Basic Setup to Use Vera

1. Basic Setup to Use Vera • module load synopsys/vera • Documentation is in: • /opt/synopsys/vera/X-2005.06-1/vera_vX-2005.061_sparcOS5/doc • user_guide.pdf is a good starting place

2. The Standard Vera Testbench Structure • There will be files for each block

3. Running Vera module adder(in0, in1, out0, clk); input[7:0] in0, in1; input clk; output[8:0] out0; reg[8:0] out0; always@(posedge clk) begin out0 <= in0 + in1; end endmodule • Making a testbench for an adder module • First, invoke the template generator > vera -tem -t adder -c clk adder.v

4. Template Generator Results • The Interface File (adder.if.vrh) • Defines the interface beween the DUT and the Vera program // adder.if.vrh #ifndef INC_ADDER_IF_VRH #define INC_ADDER_IF_VRH interface adder{ output [7:0] in0 OUTPUT_EDGE OUTPUT_SKEW; output [7:0] in1 OUTPUT_EDGE OUTPUT_SKEW; input [8:0] out0 INPUT_EDGE #-1; input clk CLOCK;} //end of interface adder #endif • Modify if additional signals need to be controlled/observed

5. More Template Generator Results • The test_top File (adder.test_top.v) • Instantiates top-level testbench (Vera shell + DUT) vera_shell vshell( //Vera shell file. .SystemClock(SystemClock), .adder_in0 (in0), .adder_in1 (in1), .adder_out0 (out0), .adder_clk (clk),); ‘ifdef emu /*DUT is in emulator, so not instantiated here*/ ‘else adder dut( .in0 (in0), .in1 (in1), .out0 (out0), .clk (clk),); ‘endif

6. More Template Generator Results • The Vera File (adder.vr.tmp -> adder.vr) • Contains the Vera code which describes the testbench #define OUTPUT_EDGE PHOLD #define OUTPUT_SKEW #1 #define INPUT_EDGE PSAMPLE #include <vera_defines.vrh> // define any interfaces, and verilog_node here #include "adder.if.vrh” program adder_test { // Here is the main Vera code }

7. More Template Generator Results • The Vera Shell File (adder_shell.v) • Verilog code which communicates with the Vera testbench Module vera_shell( SystemClock, adder_in0, adder_in1, adder_out0, adder_clk ); input SystemClock; output [7:0] adder_in0; output [7:0] adder_in1; input [8:0] adder_out0; inout adder_clk; wire SystemClock; // Other components of this file are not listed here

8. Running Vera, Remaining Steps • Compile the adder.vr file vera -cmp -vlog adder.vr • Create the executable vcs -vera adder.v adder_shell.v adder.test_top.v • Run the simulation simv +vera_load=adder.vro

9. Running Vera, The Big Picture

10. Randomization in Vera • random() returns a random integer (urandom,random_range, rand_poisson, etc.) • randomize() method built into all classes will randomly fill all rand properties of an object class Foo { rand integer x; } class Bar { rand integer y; } program main { Foo foo = new(); Bar bar = new(); integer z; void = foo.randomize(); void = bar.randomize(); } • randomize() calls itself recursively on attribute classes

11. Random Instruction Execution • randcase specifies a block of statements, one of which is executed randomly randcase { weight1 : statement1 weight2 : statement2 ... weightN : statementN } randcase{ 10: i=1; 20: i=2; 50: i=3; }

12. Constraint Based Randomization • constraints are parts of a class, just like functions, tasks, and properties (attributes) • Constraints define a relationship between variables, at least one of which should be random class Bus{ rand reg[15:0] addr; rand reg[31:0] data; constraint word_align {addr[1:0] == ‘2b0;} }

13. Randomize with Constraints • Each call to randomize fills new random data into each rand property according to the constraints of the class program test{ Bus bus = new; repeat (50){ integer result = bus.randomize(); if (result == OK) printf("addr = %16h data = %32h\n” ,bus.addr, bus.data); else printf("Randomization failed.\n"); } }

14. Random Sequence Generation • A (context free) language is defined as a set of production rules • Production rules are selected randomly when there is choice randseq (production_name){ production_definition1; production_definition2; ... production_definitionN; }

15. Random Sequence Generation • Non-terminals are replaced using production rule • Generation is complete when only terminals remain Production rule example: All possible sentences: main : top middle bottom; top : add | dec; middle : popf | pushf; bottom : mov; add popf mov add pushf mov dec popf mov dec pushf mov • Great for generating instruction sequnces • Weights can be added top : &(2) add | $(1) dec;

16. Functional Coverage • You may want to keep track of how many times a particular event occurs during simulation • ex. How often is x=1? Does y ever become 0? etc… • A coverage_group defines the interesting events that you want to keep track of coverage_group CovGroup{ sample_event = @ (posedge CLOCK); sample var1, ifc.sig1; sample s_exp(var1 + var2); } program covTest{ integer var1, var2; CovGroup cg = new(); }

17. Coverage Bins • To fully define a coverage group, you must know what signal state and transitions you want to count coverage_group CovGroup{ sample_event = @ (posedge CLOCK); sample var1, ifc.sig1; sample s_exp(var1 + var2); } • Var1 is sampled, but what values am I interested in? • State bins and transition bins define the values you want to count

18. User-Defined Coverage Bins coverage_group MyCov () { sample_event = @ (posedge clock); sample port_number { state s0(0:7); state s1(8:15); trans t1("s0"->"s1"); } } • s0 is values 0-7 • s1 is values 8-15 • t1 is a transition from s0 to s1