In an article recently written by embedded software expert Mike Barr, he predicts that 32-bit processors will eventually defeat or completely replace 8-bit products. I still recall that in 1990, an analyst once said to me with utter decisiveness that the 8-bit dollar is dead and that in the near future it will be 32-bit.
Mike's article elicited a polarized opinion. A reader named Chuck Manning believes that the 32-bit price reduction will drive the development of more small, low-end products. In the past, I have made the same point several times. When you can buy an 8-bit component with a penny, you may open a vast application market that we could not imagine today.
Chuck also pointed out that byte-wide processors eat less power and can tolerate wider supply voltages than 32-bit processors. This is the truth, low power consumption is undoubtedly the holy grail of the electronic industry at present, however, in the foreseeable future, I have not seen any sign that can develop the non-power consumption CPU.
Another reader, Miro Samek, said that "8-bit is meaningless." His theoretical basis is that the CPU itself is only a small part of a typical microprocessor, and most of the others are memory and side. On this point, there are many opinions from both support and opposition. But basically I do not agree with this statement.
Indeed, today's Cortex-M0+ with a 40nm process requires less than 0.012. Under the constraints of transistor overhead or grain size, the CPU itself will eventually become even smaller.
However, regarding the topics discussed today, we have also seen three trends that are full of contradictions and confusion:
First of all, many very low-level components are manufactured from "antique" fabs and processes that have been completely depreciated. If such components are to adopt more advanced processes in the future, higher manufacturing costs must be paid.
Second, there is another cost that will not disappear. Let's face it: The future 32-bit microcontrollers will be ARM's world, and ARM's main revenue source is to receive authorization or royalties for each component. These figures are very secretive, but I have heard some rumors that the Cortex components have to pay a few tens of cents.
Even if all other costs are zero, it is still difficult for these components to compete in extremely price-sensitive applications. I have always believed that ARM’s biggest competitor has not yet emerged: an open-source CPU vendor that is exempt from licensing fees and can support all ARM products.
Will this happen? maybe. If it does, will it succeed? The shift from the support of proprietary tools toward the free and open end has always been a major trend in the semiconductor industry. Therefore, once an open-ended CPU has emerged, it will surely meet the manufacturer's development model. But indeed, it is difficult to see such a free transfer model, how to create a huge, mostly ARM-compatible ecosystem.
Third, the cost of silicon wafers will continue to decline until they are no longer considered issues for low-end microprocessors. The most expensive part will then be packaged, but there is no reason why high-end and low-end microcontrol packages and pins are incompatible. Think of six-pin Cortex components.
In addition, I do not agree with Miro's statement: "I think the main reason for the continued strength of 8-bit is not related to technology, but it is from the habit of the embedded development community."
There is no doubt that this fact is true, but the cost is still a key factor in pushing engineering decisions. However, in addition to costs, tools must also be considered. I recently used ARM's pretty good IDEs, but it costs thousands of dollars. In contrast, Microchip's PIC tools are almost free. Of course, you can also use GCC to develop ARM and build your own environment, but the time you need and more expertise may have to be more than a developer has, so 32-bit will win a comprehensive victory. ? Maybe, as Mike said in that article, it will get the most out of most applications. But will it come soon? I still doubt it.
Mike's article elicited a polarized opinion. A reader named Chuck Manning believes that the 32-bit price reduction will drive the development of more small, low-end products. In the past, I have made the same point several times. When you can buy an 8-bit component with a penny, you may open a vast application market that we could not imagine today.
Chuck also pointed out that byte-wide processors eat less power and can tolerate wider supply voltages than 32-bit processors. This is the truth, low power consumption is undoubtedly the holy grail of the electronic industry at present, however, in the foreseeable future, I have not seen any sign that can develop the non-power consumption CPU.
Another reader, Miro Samek, said that "8-bit is meaningless." His theoretical basis is that the CPU itself is only a small part of a typical microprocessor, and most of the others are memory and side. On this point, there are many opinions from both support and opposition. But basically I do not agree with this statement.
Indeed, today's Cortex-M0+ with a 40nm process requires less than 0.012. Under the constraints of transistor overhead or grain size, the CPU itself will eventually become even smaller.
However, regarding the topics discussed today, we have also seen three trends that are full of contradictions and confusion:
First of all, many very low-level components are manufactured from "antique" fabs and processes that have been completely depreciated. If such components are to adopt more advanced processes in the future, higher manufacturing costs must be paid.
Second, there is another cost that will not disappear. Let's face it: The future 32-bit microcontrollers will be ARM's world, and ARM's main revenue source is to receive authorization or royalties for each component. These figures are very secretive, but I have heard some rumors that the Cortex components have to pay a few tens of cents.
Even if all other costs are zero, it is still difficult for these components to compete in extremely price-sensitive applications. I have always believed that ARM’s biggest competitor has not yet emerged: an open-source CPU vendor that is exempt from licensing fees and can support all ARM products.
Will this happen? maybe. If it does, will it succeed? The shift from the support of proprietary tools toward the free and open end has always been a major trend in the semiconductor industry. Therefore, once an open-ended CPU has emerged, it will surely meet the manufacturer's development model. But indeed, it is difficult to see such a free transfer model, how to create a huge, mostly ARM-compatible ecosystem.
Third, the cost of silicon wafers will continue to decline until they are no longer considered issues for low-end microprocessors. The most expensive part will then be packaged, but there is no reason why high-end and low-end microcontrol packages and pins are incompatible. Think of six-pin Cortex components.
In addition, I do not agree with Miro's statement: "I think the main reason for the continued strength of 8-bit is not related to technology, but it is from the habit of the embedded development community."
There is no doubt that this fact is true, but the cost is still a key factor in pushing engineering decisions. However, in addition to costs, tools must also be considered. I recently used ARM's pretty good IDEs, but it costs thousands of dollars. In contrast, Microchip's PIC tools are almost free. Of course, you can also use GCC to develop ARM and build your own environment, but the time you need and more expertise may have to be more than a developer has, so 32-bit will win a comprehensive victory. ? Maybe, as Mike said in that article, it will get the most out of most applications. But will it come soon? I still doubt it.
Nantong Boxin Electronic Technology Co., Ltd. , https://www.ntbosen.com