sequences and sequencer

The Sequencer’s Role

The sequencer serves as:

1. A controller for generating and managing sequences of transactions.
2. An arbiter for sequences running simultaneously, decides which sequence of the running sequences should pass its transactions to the driver.
3. the sequencer_Id is a field assigned by the sequencer to each sequence
4. within each sequence each transaction is assigned a unique Id
5. so each transaction is uniquelyidentified by the combination of its seq_id and tr_id
6. it is the sequencer responsibility to route the drivers responses correctly to the correct sequence based on the sequence_Id and tr_Id

For this, the sequencer must:

• Send transactions to the driver.
• provides implementation of the API
• Optionally receive feedback from the driver (e.g., when the driver requests the next transaction or sends status updates).

---

3. Why Use an Export?

The sequencer connects to the driver via a TLM interface (typically uvm_seq_item_pull_imp in TLM1). The sequencer needs an export because the driver calls methods implemented in the sequencer. Here’s how it works:

1. Driver Pulls Transactions:
   • The driver initiates communication by pulling transactions from the sequencer via the default_export.
   • يعني الدرايفر هو اللي بيبدا التواصل عشان كده ال API Implementation هيكون عند ال sequencer
   • Using an export ensures the sequencer provides the required TLM implementation (get_next_item(), item_done(), etc.).
2. Driver Requests and Sequencer Responds:
   • With an export, the driver can act as the active entity that makes requests (seq_item_port.get_next_item()), and the sequencer fulfills these requests.
3. Bidirectional Communication:
   • The sequencer provides callbacks (like item_done()) so that the driver can notify it when a transaction is complete or needs arbitration.
   • An export allows the driver to trigger these callbacks.

---

5. Summary of Responsibilities

Component	Interface Usage	Role
Driver	Port	Actively requests transactions from the sequencer.
Sequencer	Export	Reactively provides transactions to the driver and implements TLM methods.

The Sequence Role

• just to generate transactions
• each transaction has a unique tr_id and each sequence has a unique seq_id
• all that done inside the body() task of the sequence class

Mechanism

• the sequence class generate the transaction inside the body() task and call start_item()
• the start_item() blocks the execution in the body() untill the get_next_item() executed from the run_phase in the driver this indicates that the driver is ready to receive a transaction
• the start_item() unblocks and finish_item() executed which sends the transaction to the driver
• the item_done() in the run_phase at the driver indicates that the driver completed his job

Key Points:

• The sequence is the active component responsible for generating transactions.
• The sequencer is a reactive component that:
  • Arbitrates between sequences (if there are multiple).
  • Passes the next transaction to the driver upon request.
• The driver is also reactive and:
  • Requests transactions from the sequencer.
  • Drives the transactions to the DUT.

---

Example Flow

Imagine a testbench with one sequence, one sequencer, and one driver:

1. The sequence generates a transaction (e.g., a write operation with specific address/data).
2. The sequencer holds the transaction in its queue.
3. The driver calls seq_item_port.get_next_item() to fetch the transaction.
4. The driver processes the transaction (e.g., sends it to the DUT via an interface).
5. Once done, the driver calls item_done() to notify the sequencer.
6. The sequencer acknowledges and allows the sequence to generate the next transaction.

---

Arbitration (When Multiple Sequences Run)

If multiple sequences are active, the sequencer uses arbitration logic (e.g., round-robin, priority-based) to determine which sequence gets to send its transaction to the driver.

---

3. add_typewide_sequence(sequence_x.get_type());

• Purpose:
  • Registers a typewide sequence for this class or component.
  • A typewide sequence is a special mechanism in UVM that allows you to associate a default sequence with a sequencer. This sequence will be automatically started when the test begins, unless explicitly overridden.