The past of Thunderbolt interface

The Thunderbolt interface (also known as Thunderbolt on Apple devices) was first introduced in 2009. It is a high-speed I/O interface advocated by Intel. However, it was in the laboratory for a long time at that time. It was not until 2011 that it was integrated into Apple's Mac devices in the form of Mini DisplayPort (hereinafter referred to as miniDP) interface.

The essence of Thunderbolt

Yes, Thunderbolt is essentially just a technical protocol. It needs to "parasitize" on a specific interface standard, and miniDP is its first "host".

The transmission rate of Thunderbolt 1 technology is 10Gbps

In 2015, with the emergence of the USB Type-C interface standard, it was quickly targeted by the latest Thunderbolt 3 technology. To this day, there is still popular science content circulating on the Internet that "Thunderbolt 3 must be USB Type-C, but USB Type-C may not support Thunderbolt 3".

Compared with traditional USB interfaces, the biggest feature of Thunderbolt technology is that it is faster.

For example, the first generation of Thunderbolt has a transmission rate of 10Gbps, and can connect up to 4 devices in series. Compared with it, the USB3.0 (5Gbps) of the same period is simply "weak".

In 2013, USB-IF (USB Promotion Organization) released the USB3.1 technical specification, and it took a lot of effort to increase the theoretical transmission rate to 10Gbps. However, Intel has joined hands with Apple to launch Thunderbolt 2, which still uses the miniDP interface form, but the transmission rate has doubled. 20Gbps once again makes traditional USB "envious and jealous".

Thunderbolt 3, which was launched in 2015, is still the "king" among all I/O interfaces even from today's perspective. It has a transmission rate of up to 40Gbps, and this is in full-duplex state. If it is a one-way transmission, its theoretical speed can reach up to 80Gbps, which is comparable to the DisplayPort 2.0 (DP 2.0) standard that was just set in 2019.

In addition, Thunderbolt 3 is also compatible with various multimedia transmission protocols that you are familiar with, can expand almost all interfaces, and can even be used to connect external graphics cards, allowing a 1kg thin and light notebook to play the latest 3A game masterpieces through the desktop version of the RTX 2080 graphics card.

The integration of Thunderbolt

The war between Thunderbolt and USB finally saw the light of day in 2019. In early March of that year, Intel opened the Thunderbolt technology protocol to the USB Promotion Organization. USB-IF then released USB4, which realized the integration of Thunderbolt and USB protocols at the bottom layer, enhancing the compatibility between products based on the USB Type-C interface.

Unfortunately, even the USB3.2 standard has not been popularized until today, and USB4 and Thunderbolt 4 have just officially come to the fore with the 11th generation of Core. Before officially introducing this new interface, we still need to review the survival status of the most popular Thunderbolt 3 interface in the past.

The present life of Thunderbolt interface

We all know that Thunderbolt 3 is the "king" among many interfaces, and it is first-class in terms of practicality and publicity. However, looking at the mobile and desktop fields, only a very small number of high-end notebooks and designer motherboards have dabbled in it. The USB Type-C on most notebooks does not support Thunderbolt 3, and even the transmission rate remains at the level of USB3.1 Gen1 (5Gbps).

So, what is the reason why Thunderbolt 3 is difficult to popularize?

The thorny road of Thunderbolt

In the past, if a laptop wanted to support Thunderbolt 3, it needed to purchase and install an additional Thunderbolt 3 controller chip, and Intel also had to charge a hefty licensing fee.

These extra expenses are nothing for the already expensive Apple MacBook, but they are an unbearable burden for Windows laptops that focus on "small profits but quick turnover".

MacBook Pro 2018 disassembly diagram, with 1 Intel JHL7540 Thunderbolt 3 controller chip on each side of the motherboard

In addition, Thunderbolt 3 needs to occupy 4 full-speed PCI-E 3.0 channels (i.e. PCI-E 3.0×4) to achieve full-speed performance of 40Gbps. The problem is that before the 8th generation Core (mobile version), the entire platform had only 16 PCI-E 3.0 channels. After matching high-performance discrete graphics and high-speed PCI-E 3.0×4 M.2 SSD hard drives, if Thunderbolt 3 is added, it may cause the problem of "grabbing bandwidth". ·

In reality, many notebooks that support Thunderbolt 3 only have 2 PCI-E 3.0 channels (i.e. PCI-E 3.0×2) for this interface, which is also called "half-speed Thunderbolt 3" in the industry. Its theoretical transmission rate is only 20Gbps. Although the function is not affected, the performance loss is more obvious when connecting an external graphics card. Therefore, in the eyes of countless enthusiasts, only the full-speed Thunderbolt 3 is the real Thunderbolt 3.

By 2019, these problems "seem" to have been effectively solved.

Assumed native mechanism

As mentioned above, Intel opened the license for Thunderbolt 3 in 2019, and the relevant licensing fees were reduced in disguise. In addition, it is said that all 10th-generation Core processor platforms "natively support" Thunderbolt 3, and platforms represented by Ice Lake have increased the number of PCI-E 3.0 channels to 32. From all indications, it would be unfair if new products are not equipped with Thunderbolt 3!

We all know that the 10th-generation Core includes two major platforms: 10nm Ice Lake (thin and light notebooks) and 14nm Comet Lake (thin and light notebooks and gaming notebooks). From the underlying architecture, Ice Lake integrates the Thunderbolt 3 controller into the CPU core, and natively supports 4 Thunderbolt 3s, with a PCI-E 3.0 channel count of up to 32. Although 16 of these channels (4×4) are exclusively for Thunderbolt 3, they will no longer occupy the number of PCI-E buses provided by the CPU or compete with PCH for the already crowded bus resources.

However, do you think that Ice Lake "natively supports" Thunderbolt 3, and that notebooks equipped with this platform processor must be equipped with Thunderbolt 3 interfaces as standard?

The answer is of course that you think too much. At this stage, the Thunderbolt 3 interface needs to support USB PD protocol charging, which requires an additional PD power management chip.

In addition, a design like Ice Lake also means that the distance between the Thunderbolt 3 interface and the CPU (with integrated controller) is far, and there is a certain signal attenuation problem. For this reason, if Ice Lake platform notebooks want to support full-blooded Thunderbolt 3, they also need to be equipped with a signal enhancer called JHL8040R Retimer. Although a Retimer chip can directly connect to two full-speed (40Gbps) Thunderbolt 3 interfaces, it still costs 2.4 US dollars, and there are also R&D costs such as optimized layout of PCB motherboards.

Price and parameter information from Intel's official website

Let's take a look at the Comet Lake platform. Its underlying architecture is not much different from Intel's early platforms. The Thunderbolt 3 function is still attached to the PCH (traditional north bridge) chipset.

In other words, notebooks equipped with the Comet Lake platform still need to be equipped with additional Thunderbolt 3 controller chips to support Thunderbolt 3. It is not the legendary "native support". Not only is the cost higher, but there is also the problem of competing with other devices for PCI-E 3.0 channels and occupying DMI 3.0 bus resources.

If the Thunderbolt 3 function of the Ice Lake platform is implemented in the path of "CPU (integrated Thunderbolt master) → Retimer → interface", then the Comet Lake platform is "CPU → DMI → PCH → Thunderbolt master → interface", with more steps, and greater cost, performance and power consumption losses.

Now, everyone knows why even the 10th generation Core notebooks on the Ice Lake platform can be equipped with Thunderbolt 3 interfaces, but there are so few products?

Similarly, although the 11th generation Core is also claimed to natively support Thunderbolt 4, it also has the above-mentioned problems, so only mid-to-high-end 11th generation Core new products can use this interface. The USB and USB Type-C of most low-end models still maintain 5Gbps, which is the speed of USB3.0 (including 3.1 Gen1 and 3.2 Gen1).

The future of Thunderbolt interface

The question is again, the 11th generation Core Tiger Lake platform supports USB4 and Thunderbolt 4, what is the difference between them?

Thunderbolt 4 changes and constants

USB4 has a "half-speed version" with a transmission rate of 20Gbps, which is used to replace the existing USB3.2 Gen2×2 standard. At the same time, USB4 also has a higher-speed USB 40, which we can understand as a simplified version of Thunderbolt 3.

Thunderbolt 4 can be regarded as a "full-featured USB4". It not only needs to have a full speed of 40Gbps, but also needs to be compatible with USB PD charging and all the functions of the original Thunderbolt 3. It has higher certification requirements in terms of video output capabilities (supporting dual 4K monitors), supporting cables and safety.

With the popularity of the 10th and 11th generation Core platforms, the cost of adding Thunderbolt 3 (and future Thunderbolt 4) functions to thin and light notebooks is not too high. It can become the standard configuration of mainstream products priced at 600 to 800 USD. The real factor hindering the popularization of Thunderbolt functions is the lack of killer application environments.

For example, now the full-featured USB Type-C (supporting USB PD charging and PD video output) is sufficient for the needs of most users, so why spend more money to equip it with Thunderbolt 4? Although the external graphics card expansion dock that many players are particularly concerned about is considered an exclusive selling point of the Thunderbolt interface, the price of the relevant expansion dock is very expensive, and there are various compatibility issues between it and the notebook, and the experience is not perfect.

For notebook manufacturers, the Thunderbolt interface is costly, and there is no need to force it to be installed when the application scope is not large; for software and peripheral manufacturers, the popularity of the Thunderbolt interface is not high, and there is no need to develop more products and functions for it. In the end, this is the current situation we are facing: the Thunderbolt interface is too expensive and useless.

The emergence of USB4 will also take away a lot of market share that should have belonged to Thunderbolt 4, because the former requires lower costs. The hope of Thunderbolt 4 lies in Intel's latest Evo platform certification program, which is what we used to call Project Athena 2.0. Thunderbolt 4 and the 11th generation Core are the basic configurations of the Evo platform, but the price of new products that have passed Evo certification generally starts at $1,000. It seems that it is still a long way to go for lower-priced thin and light notebooks to enjoy the emperor-level treatment of Thunderbolt 4 with external graphics cards.