Method and apparatus for generating gate-level activity data for use in clock gating efficiency analysis
View Patent ↗A mechanism for generating gate-level activity data for use in clock gating efficiency analysis of an integrated circuit (IC) design is provided. Generating the gate-level activity data includes generating a signal behaviour description for inter-register signals, generating a gate-level netlist for the IC design, generating gate-level stimuli based at least partly on the generated signal behaviour description, and performing gate-level simulation using the generated gate-level stimuli to generate gate-level activity data for the IC design. In one embodiment, generating the signal behaviour description includes performing Register Transfer Level (RTL) simulation of the IC design, and generating the gate-level netlist includes performing RTL synthesis. The RTL simulation and RTL synthesis are performed on RTL data for the IC design.
1. A method comprising:
generating a signal behaviour description for inter-register signals by performing Register Transfer Level (RTL) simulation of an IC design;
generating a gate-level netlist for the IC design by performing an RTL synthesis;
generating gate-level stimuli using the signal behaviour description;
generating gate-level activity data for the gate-level netlist of IC design by gate level simulation using the generated gate level stimuli;
deriving at least one clock gating metric for at least one register within the IC design based on gate-level activity data therefor;
calculating, at a data processor, at least one clock gating power cost function for the at least one register within the IC design based on the clock gating metric; and
determining, at the data processor, whether efficient clock gating may be achieved for the at least one register within the IC design based on the clock gating power cost function(s) therefor.
2. The method of claim 1 , wherein said generating the signal behaviour description for inter-register signals comprises performing an Register Transfer Level (RTL) simulation on RTL data for the IC design.
3. The method of claim 2 wherein said performing RTL simulation comprises using user-defined test parameters.
4. The method of claim 3 , wherein the method comprises performing RTL synthesis on RTL data for the IC design to generate the gate-level netlist.
5. The method of claim 3 , wherein generating gate-level stimuli comprises mapping RTL registers within the signal behaviour description to registers within the gate-level netlist, and generating a gate-level stimuli for registers within the gate-level netlist based at least partly on waveforms for the RTL registers within the signal behaviour description.
6. The method of claim 2 , wherein generating gate-level stimuli comprises mapping RTL registers within the signal behaviour description to registers within the gate-level netlist, and generating a gate-level stimuli for registers within the gate-level netlist based at least partly on waveforms for the RTL registers within the signal behaviour description.
7. The method of claim 2 , wherein performing gate-level simulation comprises monitoring, in response to the generated gate-level stimuli, at least one from a group comprising at least one of:
at least one clock input of at least one register within the IC design;
at least one data input of at least one register within the IC design; and
at least one data output of at least one register within the IC design.
8. The method of claim 2 , wherein the gate-level activity data comprises at least one from a group comprising at least one of:
clock input activity for at least one register within the IC design;
data input activity for at least one register within the IC design; and
data output activity for at least one register within the IC design.
9. The method of claim 1 , wherein said generating the gate-level netlist for the IC design comprises performing the RTL synthesis on RTL data for the IC design.
10. The method of claim 9 , wherein the RTL synthesis comprises a standard, un-customized synthesis flow.
11. The method of claim 1 , wherein said generating the gate-level stimuli further comprises:
mapping RTL registers within the signal behaviour description to registers within the gate-level netlist; and
generating a gate-level stimuli for registers within the gate-level netlist based on waveforms for the RTL registers within the signal behaviour description.
12. The method of claim 1 , wherein said performing gate-level simulation comprises:
monitoring, in response to the generated gate-level stimuli, at least one from a group comprising at least one of:
at least one clock input of at least one register within the IC design,
at least one data input of at least one register within the IC design, and
at least one data output of at least one register within the IC design.
13. The method of claim 1 , wherein the gate-level activity data comprises at least one from a group comprising at least one of:
clock input activity for at least one register within the IC design,
data input activity for at least one register within the IC design, and
data output activity for at least one register within the IC design.
14. The method of claim 1 , wherein the at least one clock gating metric comprises at least one from a group comprising
data output/clock activity ratio for the at least one register within the IC design; and
maximum clock pulses per idle output window for the at least one register within the IC design.
15. The method of claim 1 , wherein the method further comprises performing said calculating clock gating power cost functions based at least partly on at least one from a group comprising at least one of:
wasted clock toggles for the at least one register within the IC design, determined based at least partly on the at least one clock gating metric;
register fan-out size from the at least one register within the IC design;
added clock skew attributable to gating at least one clock signal within the IC design; and
gated clock leakage attributable to gating at least one clock signal within the IC design.
16. The method of claim 1 further comprising:
performing efficiency analysis of an integrated circuit design based upon the gate-level activity data.
17. A non-transitory computer-readable storage medium storing instructions configured to be executed by a signal processing module to generate gate-level activity data for use in clock gating efficiency analysis of an integrated circuit (IC) design, the instructions configured to:
generate a signal behaviour description for inter-register signals;
generate a gate-level netlist for the IC design using Register Transfer Level (RTL) synthesis;
generate gate-level stimuli based at least partly on the generated signal behaviour description;
generate gate-level activity data for the IC design by performing gate level simulation using the generated gate level stimuli;
derive at least one clock gating metric for at least one register within the IC design based on gate-level activity data therefor:
calculate at least one clock gating power cost function for the at least one register within the IC design based on the clock gating metric; and
perform a clock sating efficiency analysis based upon the gating power cost function(s).
18. The non-transitory computer-readable storage medium of claim 17 wherein the instructions configured to generate the signal behaviour description for inter -register signal further comprise instructions configured to generate the signal behaviour description using RTL simulation on RTL data for the IC design.
19. The non-transitory computer-readable storage medium of claim 18 wherein the instructions configured to generate the gate-level netlist for the IC design further comprise instructions configured to perform RTL synthesis on RTL data for the IC design to generate the gate-level netlist.
20. A method comprising:
generating a signal behaviour description for inter-register signals by performing Register Transfer Level (RTL) simulation of the IC design using user defined test patterns;
generating, substantially directly from RTL data of the IC design, a frontend gate-level netlist for the IC design by performing an RTL synthesis that does not include backend synthesis modifications to the IC design;
mapping RTL registers within the signal behaviour description to gate-level registers within the gate-level netlist;
generating, based on waveforms for the RTL registers within the signal behaviour description, gate-level stimuli for gate-level registers using the signal behaviour description;
generating gate-level activity data for the gate-level netlist of IC design by gate level simulation using the generated gate level stimuli, wherein the gate level simulation comprises monitoring, in response to the generated gate-level stimuli, at least one of a clock input of a first register of the gate-level registers, a data input activity of the first register, and a data output activity of the first register;
deriving, at a data processor, at least one clock gating metric for the first register based on the gate-level activity data;
calculating, at a data processor, at least one clock gating power cost function for the first register; and
modifying, at a data processor, the IC design to clock gate the first register in response to determining, based on the power cost function, that the first register can be efficiently clock gated;
producing an integrated circuit based on the modified IC design.