7.2.5 Self-timed Circuits

TL;DR

The content explains the concept of synchronous and asynchronous handshake protocols in processing pipelines, highlighting the benefits of asynchronous systems for increasing throughput.

Transcript

We've been designing our processing pipelines to have all the stages operate in lock step, choosing the clock period to accommodate the worst-case processing time over all the stages. This is what we'd call a synchronous, globally timed system. But what if there are data dependencies in the processing time, i.e., if for some data inputs a particula... Read More

Key Insights

• ⌛ Synchronous, globally timed systems have been traditionally used in processing pipelines.
• 😒 Asynchronous systems use a handshake protocol with two signals to transfer data between stages when they are ready.
• 🤝 The handshake protocol in asynchronous systems involves four phases for signaling and coordinating data transfers.
• 🌐 Asynchronous systems can operate without a global clock, relying on transition signaling between stages.
• 👻 Allowing stages to operate asynchronously can increase throughput by exploiting data dependencies and reducing waiting time.
• ❓ Asynchronous systems provide more flexibility and efficiency compared to synchronous systems in processing pipelines.
• ❓ Multiple downstream modules in asynchronous systems can be coordinated using special circuits.

Questions & Answers

Q: What is the difference between synchronous and asynchronous systems in processing pipelines?

In synchronous systems, all stages operate in lock step based on a single system clock, while asynchronous systems allow stages to operate at their own internal timing, improving throughput.

Q: How does the handshake protocol work in asynchronous systems?

The handshake protocol involves four phases: the upstream stage asserts HERE-IS-X, the downstream stage asserts GOT-X to consume the input, the downstream stage waits for HERE-IS-X to go low, and finally, both stages deassert their signals to complete the transfer.

Q: Can an asynchronous system operate without a global clock?

Yes, an asynchronous self-timed system can operate without a global clock by relying on transition signaling between stages based on the conditions required by the handshake protocol.

Q: What is the benefit of using an asynchronous system?

Asynchronous systems allow stages to operate at their own timing, which can exploit data dependencies and increase throughput by processing data at the fastest possible speed instead of waiting for worst-case processing delays.

Summary & Key Takeaways

• The content discusses the traditional approach of synchronous, globally timed systems in processing pipelines, where all stages operate in lock step based on a single system clock.

• It introduces the concept of asynchronous systems, which use a handshake protocol with two signals (HERE-IS-X and GOT-X) to transfer data between stages when they are ready.

• The content explains the four phases of the handshake protocol in asynchronous systems and highlights the advantages of allowing stages to operate at their own internal timing, increasing throughput.