When you try to design or evaluate using PCI Express compatible ICs, have you ever been confused by questions like "What is a lane?" or "How do I view bandwidth?"
When understanding how PCI Express works and the differences between generations, it's helpful to clarify these terms a little at the beginning, as this will make future understanding much easier. This article is aimed at engineers who are new to PCI Express and provides an easy-to-understand summary, with illustrations, of the "ultra-basic" terms and concepts that you should know before starting your design.
Please use this as your first step in learning about PCI Express.
What is an interface?
In order for ICs to communicate with each other, a "path" for signals to pass through is required. This path is called an "interface."
An interface includes the type of signal, communication method, physical connection method, etc., and is like a "common rule" for ICs to operate correctly. PCI Express is one such interface, and enables high-speed data exchange between ICs based on a specific communication method and physical specifications.
Serial communication and lane concept
Difference between serial and parallel communication
Methods for exchanging data between ICs can be broadly divided into "parallel communication" and "serial communication." Parallel communication is a method of sending data simultaneously using multiple signal lines, and was often used in older bus interfaces. Serial communication, on the other hand, is a method of sending data sequentially over a single signal line, and is characterized by its resistance to noise and ease of speed increase. Figure 1 provides a simple comparison of the differences between parallel communication and serial communication. With serial communication, you can see how data flows sequentially over a single signal line.
Lane's way of thinking
PCI Express uses this serial communication, and the signal lines are called "lanes." A lane is a set of transmit and receive lines, and by bundling multiple lanes together depending on the application, such as x1 (1 lane), x4 (4 lanes), or x16 (16 lanes), more data can be exchanged simultaneously.
The difference between bandwidth and transfer speed
When ICs communicate with each other, "how much data can be exchanged and at what speed" is a very important point.
The terms that are often used in this context are "bandwidth" and "transfer speed."
Both words are related to "speed," but their meanings are slightly different. Let's think about this difference using the analogy of roads and cars.
Bandwidth is an indicator of how much information a communication channel can carry, and is usually expressed in bits per second (bps).
→ This can be likened to the number of lanes on a road. The more lanes there are, the more cars (data) can pass through at the same time, so the wider the bandwidth.
Transfer speed (Throughput) is an indicator of the amount of data actually exchanged, and is often expressed in bytes (B/s). In some cases, the unit used is transfer count (T/s), which indicates how many times data can be transferred per second.
→ This can be likened to the number of cars actually on the road. Even if the road is wide, if there are few cars, the transfer speed will be low.
Figure 2 uses road lanes and the flow of cars to visualize the difference between bandwidth and transfer speed. When designing or selecting ICs, it is important to consider not only the width of the communication path (bandwidth) but also the actual amount of data flowing (transfer speed).
Protocol Layer Overview
Communications are organized in layers
When digital devices exchange data with each other, it is not enough to simply send signals; rules must be established regarding the order, format, how to prevent errors, etc. These rules are organized by role and are called "protocol layers."
Why divide into layers?
Dividing the communication mechanism into layers has the advantage of making design and troubleshooting easier. For example, if you separate error handling into "this layer" and signal transmission and reception into "this layer," you can improve and change only those layers.
Three commonly used layers
Communication protocols often have three layers:
・Transaction Layer
This layer defines the purpose of communication (reading, writing, etc.) and issues instructions for data exchange.
・Data Link Layer
A layer that ensures error-free communication between adjacent devices
・Physical layer
The layer that sends and receives data as actual electrical signals
PCI Express communicates using the three layers mentioned above: the transaction layer, data link layer, and physical layer. Understanding the role of each layer will help you understand where ICs are involved and what you need to pay attention to.
Summary: Before you start designing, understand the meaning of the words.
Before starting a PCI Express design, it's a good idea to first have a rough understanding of the terms and concepts that are frequently used, which will make it easier to read the documents and specifications that follow. Even if you just clarify the meaning of the terms before getting into the detailed specifications and design, you should feel a little closer to the world of PCI Express.
Recommended related information
For more information on Parade Technologies' PCI Express products, please see the following page:
Inquiry
If you have any questions regarding this article, please contact us below.
Parade Technologies Manufacturer Information Top
Parade Technologies Manufacturer Information If you would like to return to the top page, please click below.