Design Compiler工具学习笔记（5）

---

 

---

目录

引言

知识储备

代码风格

DFT

实际操作

---

---

---

引言

本篇继续学习 DC的基本使用。本篇主要学习 DC 工作机理和工作过程 以及简单介绍 DFT。

前文链接：

Design Compiler工具学习笔记（1）

Design Compiler工具学习笔记（2）

Design Compiler工具学习笔记（3）

Design Compiler工具学习笔记（4） 

---

---

---

---

知识储备

自底向上模式：

假设一个top设计文件包含 A，B，C 3个子模块，那么可以先将  A 模块设为顶层模块，并且添加对其的时序和面积约束，而后进行综合。综合完成后将该模块设置为 dont_touch ；

然后将 B 模块设置为顶层，继续执行和 A 相同的操作流程，直到遍历所有模块。

如果子模块之间还存在逻辑(glue logic)，还需要对最终的顶层模块设置约束跑一遍综合。

该模式费时费力。

 

 compile_ultra 是一个功能更强大的 compile 命令，可以将模块之间的层次化结构关系取消。有利于优化面积。但是这不太利于子模块后仿真时debug。此优化也可以通过下面的选项关闭。

 

 

 

 

 

 

 适当插入寄存器可以提高设计的速度。

 流水线结构综合步骤：

 多核优化：

代码风格

DFT

  

---

  

---

---

---

  

实际操作

此处仍以上篇文章的设计文件作为示例工程。主要是更新一下脚本内容：

仅供参考~~~

# |===========================================================
# | Author 		: Xu Y. B.
# | Date   		: 2022-11-21
# | Description : tcl script for top design 
# |===========================================================
 
 
# |===========================================================
# |STEP 1: Read & elaborate the RTL design file list & check
# |===========================================================
set TOP_MODULE TOP
analyze        -format verilog [list TOP.v CAL_FUNC_MDL.v CAL_ABCD_TOP.v]
elaborate      $TOP_MODULE     -architecture verilog
current_design $TOP_MODULE
 
if {[link] == 0} {
	echo "Your Link has errors !";
	exit;
}
 
if {[check_design] == 0} {
	echo "Your check design has errors !";
	exit;
}
 
# |===========================================================
# |STEP 2: reset design
# |===========================================================
reset_design
 
 
# |===========================================================
# |STEP 3: Write unmapped ddc file
# |===========================================================
uniquify
set uniquify_naming_style "%s_%d"
write -f ddc -hierarchy -output ${UNMAPPED_PATH}/${TOP_MODULE}.ddc
 
 
# |===========================================================
# |STEP 4: define clocks
# |===========================================================
set       CLK_NAME          	I_CLK_100M
set       CLK_PERIOD        	10
set       CLK_SKEW	        	[expr {$CLK_PERIOD*0.05}]						
set       CLK_TRANS         	[expr {$CLK_PERIOD*0.01}]						
set       CLK_SRC_LATENCY   	[expr {$CLK_PERIOD*0.1 }]						
set       CLK_LATENCY       	[expr {$CLK_PERIOD*0.1 }]						
 
create_clock 			-period 	$CLK_PERIOD  	  [get_ports $CLK_NAME]
set_ideal_network 						 			  [get_ports $CLK_NAME]
set_dont_touch_network 					 			  [get_ports $CLK_NAME]
set_drive 				0 							  [get_ports $CLK_NAME]
 
set_clock_uncertainty   -setup       $CLK_SKEW        [get_clocks $CLK_NAME]
set_clock_transition    -max         $CLK_TRANS       [get_clocks $CLK_NAME]
set_clock_latency       -source -max $CLK_SRC_LATENCY [get_clocks $CLK_NAME]
set_clock_latency       -max         $CLK_LATENCY     [get_clocks $CLK_NAME]
 
 
# |===========================================================
# |STEP 5: define reset
# |===========================================================
set RST_NAME 					I_RSTN
set_ideal_network 				[get_ports $RST_NAME]
set_dont_touch_network          [get_ports $RST_NAME]
set_drive             0         [get_ports $RST_NAME]
 
 
# |===========================================================
# |STEP 6: set input delay using timing budget
# |Assume a weak cell to drive the input pins
# |===========================================================
set 		LIB_NAME 			typical
set 		WIRE_LOAD_MODEL 	smic18_wl10
set 		DRIVE_CELL 			INVX1
set 		DRIVE_PIN 			Y
set 		OPERATE_CONDITION   typical
 
set 		ALL_INPUT_EXCEPT_CLK [remove_from_collection [all_inputs] [get_ports "$CLK_NAME"]]
set         INPUT_DELAY 		 [expr {$CLK_PERIOD*0.6}]
 
set_input_delay $INPUT_DELAY -clock $CLK_NAME $ALL_INPUT_EXCEPT_CLK
# set_input_delay -min 0 -clock $CLK_NAME $ALL_INPUT_EXCEPT_CLK
set_driving_cell -lib_cell ${DRIVE_CELL} -pin ${DRIVE_PIN} $ALL_INPUT_EXCEPT_CLK
 
 
# |===========================================================
# |STEP 7: set output delay 
# |===========================================================
set OUTPUT_DELAY  [expr {$CLK_PERIOD*0.6}]
set MAX_LOAD      [expr {[load_of $LIB_NAME/INVX8/A] * 10}]
 
set_output_delay  $OUTPUT_DELAY -clock $CLK_NAME 	 [all_outputs]
set_load 		  [expr {$MAX_LOAD * 3}] 			 [all_outputs]
set_isolate_ports -type buffer 					 	 [all_outputs]
 
 
# |===========================================================
# |STEP 8: set max delay for comb logic 
# |===========================================================
# set_input_delay  [expr $CLK_PERIOD * 0.1] -clock $CLK_NAME -add_delay [get_ports I_1]
# set_output_delay [expr $CLK_PERIOD * 0.1] -clock $CLK_NAME -add_delay [get_ports O_1]
 
 
# |===========================================================
# |STEP 9: set operating condition & wire load model 
# |===========================================================
set_operating_conditions -max 			$OPERATE_CONDITION \
						 -max_library 	$LIB_NAME
 
set 					 auto_wire_load_selection false
set_wire_load_mode  	 top
set_wire_load_model		 -name    $WIRE_LOAD_MODEL \
						 -library $LIB_NAME
 
 
# |===========================================================
# |STEP 10: set area constraint (Let DC try its best) 
# |===========================================================
set_max_area			 10000
 
 
# |===========================================================
# |STEP 11: set DRC constraint 
# |===========================================================
# set MAX_CAPACITANCE [expr {[load_of $LIB_NAME/NAND4X2/Y] * 5}]
# set_max_capacitance $MAX_CAPACITANCE $ALL_INPUT_EXCEPT_CLK
 
 
# |===========================================================
# |STEP 12: set group path
# |Avoid getting stack on one path
# |===========================================================
# group_path -name $CLK_NAME -weight 5 				\
# 						   -critical_range  [expr {$CLK_PERIOD * 0.1}] 
 
# group_path -name INPUTS    -from [all_inputs] 		\
# 					 	   -critical_range  [expr {$CLK_PERIOD * 0.1}] 
 
# group_path -name OUTPUTS   -to [all_outputs] 		\
# 						   -critical_range  [expr {$CLK_PERIOD * 0.1}] 
 
# group_path -name COMBS     -from [all_inputs] 		\
# 						   -to [all_outputs] 		\
# 						   -critical_range  [expr {$CLK_PERIOD * 0.1}] 	
# report_path_group
 
 
# |===========================================================
# |STEP 13: Elimate the multiple-port inter-connect &
# |			define name style
# |===========================================================
# set_app_var 						verilogout_no_tri 					true
# set_app_var						verilogout_show_unconnected_pins 	true
# set_app_var						bus_naming_style					{%s[%d]}
# simplify_constants 				-boundary_optimization
# set_boundary_optimization 		[current_design] 					true
# set_fix_multiple_port_nets 		-all 								-buffer_constants
 
 
# |===========================================================
# |STEP 14: timing exception define
# |===========================================================
# set_false_path -from [get_clocks I_CLK_100M] -to [get_clocks I_CLK_100M]
# set ALL_CLKS [all_clocks]
# foreach_in_collection CUR_CLK $ALL_CLKS 
# {
# 	set OTHER_CLKS [remove_from_collection [all_clocks] $CUR_CLK]
# 	set_false_path -from $CUR_CLK $OTHER_CLKS
# }
 
# set_false_path -from [get_clocks I_CLK_100M] -to [get_clocks I_CLK_100M]
# set_false_path -from [get_clocks I_CLK_100M] -to [get_clocks I_CLK_100M]
 
# set_disable_timing TOP/U1 -from a -to y
# set_case_analysis 0 [get_ports sel_i]
 
# set_multicycle_path -setup 6 -from FFA/CP -through ADD/out -to FFB/D
# set_multicycle_path -hold 5 -from FFA/CP -through ADD/out -to FFB/D
# set_multicycle_path -setup 2 -to FFB/D
# set_multicycle_path -hold 1 -to FFB/D
 
 
 
# |===========================================================
# |STEP 15: compile flow
# |===========================================================
# ungroup -flatten -all
 
# 1st-pass compile
# compile -map_effort high -area_effort high
# compile -map_effort high -area_effort high -boundary_optimization
compile -map_effort high -area_effort high -scan
 
# simplify_constants -boundary_optimization
# set_fix_multiple_port_nets -all -buffer_constants
 
# compile -map_effort high -area_effort high -incremental_mapping -scan
 
# 2nd-pass compile
# compile -map_effort high -area_effort high -incremental_mapping -boundary_optimization
# compile_ultra -incr
 
 
# |===========================================================
# |STEP 16: write post-process files
# |===========================================================
# change_names -rules verilog -hierarchy
# remove-unconnected_ports [get_cells -hier *] -blast_buses
# Write the mapped files
write -f ddc -hierarchy -output $MAPPED_PATH/${TOP_MODULE}.ddc
# write 	-f verilog 	-hierarchy -output 	$MAPPED_PATH/${TOP_MODULE}.v
# write_sdc -version 	1.7 				$MAPPED_PATH/${TOP_MODULE}.sdc
# write_sdf -version 	2.1 				$MAPPED_PATH/${TOP_MODULE}.sdf
 
# |===========================================================
# |STEP 17: generate report files
# |===========================================================

编译指令为：compile -map_effort high -area_effort high 

用design version打开ddc文件：

 触发器：

 编译指令为：compile -map_effort high -area_effort high -scan

触发器：