集成电路设计开源EDA软件yosys详解1:工具安装

yosys为一套开源的针对verilog的rtl综合框架,从本节开始将详细介绍工具的使用,并详细对源代码进行分析和讲解,首先介绍一下工具的安装和使用。yosys的git网址为https://github.com/YosysHQ/yosys。

下面将介绍如何在ubuntu环境下安装,使用的环境是ubuntu16.0。

安装

准备工作

安装Tabby CAD Suite工具,网址为:Releases · YosysHQ/oss-cad-suite-build · GitHub,根据系统下载指定的安装包:

集成电路设计开源EDA软件yosys详解1:工具安装_第1张图片

这里选择的是linux-x64,然后根据安装指南GitHub - YosysHQ/oss-cad-suite-build: Multi-platform nightly builds of open source digital design and verification tools, 设置环境变量即可,这部分很简单。

安装必要的软件包:

apt-get install build-essential clang bison flex \
	libreadline-dev gawk tcl-dev libffi-dev git \
	graphviz xdot pkg-config python3 libboost-system-dev \
	libboost-python-dev libboost-filesystem-dev zlib1g-dev

编译和安装

拷贝代码:

git clone https://github.com/YosysHQ/yosys.git

进入yosys目录,执行编译和安装指令

make
make install

正常情况下,会安装成功,当前目录下会生成yosys等可执行程序:

EDA软件使用

在当前目录执行:

./yosys

正常的话会进入yosys界面:

集成电路设计开源EDA软件yosys详解1:工具安装_第2张图片

 我们以fiedler-cooley.v(目录/tests/simple/下)文件为例来说明,verilog源代码如下:

module up3down5(clock, data_in, up, down, carry_out, borrow_out, count_out, parity_out);

input [8:0] data_in;
input clock, up, down;

output reg [8:0] count_out;
output reg carry_out, borrow_out, parity_out;

reg [9:0] cnt_up, cnt_dn;
reg [8:0] count_nxt;

always @(posedge clock)
begin
	cnt_dn = count_out - 3'b 101;
	cnt_up = count_out + 2'b 11;

	case ({up,down})
		2'b 00 : count_nxt = data_in;
		2'b 01 : count_nxt = cnt_dn;
		2'b 10 : count_nxt = cnt_up;
		2'b 11 : count_nxt = count_out;
		default : count_nxt = 9'bX;
	endcase

	parity_out  <= ^count_nxt;
	carry_out   <= up & cnt_up[9];
	borrow_out  <= down & cnt_dn[9];
	count_out   <= count_nxt;
end

endmodule

一个简单的逻辑电路代码实现,执行指令:

read -sv tests/simple/fiedler-cooley.v

读取verilog文件并用数字前端描述该电路,结果如下:

 接着执行指令:

hierarchy -top up3down5

生成RTL结构,结果如下:

集成电路设计开源EDA软件yosys详解1:工具安装_第3张图片

 接着执行指令:

write_ilang

将设计写入Yosys内部文件格式,结果如下:

4. Executing RTLIL backend.
Output filename: 
# Generated by Yosys 0.17+50 (git sha1 f698a0514, gcc 5.4.0-6ubuntu1~16.04.12 -fPIC -Os)
autoidx 7
attribute \hdlname "\\up3down5"
attribute \top 1
attribute \src "tests/simple/fiedler-cooley.v:3.1-33.10"
module \up3down5
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire $0\borrow_out[0:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire $0\carry_out[0:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire width 10 $0\cnt_dn[9:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire width 10 $0\cnt_up[9:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire width 9 $0\count_nxt[8:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire width 9 $0\count_out[8:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire $0\parity_out[0:0]
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  wire width 9 $1\count_nxt[8:0]
  attribute \src "tests/simple/fiedler-cooley.v:17.11-17.29"
  wire width 10 $add$tests/simple/fiedler-cooley.v:17$3_Y
  attribute \src "tests/simple/fiedler-cooley.v:28.17-28.31"
  wire $and$tests/simple/fiedler-cooley.v:28$5_Y
  attribute \src "tests/simple/fiedler-cooley.v:29.17-29.33"
  wire $and$tests/simple/fiedler-cooley.v:29$6_Y
  attribute \src "tests/simple/fiedler-cooley.v:27.17-27.27"
  wire $reduce_xor$tests/simple/fiedler-cooley.v:27$4_Y
  attribute \src "tests/simple/fiedler-cooley.v:16.11-16.30"
  wire width 10 $sub$tests/simple/fiedler-cooley.v:16$2_Y
  attribute \src "tests/simple/fiedler-cooley.v:9.23-9.33"
  wire output 6 \borrow_out
  attribute \src "tests/simple/fiedler-cooley.v:9.12-9.21"
  wire output 5 \carry_out
  attribute \src "tests/simple/fiedler-cooley.v:6.7-6.12"
  wire input 1 \clock
  attribute \src "tests/simple/fiedler-cooley.v:11.19-11.25"
  wire width 10 \cnt_dn
  attribute \src "tests/simple/fiedler-cooley.v:11.11-11.17"
  wire width 10 \cnt_up
  attribute \src "tests/simple/fiedler-cooley.v:12.11-12.20"
  wire width 9 \count_nxt
  attribute \src "tests/simple/fiedler-cooley.v:8.18-8.27"
  wire width 9 output 7 \count_out
  attribute \src "tests/simple/fiedler-cooley.v:5.13-5.20"
  wire width 9 input 2 \data_in
  attribute \src "tests/simple/fiedler-cooley.v:6.18-6.22"
  wire input 4 \down
  attribute \src "tests/simple/fiedler-cooley.v:9.35-9.45"
  wire output 8 \parity_out
  attribute \src "tests/simple/fiedler-cooley.v:6.14-6.16"
  wire input 3 \up
  attribute \src "tests/simple/fiedler-cooley.v:17.11-17.29"
  cell $add $add$tests/simple/fiedler-cooley.v:17$3
    parameter \A_SIGNED 0
    parameter \A_WIDTH 9
    parameter \B_SIGNED 0
    parameter \B_WIDTH 2
    parameter \Y_WIDTH 10
    connect \A \count_out
    connect \B 2'11
    connect \Y $add$tests/simple/fiedler-cooley.v:17$3_Y
  end
  attribute \src "tests/simple/fiedler-cooley.v:28.17-28.31"
  cell $and $and$tests/simple/fiedler-cooley.v:28$5
    parameter \A_SIGNED 0
    parameter \A_WIDTH 1
    parameter \B_SIGNED 0
    parameter \B_WIDTH 1
    parameter \Y_WIDTH 1
    connect \A \up
    connect \B $add$tests/simple/fiedler-cooley.v:17$3_Y [9]
    connect \Y $and$tests/simple/fiedler-cooley.v:28$5_Y
  end
  attribute \src "tests/simple/fiedler-cooley.v:29.17-29.33"
  cell $and $and$tests/simple/fiedler-cooley.v:29$6
    parameter \A_SIGNED 0
    parameter \A_WIDTH 1
    parameter \B_SIGNED 0
    parameter \B_WIDTH 1
    parameter \Y_WIDTH 1
    connect \A \down
    connect \B $sub$tests/simple/fiedler-cooley.v:16$2_Y [9]
    connect \Y $and$tests/simple/fiedler-cooley.v:29$6_Y
  end
  attribute \src "tests/simple/fiedler-cooley.v:27.17-27.27"
  cell $reduce_xor $reduce_xor$tests/simple/fiedler-cooley.v:27$4
    parameter \A_SIGNED 0
    parameter \A_WIDTH 9
    parameter \Y_WIDTH 1
    connect \A $1\count_nxt[8:0]
    connect \Y $reduce_xor$tests/simple/fiedler-cooley.v:27$4_Y
  end
  attribute \src "tests/simple/fiedler-cooley.v:16.11-16.30"
  cell $sub $sub$tests/simple/fiedler-cooley.v:16$2
    parameter \A_SIGNED 0
    parameter \A_WIDTH 9
    parameter \B_SIGNED 0
    parameter \B_WIDTH 3
    parameter \Y_WIDTH 10
    connect \A \count_out
    connect \B 3'101
    connect \Y $sub$tests/simple/fiedler-cooley.v:16$2_Y
  end
  attribute \src "tests/simple/fiedler-cooley.v:14.1-31.4"
  process $proc$tests/simple/fiedler-cooley.v:14$1
    assign { } { }
    assign { } { }
    assign { } { }
    assign { } { }
    assign { } { }
    assign { } { }
    assign { } { }
    assign $0\cnt_dn[9:0] $sub$tests/simple/fiedler-cooley.v:16$2_Y
    assign $0\cnt_up[9:0] $add$tests/simple/fiedler-cooley.v:17$3_Y
    assign $0\count_nxt[8:0] $1\count_nxt[8:0]
    assign $0\parity_out[0:0] $reduce_xor$tests/simple/fiedler-cooley.v:27$4_Y
    assign $0\carry_out[0:0] $and$tests/simple/fiedler-cooley.v:28$5_Y
    assign $0\borrow_out[0:0] $and$tests/simple/fiedler-cooley.v:29$6_Y
    assign $0\count_out[8:0] $1\count_nxt[8:0]
    attribute \src "tests/simple/fiedler-cooley.v:19.2-25.9"
    switch { \up \down }
      attribute \src "tests/simple/fiedler-cooley.v:0.0-0.0"
      case 2'00
        assign { } { }
        assign $1\count_nxt[8:0] \data_in
      attribute \src "tests/simple/fiedler-cooley.v:0.0-0.0"
      case 2'01
        assign { } { }
        assign $1\count_nxt[8:0] $sub$tests/simple/fiedler-cooley.v:16$2_Y [8:0]
      attribute \src "tests/simple/fiedler-cooley.v:0.0-0.0"
      case 2'10
        assign { } { }
        assign $1\count_nxt[8:0] $add$tests/simple/fiedler-cooley.v:17$3_Y [8:0]
      attribute \src "tests/simple/fiedler-cooley.v:0.0-0.0"
      case 2'11
        assign { } { }
        assign $1\count_nxt[8:0] \count_out
      attribute \src "tests/simple/fiedler-cooley.v:0.0-0.0"
      case 
        assign { } { }
        assign $1\count_nxt[8:0] 9'x
    end
    sync posedge \clock
      update \count_out $0\count_out[8:0]
      update \carry_out $0\carry_out[0:0]
      update \borrow_out $0\borrow_out[0:0]
      update \parity_out $0\parity_out[0:0]
      update \cnt_up $0\cnt_up[9:0]
      update \cnt_dn $0\cnt_dn[9:0]
      update \count_nxt $0\count_nxt[8:0]
  end
end

接下来执行指令:

proc; opt

将process转换为网络元素并做一些优化,结果如下:

5. Executing PROC pass (convert processes to netlists).

5.1. Executing PROC_CLEAN pass (remove empty switches from decision trees).
Cleaned up 0 empty switches.

5.2. Executing PROC_RMDEAD pass (remove dead branches from decision trees).
Removed 1 dead cases from process $proc$tests/simple/fiedler-cooley.v:14$1 in module up3down5.
Marked 1 switch rules as full_case in process $proc$tests/simple/fiedler-cooley.v:14$1 in module up3down5.
Removed a total of 1 dead cases.

5.3. Executing PROC_PRUNE pass (remove redundant assignments in processes).
Removed 0 redundant assignments.
Promoted 7 assignments to connections.

5.4. Executing PROC_INIT pass (extract init attributes).

5.5. Executing PROC_ARST pass (detect async resets in processes).

5.6. Executing PROC_ROM pass (convert switches to ROMs).
Converted 0 switches.


5.7. Executing PROC_MUX pass (convert decision trees to multiplexers).
Creating decoders for process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
     1/1: $1\count_nxt[8:0]

5.8. Executing PROC_DLATCH pass (convert process syncs to latches).

5.9. Executing PROC_DFF pass (convert process syncs to FFs).
Creating register for signal `\up3down5.\count_out' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$12' with positive edge clock.
Creating register for signal `\up3down5.\carry_out' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$13' with positive edge clock.
Creating register for signal `\up3down5.\borrow_out' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$14' with positive edge clock.
Creating register for signal `\up3down5.\parity_out' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$15' with positive edge clock.
Creating register for signal `\up3down5.\cnt_up' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$16' with positive edge clock.
Creating register for signal `\up3down5.\cnt_dn' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$17' with positive edge clock.
Creating register for signal `\up3down5.\count_nxt' using process `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
  created $dff cell `$procdff$18' with positive edge clock.

5.10. Executing PROC_MEMWR pass (convert process memory writes to cells).

5.11. Executing PROC_CLEAN pass (remove empty switches from decision trees).
Found and cleaned up 1 empty switch in `\up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
Removing empty process `up3down5.$proc$tests/simple/fiedler-cooley.v:14$1'.
Cleaned up 1 empty switch.

5.12. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.


6. Executing OPT pass (performing simple optimizations).

6.1. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.

6.2. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.

Removed a total of 1 cells.

6.3. Executing OPT_MUXTREE pass (detect dead branches in mux trees).
Running muxtree optimizer on module \up3down5..
  Creating internal representation of mux trees.
  Evaluating internal representation of mux trees.
  Analyzing evaluation results.
Removed 0 multiplexer ports.


6.4. Executing OPT_REDUCE pass (consolidate $*mux and $reduce_* inputs).
  Optimizing cells in module \up3down5.
Performed a total of 0 changes.

6.5. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.
Removed a total of 0 cells.

6.6. Executing OPT_DFF pass (perform DFF optimizations).

6.7. Executing OPT_CLEAN pass (remove unused cells and wires).
Finding unused cells or wires in module \up3down5..
Removed 2 unused cells and 10 unused wires.


6.8. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.

6.9. Rerunning OPT passes. (Maybe there is more to do..)

6.10. Executing OPT_MUXTREE pass (detect dead branches in mux trees).
Running muxtree optimizer on module \up3down5..
  Creating internal representation of mux trees.
  Evaluating internal representation of mux trees.
  Analyzing evaluation results.
Removed 0 multiplexer ports.


6.11. Executing OPT_REDUCE pass (consolidate $*mux and $reduce_* inputs).
  Optimizing cells in module \up3down5.
Performed a total of 0 changes.

6.12. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.
Removed a total of 0 cells.

6.13. Executing OPT_DFF pass (perform DFF optimizations).

6.14. Executing OPT_CLEAN pass (remove unused cells and wires).
Finding unused cells or wires in module \up3down5..

6.15. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.

6.16. Finished OPT passes. (There is nothing left to do.)

接着执行指令

show

可以看到xdot格式的网络图:

集成电路设计开源EDA软件yosys详解1:工具安装_第4张图片

执行指令:

techmap; opt

将网络表转换为逻辑门,并做一些优化,执行结果如下:

10. Executing TECHMAP pass (map to technology primitives).

10.1. Executing Verilog-2005 frontend: /home/joezhoushen/yosys/share/techmap.v
Parsing Verilog input from `/home/joezhoushen/yosys/share/techmap.v' to AST representation.
Generating RTLIL representation for module `\_90_simplemap_bool_ops'.
Generating RTLIL representation for module `\_90_simplemap_reduce_ops'.
Generating RTLIL representation for module `\_90_simplemap_logic_ops'.
Generating RTLIL representation for module `\_90_simplemap_compare_ops'.
Generating RTLIL representation for module `\_90_simplemap_various'.
Generating RTLIL representation for module `\_90_simplemap_registers'.
Generating RTLIL representation for module `\_90_shift_ops_shr_shl_sshl_sshr'.
Generating RTLIL representation for module `\_90_shift_shiftx'.
Generating RTLIL representation for module `\_90_fa'.
Generating RTLIL representation for module `\_90_lcu'.
Generating RTLIL representation for module `\_90_alu'.
Generating RTLIL representation for module `\_90_macc'.
Generating RTLIL representation for module `\_90_alumacc'.
Generating RTLIL representation for module `\$__div_mod_u'.
Generating RTLIL representation for module `\$__div_mod_trunc'.
Generating RTLIL representation for module `\_90_div'.
Generating RTLIL representation for module `\_90_mod'.
Generating RTLIL representation for module `\$__div_mod_floor'.
Generating RTLIL representation for module `\_90_divfloor'.
Generating RTLIL representation for module `\_90_modfloor'.
Generating RTLIL representation for module `\_90_pow'.
Generating RTLIL representation for module `\_90_pmux'.
Generating RTLIL representation for module `\_90_demux'.
Generating RTLIL representation for module `\_90_lut'.
Successfully finished Verilog frontend.

10.2. Continuing TECHMAP pass.
Using extmapper simplemap for cells of type $eq.
Using template $paramod$1e99ce38f701dd11f85f107c1bfc7d0aa5d10769\_90_pmux for cells of type $pmux.
Using extmapper simplemap for cells of type $and.
Using extmapper simplemap for cells of type $logic_not.
Using extmapper simplemap for cells of type $dff.
Running "alumacc" on wrapper $extern:wrap:$sub:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=3:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47.
Using template $extern:wrap:$sub:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=3:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47 for cells of type $extern:wrap:$sub:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=3:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47.
Running "alumacc" on wrapper $extern:wrap:$add:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=2:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47.
Using template $extern:wrap:$add:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=2:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47 for cells of type $extern:wrap:$add:A_SIGNED=0:A_WIDTH=9:B_SIGNED=0:B_WIDTH=2:Y_WIDTH=10:394426c56d1a028ba8fdd5469b163e04011def47.
Using extmapper simplemap for cells of type $reduce_xor.
Using extmapper simplemap for cells of type $reduce_or.
Using extmapper simplemap for cells of type $mux.
Using template $paramod$9ba033cb42591e01e61f078c95ea0ab8e1e30f70\_90_alu for cells of type $alu.
Using template $paramod$de9286bb3f6c9573089ef1694e580944d318189a\_90_alu for cells of type $alu.
Using extmapper simplemap for cells of type $xor.
Using template $paramod\_90_lcu\WIDTH=32'00000000000000000000000000001010 for cells of type $lcu.
Using extmapper simplemap for cells of type $pos.
Using extmapper simplemap for cells of type $not.
Using extmapper simplemap for cells of type $or.
No more expansions possible.


11. Executing OPT pass (performing simple optimizations).

11.1. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.


11.2. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.

Removed a total of 19 cells.

11.3. Executing OPT_MUXTREE pass (detect dead branches in mux trees).
Running muxtree optimizer on module \up3down5..
  Creating internal representation of mux trees.
  No muxes found in this module.
Removed 0 multiplexer ports.

11.4. Executing OPT_REDUCE pass (consolidate $*mux and $reduce_* inputs).
  Optimizing cells in module \up3down5.
Performed a total of 0 changes.

11.5. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.
Removed a total of 0 cells.

11.6. Executing OPT_DFF pass (perform DFF optimizations).

11.7. Executing OPT_CLEAN pass (remove unused cells and wires).
Finding unused cells or wires in module \up3down5..
Removed 1 unused cells and 145 unused wires.


11.8. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.

11.9. Rerunning OPT passes. (Maybe there is more to do..)

11.10. Executing OPT_MUXTREE pass (detect dead branches in mux trees).
Running muxtree optimizer on module \up3down5..
  Creating internal representation of mux trees.
  No muxes found in this module.
Removed 0 multiplexer ports.

11.11. Executing OPT_REDUCE pass (consolidate $*mux and $reduce_* inputs).
  Optimizing cells in module \up3down5.
Performed a total of 0 changes.

11.12. Executing OPT_MERGE pass (detect identical cells).
Finding identical cells in module `\up3down5'.
Removed a total of 0 cells.

11.13. Executing OPT_DFF pass (perform DFF optimizations).

11.14. Executing OPT_CLEAN pass (remove unused cells and wires).
Finding unused cells or wires in module \up3down5..

11.15. Executing OPT_EXPR pass (perform const folding).
Optimizing module up3down5.

11.16. Finished OPT passes. (There is nothing left to do.)

基本操作先介绍到这里,后续会对代码做详细解析。

你可能感兴趣的:(FPGA+EDA,fpga开发,EDA)