Page 8 of 40

Re: What would you like to see in The Next Chip?

Posted: Sun Sep 24, 2017 1:35 pm
by halfro
loboris wrote:
halfro wrote:1. Well radio implementations such as LORA for LPWAN, nothing more than that for in my side of the world(Kenya). It can even be an off chip addition that can be done as a module that one can buy in bulk from a single supplier. I have done deployments for such and the ESP32 is a real contender if this was taken up. The advantage is actually obtaining this from a single supplier.

2. The ability to get the WROOM32 with a uFL connector that one can add an eternal antenna in the case where one is using a product case.
If nothing else, than you already have such a board and module satisfying both of your requirements:
Pycom's LoPy (specs)
L01 OEM module (specs)
Thank you for this, especially on the L_01 module.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 12:15 am
by Leitukey
Attempts at clever quips by an Espressif representative regarding an obvious fundamental problem with Espressif products do not warm the cockles of my heart. No. They do not. Instead, they make my blood run cold.

Espressif seems like a typical tech company that has created some remarkably innovative and popular products yet, unbeknownst to its key investors, is rapidly heading towards insolvency. Typically, arguing against spectacular success is as futile as "yelling into a hurricane." But I will attempt that very thing.

I would much rather see the ESP32 remain as is, with no new significant features added for say six to twelve months, and instead see the ESP32 become a rock solid platform. Apparently the chances of that are currently extremely remote.

Engineers and engineering clearly dominate the Espressif corporate culture. To many this seems cool, fun and most importantly "The Right Way". After all Espressif is a classic disruptor which is following the path laid out in a pseudo-holy tome reverently adhered to by its worshippers which is known as the, "Innovator's Dilemma." Right? Therefore, what's not to like? Although he coined the currently popular phrase, Harvard professor Clayton Christensen did not invent the concept of disruptive innovation. It is not a new concept at all.

As most Westerners and followers of Abrahamic faiths around the world know, David famously slew Goliath. In part he did so by implementing an existing technology, a sling, in an innovative manner. But David was able to not merely transform (that is, to be a disruptor) he became the King David we know today because he actually successfully consolidated his power and then transferred that power to his son Solomon.

Moving from transformation to consolidation might seem natural and easy but in fact typically it is neither. Moses was the great law giver who brought the Jewish people out of slavery in Egypt and into the desert but it was his student Joshua who was tasked with leading the Jewish people out of the desert and into the land of Israel (the Promised Land). Moses was a transformative leader whereas Joshua was a consolidating leader. Both were necessary, but neither was properly suited to perform the task of the other.

Espressif's key investors are perched atop a thorny dilemma which is common for successful startup companies: the transformative CEO who "made the company what it is" is usually the wrong person to "make the company what it could be." If you read the headlines of tech websites and blogs carefully you have probably noticed that every few months a well-known tech company CEO fights angrily, vociferously, but usually unsuccessfully to keep his job.

More to the point, Espressif faces a typical yet pernicious problem which virtually all new and successful tech companies face: its merry band of innovative engineers are naturally obsessively creating new features while paying scant attention to actual user benefits for the majority of its potential customers and potential customers' customers (end users).

Espressif is a darling tech company to legions of makers/hackers right now because it has created some remarkably innovative products. And I guess Chinese companies that make toys and trinkets actually ship products with the ESP8266 and ESP32. But if Espressif does not shift its focus soon from cranking out "bright new shiny objects" to building "stuff that works reliably" it will soon be crushed by competitors.

The most important thing key Espressif investors can do right now is to replace the current Espressif CEO with a CEO who knows how to properly satisfy customers with benefits, instead of one who is myopically focused on churning out unreliable features at breakneck speed.

What is my stake in this discussion? I want to develop a commercial product. I have used PIC chips (microcontrollers made by Microchip) before. They were very reliable. Very reliable. But compared to the ESP32 PIC chips and other similarly mature microcontrollers are remarkably limited, expensive, and difficult to program.

Yet the idea of shipping a product to customers that contains the ESP32 actually keeps me up at night because despite its remarkable potential value, I shudder at the thought of selling customers a product that seems to work at first, but gradually is shown to be unreliable.

See, I have been involved in a product recall before that was the direct result of embracing a new wireless hardware platform that seemed to work properly but turned out to be unreliable. The headaches, hassles, financial costs, and loss of goodwill from customers was something I remember vividly.

The ESP8266 was a little too limited. Therefore, creating the ESP32 therefore made sense. But the ESP32 is remarkably powerful. It will suffice for myriad IoT devices for many years to come. Espressif as a company should take a break for, say six to twelve months, from adding features to the ESP32. Instead of adding new features, making the ESP32 rock solid should be the number one priority for Espressif's engineering team right now.

Naturally, the innovative engineers at Espressif (including Espressif's current CEO) will resist this. Engineers—particularly clever ones who build innovative products—typically love development but loathe maintenance.

That is why Espressif needs a new CEO.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 1:01 am
by Lucas.Hutchinson

You do realise this thread is asking about community suggestions for the next espressif chip?
E.g. the ESP64 or something like that? As in the product to replace this one?

What you are talking about is in reference to the IDF which, as they have stated before in this thread, is developed in parallel and separately to the actual silicon.

So the result of this discussion will likely not see the light of day for another 12 or more months at a minimum.

So in regard to the IDF, yes there are bugs and it could be more stable. But has almost nothing to do with what features future silicon will have.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 1:21 am
by ESP_Sprite
We are aware that there are bugs in the ESP32 silicon; the ECO document is proof of the fact that we're open about them. Making a new chip, however, does not mean we'll be throwing away all our old designs and starting from scratch again: a lot of the IP has migrated from the ESP8266 to the ESP32, and a lot of IP will migrate from the ESP32 to whatever new chip we'll be making next. In the mean time, we will be fixing all the bugs in these peripherals we find in the current revision of chip so we can offer a more bug-free experience in the new chip.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 1:29 am
by Leitukey
charma wrote: Personally I'd like to see an external memory interface. It could be slow (multiplexed bus with external transparent latches) like the PIC has, but that would be an easy way to attach memory and memory-mapped devices where speed isn't critical but you don't want to go through many GPIO expanders to mimic a parallel bus.

At present, I assume that asking you guys to cram more memory on the silicon is unrealistic. Besides, slow, external memory will very likely suffice for almost all of my intended needs.

Eventually (say in a year or two) I would like to see that feature added. For now, however, I would prefer that the ESP32 become rock solid stable instead of equipped with new functionality.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 1:33 am
by Lucas.Hutchinson

Have you seen anything about the wRover modules?
They include 4MB of external PSRAM. Is this what you would be looking for?

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 2:37 am
by Leitukey
Lucas.Hutchinson wrote:@Leitukey

Have you seen anything about the wRover modules?
They include 4MB of external PSRAM. Is this what you would be looking for?
Yes. I have you seen the ESP32-WROVER.

I intend to develop in MicroPython but I want MicroPython to keep approaching Python until it finally gets there. 4MB is a nice start. Eventually (but not now) I would like to be able plugin additional RAM to the ESP32 (either actually as a USB device or by soldering it directly on the PCB) like I can plug in an external storage device (such as a hard drive or USB stick) into my laptop.

I want to reiterate: this is a low priority feature for me, one that I will gladly wait one to two years to see. For now, rock solid reliability of the ESP32 is the most important feature to me because it would make development go much faster and easier. And more importantly it would allow me to provide customers with a reliable product that "just works."

I have spent many unpleasant hours on tech support calls with justifiably frustrated (and occasionally even angry) users that would been avoided had the product in question worked properly.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 2:59 am
by Leitukey
WiFive wrote:Seems as though some people want low power low cost chip for battery powered nodes and others want a workhorse with audio/video features. Sounds like 2 different chips.
Going from scope creep to product creep would be a terrible idea.

Esperrif should stick with "small, weak, and cheap" at this point. "Big, strong, and expensive" is territory for mature corporations with deep pockets (much money). If for example, Samsung were to see Esperrif as a serious competitor, they would likely work to put Esperrif out of business.

If the executives at Samsung are even aware of Esperrif, they probably do not consider the ESP32 to be a serious competitor of theirs.

Watch forty seconds of the following interview with Sir Robin Saxby from the point I linked to. He basically explains how ARM employed a strategy very similar to that which Esperrif is employing.

Sir Robin Saxby, first ARM CEO and Chairman

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 3:06 am
by Leitukey
EndlessDelirium wrote:There is little benefit in a new chip with an even broader feature set if Espressif keeps on failing to deliver on the SDK side. Take a closer look at the ESP32. The feature set of that chip already is rather awesome. But many of those features are either hard to use or simply broken.

So please do us the favor and invest into bugfixing/stabilizing/enhancing the ESP IDF and a new silicon revision instead.

At least in this case, @EndlessDelirium is not endlessly delirious. He is eminently practical. Please heed his suggestions! Please put all of the feature requests in this thread on hold for now. We do not need them until the ESP32 is rock-solid stable.

Re: What would you like to see in The Next Chip?

Posted: Mon Sep 25, 2017 3:16 am
by Leitukey
permal wrote: ....I too would like to see the IDF on the ESP32 feature complete and stable before a new chip...

Please do not quibble. We are in a small minority who will likely lose this argument against the sea of feature-loving makers/hackers (which, please remember, likely includes the CEO of Esperrif) who keep posting their "wish lists" in this thread.

Please muster your arguments for this one particular salient point for now, "...I too would like to see the IDF on the ESP32 feature complete and stable before a new chip..." Please back that assertion with as many powerful arguments as you care to make.

If we do happen to win this argument, then feel free to quibble.