Sram 10 megabytes

[MTS]

Hello... Tell me, is it possible that microcontrollers with sram 10 megabytes and more will appear soon?

[Tesla DeLorean]

Intel have chips with 10's of megabytes of memory on-board.

For STM32 it would create very large die, with large area for failure..

Might be more economic to use stacked die.

[MTS]

thank you... this task should be solved by STM engineers. Microcontrollers are not just for blink LEDs. And 1 megabyte is not enough to realize all the capabilities of the microcontroller. everything has already come to the point where it is possible to deploy a full-fledged operating system on stm32, but 1 megabyte of RAM negates everything. I think it's interesting for everyone to get not only some timers in microcontrollers, but also other features that will help solve more complex tasks.

[MTS]

and I think that the future belongs to STM and not to intel. Intel and amd are pulling into the past, but we need to go into the future.

[Tesla DeLorean]

You really don't understand the economics here.

The volume market is for low cost MCU solutions, with very high die yields from a wafer.

A product for a $5-10 MCU is *NOT* going to sustain a part that costs $100, and the market space for a $1-10 IC is substantially more massive than whatever niche application you have. And it's a battle you're not going to win, as the 1MB part is going to have a better than 10x yield than your 10MB part per wafer.

If you're crunching 10MB of numbers on a continuous basis, a general purpose MCU is probably not the best solution. You'd be doing a custom design, and you'd be doing things in a massively pipelined and parallel fashion.

>>Microcontrollers are not just for blink LEDs.

Sure, most people using 1MB or less, are doing MUCH more than that..

[Tesla DeLorean]

No, don't think so, they are much larger, have a deeper bench, and do more technically superior design work. Its generally a lot easier to pivot into doing less complicated things, than it is to build a team to do more complicated/challenging things..

With lead times pushing beyond 52 weeks, those controlling fab capacity/utilization are the ones with a future.

[MTS]

the revolution in IT costs a lot of money. Intel and amd are just commercial companies that have only profit as their goal. They don't need to give us a revolution in technology, they need us to buy what they have in warehouses for as long as possible. If they give you a new one, will you buy the old one from them? They add only a little bit of megahertz frequency and the number of cores. you won't get anything else. If STM gives today more than just microcontrollers, then it will act very smart. There are a lot of people who want not just to make blink led, but also to create serious software. and serious software requires good hardware. everyone is tired of x86. many are tired of windows and linux. people are ready to take a step into the new, only they need a platform on which they can build this new. give it to them. If STM decides to wait, then maybe someone else will come and give what they want. then you will lose.