AVR architecture inventor: should be different

The microprocessor architecture has evolved from significant differences in the past to more and more subtle differences. This is a matter of great concern.

The transition from AVR to ARM is a long process. How long will this trend be maintained? Where will you go next?

Since the 1970s, microcontrollers have quickly replaced discrete logic and support more functions such as advanced control, complex calculations, high-speed communications, and intuitive users even on low-cost micro-systems. interface.

Most of the earlier MCU code was written in assembly language, and this near-hardware programming has become an embedded designer's personality because no one can (or should) modify other people's code.

In the early 90s, engineers realized that there were more functional requirements for the microcontroller system. C language has always played an important role in 8-bit microprocessors. The biggest problem is that the microcontroller does not support the C language very well, and the amount of code is increasing and the integration is not high.

At this time we designed the AVR architecture. The AVR architecture was designed to support the C language while maximizing the usage and efficiency of the C language.

The Risc architecture consisting of single-cycle instructions, register files, cache-independent buses, and I/O is absolutely superior to any other 8-bit world architecture, and it may still be 20 years later.

The performance of the architecture is particularly important, but other factors such as built-in high performance, programmable flash, and EEPROM memory are equally important in the eyes of engineers.

Low-cost programming tools have gained a lot of interest in the programmer community, from a high price of several thousand dollars to almost free promotion of micro controllers.

At the same time ARM has also begun its journey of embedded applications, ARM and other processors are very different.

First, the computational performance of 32-bit von Neumann architectures used by other processors in the kernel design is inferior to that of Risc. Because of its 32-bit wide bus architecture, ARM devices can use data storage extensions of small microprocessors.

A 32-bit data unit can store 4 bytes, which is twice as large as most control program data. The Thumb instruction set is specifically designed to overcome these coding space issues.

This is a subset of the instruction set, it can realize the function of the application program with less code quantity and data when 32 numerical calculations are not needed. ARM's superior performance makes programmers shine and ARM can support. 32-bit applications can also support 8/16-bit applications.

Secondly, ARM has a different business model, which is to obtain the processor core in the simplest way through the authorization model, and then to conduct commercial mass production after secondary development. In today's increasingly competitive microcontroller market, it is clear that their processors are more loved by buyers, but it is well known that the same core from different vendors does not mean that it can completely replace the old kernel overnight. .

There are several key milestones in the success of ARM. First, they have succeeded in the design of cores such as Nokia's GSM mobile phones. This is of great significance to microprocessor companies. He broke the existing industry chain model to sell the core, but also created a brand new business cooperation model, which is almost impossible for other processor vendors.

In addition, due to this licensing business model, ARM's customers are very willing to develop various real-time operating systems, debuggers, and other applications for ARM cores in their Asics.

With several high-end vendors expressing their support for ARM, ARM has had low-cost processor development tools a few years later, and ARM has become popular among most of the world's programmers.

Today, 20 years later, the ARM processor occupies a dominant position in embedded 32-bit microprocessors. At the same time, ARM has used its various kernels to erode the majority of 8/16-bit applications.

The development of manufacturing processes in recent years has also made a very important contribution to the migration of 8-bit processors to 32-bit.

Because the on-chip system has integrated huge memory, peripheral devices and function modules, the number of processor cores is no longer important. On a 1 MB flash Cortex-3 device, the core only accounts for only a few percent of the silicon cost.

Another interesting phenomenon is that ARM's early success on mobile phones may have caused ARM to enter the tablet market, but also foreshadowed the arrival of a new handheld computer era. This may be the most attractive part of ARM's history. Who would have thought that a small core processor-based company could challenge the leader of this industry, Intel?

So I sincerely pay tribute to ARM's completion of this great industry of all kinds of high performance cores, a powerful electronic ecosystem and a great market machine.

Now, when most companies think that micro-innovation is particularly important when using the same technology.

Only companies with innovative teams can make innovations and create miracles. Adventurous, tolerant, persistent, curious, and innovative is the way to success for a company.