目录
1.Use of Macros to Overcome Errors Faced in Package
5. 最小的factory override for stimulus objects
6.避免使用uvm_printer,使用conver2string()
8. Loop performance optimization
1.Use of Macros to Overcome Errors Faced in Package
package pkg1;
`include "adder_design.sv"
`include "tb.sv"
endpackage
package pkg2;
`include "adder_design.sv"
`include "adder_tb.sv"
endpackage
module top();
`ifndef ADDER_DESIGN
`define ADDER_DESIGN
import pkg2::*;
`endif
...
endmodule
2.fork-join_none 中使用loop
使用automatic
module top;
initial begin
for (int i = 0; i < 4; i++) begin
fork
automatic int l = i;
display(l);
join_none
end
end
task display(int i);
$display("i = %d", i);
endtask
endmodule
3.random 机制
3.1 如果名字重复,使用local
rand_success = trans.randomize() with {trans.addr == local::addr; };
3.2 动态数组需要定义size
class Ethernet;
rand bit [3:0]payload[];
constraint c { payload.size() ==4;}
endclass
3.3 检查randomize 结果
4.uvm_config_db的使用要注意
a.尽量避免传递数值,有需要的话可以封装成class
b. wildcard谨慎使用
c. it is recommended to minimize the number of uvm_config_db entries.
...
5. 最小的factory override for stimulus objects
Low performance, 因为多次调用了factory中的create
//Low performance code
task body;
seq_item item;
repeat (200) begin
item = seq_item::type_id::crease("item"0;
start_item(item);
if (item.randomize()) begin ... end
finish_item(item);
endtask
High performane, 只第一次create,然后clone 它
//High performance
task body;
seq_item orig_item = seq_item::type_id::create("item");
seq_item item;
repeat(200) begin
$cast(item, orig_item.clone());
start_item(item);
assert(item.randomize());
finish_item(item);
end
endtask
6.避免使用uvm_printer,使用conver2string()
在class 自定义conver2string
virtual function string convert2string();
string s = super.convert2string();
s = { s, $psprintf( "\nname : %s", get_name() ) };
s = { s, $psprintf( "\nflavor : %s", flavor.name() ) };
s = { s, $psprintf( "\ncolor : %s", color.name() ) };
return s;
endfunction: convert2string
7.UVM objections 使用
Objections should only be used by the controlling threads, and it is also very necessary to place the objections in the run-time method of the top level test class, or in the body method of a virtual sequence. Using them in any other place is likely to be unnecessary and also cause a degradation in performance.
在virtual sequence中的body;或者在top level的testcase的run-time method
//High Performance code
class sequence extends uvm_sequence#(seq_item);
task body;
seq_item item = seq_item::type_id::create("item");
repeat(5) begin
start_item(item);
assert(item.randomize());
finish_item(item);
end
sequencer seqr;
task body;
`uvm_objection objection = new("objection");
sequence seq= sequence::type_id::create("seq");
objection.raise_objection(seqr);
seq.start(seqr);
objection.drop_objection(seqr);
endtask
8. Loop performance optimization
1.loop的性能取决于loop中的工作
下面高效代码,减少了每次循环化size的计算
//Less efficient code
int arr[];
int total = 0;
for(int i = 0;i< arr.size();i++) begin
total += arr[i];
end
//High Performance Code
int arr[];
int arr_size;
int tot = 0;
arr_size = arr.size();
for(int i = 0; i < arr_size; i++) begin
tot += arr[i];
end
9.文件目录