Espressif and product upgrade path

[expresspotato]

Hello,

We're using the ESP32-Pico-v3-02 for our project, to be mass produced this spring. As we look for a path for future upgrades, we're struggling with our ability to continue with the ESP32...

There doesn't seem to be something in the pipeline that has:
- Required: Dual Core (240 Mhz or higher)
- Required: WiFi 6 / 6E, on 5 Ghz / 6 Ghz
- Required: BLE + LE Audio
- Optional: BT Classic
- Optional: Dual Antenna
- Optional: Faster CPU, or CPU with hard FPU

For some reason Classic was abandoned. We could deal without Classic, if LE Audio was supported but we haven't read about any hints of that...

Seems like the product line has gone in the wrong direction...

Just my opinion, but what's the point of the C3 exactly? It's still a single core, just on a RISC-V architecture, instead of the tried and tested Xtensa ARM cores. The S3 is dual core and with Xtensa CPU's, but as the same clock and RAM as the ESP-32...

There will always be the market for smart lightbulbs, but for more serious projects like ours, people really want:
1) More internal RAM
2) Floating Point Unit
3) Faster core clock or more core count
4) More reliable software. We have found all versions of ESP-IDF are usually utterly broken in some regard, that makes it very hard to build a consumer facing product that doesn't get us skewered with bad reviews. We really wish Espressif would simply stop issuing new features for a month and fix what they have.

[MicroController]

Just my opinion, but what's the point of the C3 exactly?

Huh? Are you actually wondering what be the benefits could be of a more simple, smaller, cheaper SoC with lower power consumption?!

tried and tested Xtensa ARM cores

Uh-oh...

You do realize that RISC-V is a more modern, more efficient and likely more 'cost-effective' ISA? - You say you're concerned about upgrade paths... then RISC-V is actually what you're looking for.

Btw, you cannot equate ESP32 and ESP32-S3 just by the MHz and KB.

1) More internal RAM
2) Floating Point Unit
3) Faster core clock or more core count
4) More reliable software.

Sounds like you're planning on leaving the realm of microcontrollers.
Next step would probably be Espressif's P4, or some other single-board computer potentially running Linux: Multi-core, much more MHz, dozens of MB of RAM &c. Search the internet and you'll find several SBCs available <€10 which tick the boxes - except maybe for the wireless part, but that doesn't have to be part of a single SoC.

[expresspotato]

Huh? Are you actually wondering what be the benefits could be of a more simple, smaller, cheaper SoC with lower power consumption?!
Uh-oh...

Well perhaps lower power consumption. Smaller and cheaper? Well STM32 has some absolutely tiny micro processors (with terrible software) that are in fact cheaper as well. The price wasn't really the concern with the ESP32, it's already competitively priced. The software is generally much more comprehensive, and the development tools are very good. It would have taken us years to develop anything meaningful on the STM32, vs months with the ESP32.

Espressif's P4

Well that's an MCU without any connectivity, and is even less suited than what's out there already.

I'm sorry, my vision is not for "cheaper". It's for more relevant solutions to building real world products.

[MicroController]

Thank you for sharing those interesting views. I definitely learned something today.

[ESP_Sprite]

Well that's an MCU without any connectivity, and is even less suited than what's out there already.

Note that the intent of the P4 is for it to easily and transparently pair up with one of our C-series chips (e.g. the C3 or C6 or in the future the C5) allowing it to connect to things in the same way (from a software perspective) as our current chips do.

In general, while recently we happen to have released some C-series chips, it certainly doesn't mean we're focusing on that market alone. To illustrate this: For the P4, we put quite a bit of development time in getting faster CPU cores working (as a comparison: the P4 CPUs not only run at 400MHz but also can execute more instructions per cycle than our older RiscV cores, which already left the Xtensa cores in the dust) and we're certainly not going to waste all that development time on only one chip.

Also, on the cost perspective: I can imagine that if you make things that also contain loads of other hardware, or where the development time and work consists of the bulk of the cost, the difference between an ESP32-S3 and an ESP32-C3 is negligible. However most of our chips go into products that are made in very large numbers, and the difference there is a lot more impactful: we specifically slimmed down the -C3 into the even more lean and cheap -C2 because of customer request. So while I can imagine you're not very interested in those chips yourself, there absolutely are others who are.

[expresspotato]

And I appreciate that, you'd know this better than me, but my suspicion is companies use the ESP32 because it's an MCU, with built in WiFi, BLE, for IOT and other devices.

Pairing one ESP32 with another is much more expensive, time consuming, is more difficult to maintain, double the power consumption... etc.

We are just hoping for something with more reliable connectivity, faster clock, and potentially more IRAM.

The same offering in different incarnations isn't really a way forward for us to happier customers and better products.