If a reset is needed, AMD recommends using synchronous resets. Synchronous resets have the following advantages over asynchronous resets:
- Synchronous resets can directly map to more resource elements in the device architecture.
- Asynchronous resets impact the maximum clock frequency of the general logic structures. Because all AMD device general-purpose registers can program the set/reset as either asynchronous or synchronous, it might seem like there is no penalty in using asynchronous resets. If a global asynchronous reset is used, it does not increase the control sets. However, the need to route this reset signal to all register elements increases routing complexity.
- Asynchronous resets have a greater probability of corrupting memory contents of block RAMs, LUTRAMs, and SRLs during reset assertion. This is especially true for registers with asynchronous resets that drive the input pins of block RAMs, LUTRAMs, and SRLs.
- Synchronous resets offer more flexibility for control set remapping when higher density or fine tuned placement is needed. A synchronous reset can be remapped to the data path of the register if an incompatible reset is found in the more optimally placed slice. This can reduce routing resource utilization and increase placement density where needed to allow proper fitting and improved achievable clock frequency.
- Some resources such as the DSP48 and block RAM have only synchronous resets for the register elements within the block. When asynchronous resets are used on register elements associated with these elements, those registers may not be inferred directly into those blocks without impacting functionality.
Following are additional considerations:
- The clock works as a filter for small reset glitches for synchronous resets. However, if these glitches occur near the active clock edge, the flip-flop might become metastable.
- Synchronous resets might need to stretch the pulse width to ensure that the reset signal pulse is wide enough for the reset to be present during an active edge of the clock.
- When using asynchronous resets, remember to synchronize the deassertion of the asynchronous reset. Although the relative timing between clock and reset can be ignored during reset assertion, the reset release must be synchronized to the clock. Avoiding the reset release edge synchronization can lead to metastability. During reset release, setup and hold timing conditions must be satisfied for the reset pin relative to the clock pin of a register. A violation of the setup and hold conditions for asynchronous reset (e.g., reset recovery and removal timing) might cause the flip-flop to become metastable, causing design failure due to switching to an unknown state. Note that this situation is similar to the violation of setup and hold conditions for the flip-flop data pin.