd2); end end endmodule
注:此代码仅供参考,仅支持所声明指令。
CPU实现算是大学第一次遇到的有一定代码量的编程,它给我带来的更多的是面对复杂编程的经验:越复杂,越要先抽象再具体。后面的流水线CPU会更加体现这一点,表格(设计)的重要性对CPU的编程很重要。
Copyright © 2003-2013 www.wpsshop.cn 版权所有,并保留所有权利。