|
Sesame BTF
Backtracking Free
for optimization without regression
A patented architecture
The third generation of SESAME library stems is based on a patented architecture with separation of logic cell functions from the analog net drives.
This novel architecture is directly implementable without any design flow modification.
Moreover, specific guidelines are provided to help optimize the integration.
Fig; 1: Traditional cell structure with embedded drive
|
|
Fig. 2: SESAME BTF cell structure with separation between drive and cell
|
Key features of the SESAME BTF libraries
- Typically 25 functions
- Each function is proposed with two layouts:
- one for achieving the best density
- one for the lowest power consumption
- Each function is associated with 4 types of drives enabling the best performances:
- buffer,
- inverter,
- buffer-inverter,
- tristate buffer
- 12 strengths are provided for each type of drive
By combination a total number of 2,400 equivalent traditional cells is achieved: 25 functions x 2 layout x 4 types of drives x 12 strengths, while only 98 cells must be verified… |
 |
Fig. 3: DNA of a traditional library |
Fig. 4: DNA of SESAME BTF library |
Doing at best with a traditional library
The convergence of P&R with traditional libraries can take several weeks or even months and the timeline is unpredictable. It depends on the constraints on area and timing: the higher this constraint, the longer the timeline and the higher the risk of non-convergence (see fig.1).
This phenomenon is made even worse with traditional libraries, due to the absence of control on the load capacitance when the cell drive increases. As a result each modification in a critical path impacts all the paths upstream! (see fig.2) |
|
|
Fig. 1: Relationship between Time-To-Fab and constraint on area/timing
Fig. 2: Optimization of a critical path with a traditional library: any drive replacement can impact all the nets upstream! |
Benefits of the BTF approach
Mastering the lead-time and cost of design
With SESAME BTF the cells are separated from drives: a designer can choose to replace any function or drive on a critical path in order to meet the performance targets, without the penalty of any need for adaptation of the nets upstream. A local improvement no longer disturbs the rest of the paths! (see fig.3).
Moreover, this enables to optimize the non critical paths with another objective: dynamic power or leakage.
Fine tuning the optimization with an objective and a constraint
A total of 12 drive strengths enables the largest flexibility so that the Place&Route solution can choose the most relevant drive for the best trade-off between power consumption and speed. |
Fig.3 : Local optimization of a critical path thanks to the SESAME BTF library

|
A user guide for optimized integration
Specific guidelines are provided progressively by our experienced Application Engineering for benefiting fully from the possibilities granted by SESAME BTF :
- Synthesis optimization thanks to a hierarchical approach
- Synthesizer configuration
- Usage of synthesis scripts
- Usage of WLM and better…
- IR drop, CΔV resilience and noise management
- Specific methodologies for optimization towards a specific criterion:
- Power optimization
- Area optimization
- Power consumption or area optimization with a speed constraint
- Place & route:
- Optimization of critical paths at placement: drive and cell replacement
- Clock tree optimization
- …
|
|