It could be that they just use the same processor for multiple products even if it is overpowered to be able to reduce cost in acquiry, hardware design and software development.
This gets said all the time, but what do people expect? Chips that can run doom are basically the bottom tier chip now. There is no point making lower power chips because they would use the same physical resources to produce.
Do you find it odd that most people buy chairs capable of supporting 200kg when they know that no one over 100kg will ever sit on it?
The latter is overengineering for safety, using a fast chip isn't. But I imagine it's the cheapest, small SoC that offers bluetooth or wifi, or whatever a "smart" lamp needs. If someone wants to rail against the need for smart lamps, however, they'd have a point.
A Cortex-M33 microcontroller costs like a coffee, and much less that that for Ikea that can then sell more lamps and at a higher margin, because people like introducing more complications in their lives, like software updates for what used to be a wire and a switch.
> Do you find it odd that most people buy chairs capable of supporting 200kg when they know that no one over 100kg will ever sit on it?
it solved through more materials or technologies per $ input. Chairs that can support more mass require non-trivial production and/or rare materials. You likely can produce a chair out of titanium or carbon fabric that could support more weight.
There're no chairs that are both cheap and can support variety of use cases.
That is not really accurate. There are plenty of chips being made that are less powerful. Such as the Epson S1C60 which is a 4-bit MCU capable of running at a whopping 1MHz frequency
Chairs are given large safety margins, and designed to accommodate a wide range of people. Chips also have safety margins, but chips are not chairs, and there are other forces causing overpowered ARMs to be put into simple electronics, namely prevalence and mass production bringing the price down to pennies, so much so that it costs far less to develop some high level code for an cheap overpowered ARM chip than develop some more specialised embedded code for more "appropriately" sized microcontroller for the job. Sometimes things will still sway back to the specialised chips when things like power constraints or more specific hardware features carry enough weight for the purpose.
It's not particlarly powerful by modern standards. The processor in question is an ARM Cortex M33, which is available with single or dual 32-bit cores, at clock speeds of 150 to 240 MHz. But when Doom was first released, a typical PC would only have run at 66 MHz.
You might not need that kind of processing power continuously, but consider that the lamp has to connect to WiFi securely, then talk to your smart home HUB securely, meaning setting up TLS connections, and you want it to respond to commands in much less than a second, you want that kind of processing power in bursts.
We do not need a processor this powerful to do this at all. Garage door openers have been around for ages and the only novel thing here is the strength of the encryption, which could be done in hardware.
It's simply that this is "easy" because these chips are plentiful and cheap enough.
The stories of Russia harvesting chips from washing machines and dryers to put into munitions systems[1][2] demonstrates just how much "excess" there is in terms of compute power in the average smart device.
And yet i purchased a number of LEDs with mr16 gu5.3 bulbs just like the photo from TFA, and none of them work well in the minimum dimmer settings, they're all way too bright.
Maybe I've got it all wrong and i should have played doom on them instead.
You want your "smart" lamp to respond quickly to reduce power consumption and air time. Just a short sprint once is a while instead of a slower 8 bit cpu crunching on algorithms designed around 32 bit operations. The wider ALU and registers are cheap. The fetch and decode logic is almost identical in area and uses less power because you can implement your application executing fewer instructions. Have a look at the code size produced by common C compilers for something like AES128 targeting a Cortex M0+ (or M33 in this case) and a ATmega. You'll see lots of short helper functions that depending on compiler settings are either called very often, inlined or even inlined and unrolled.
Do you find it odd that most people buy chairs capable of supporting 200kg when they know that no one over 100kg will ever sit on it?
Now just a moment, I'm 6'2, with a long torso, and 100kg is literally unhealthy for me, unless I decide to lose muscle mass.
(No, other 6'2 people, I don't care what your healthy weight is. I said long torso. Most 6'2 men are 8" shorter when sitting beside me. The torso is the heaviest part of the body...)
BMI is a poor tool, especially shit to anyone who builds muscle. Bad tool, the metric measurement is only valid to the 'middle sized sedentary lifestylers'
It's also a poor tool for very tall people, since it divides by height squared, even though our bodies are obviously 3D objects. I see some research that supports using height^2.5 as a divisor:
I think people focused on the wrong metric, the BMI. They should have focused on the self-diagnosis that at 6'2" (1.88m) it would be "literally unhealthy" for you to weigh less than 100Kg. Even if you were a special case, like a body builder, to illustrate the weight we're talking about, Arnold Schwarzenegger was 6'2" and 107Kg [0] in competitions.
BMI really is a poor indicator if your body shape is far from average size, and no amount of showing off your ignorance in the topic here is going to change that.
I am 188cm, and I am firmly in overweight part of the scale according to general BMI graphs you can find anywhere. But we talk about a person who has cca 8% body fat according to various measurements taken (which is normal amount for sporty folks, and I am relatively extremely active).
The thing is, I have very wide chest and shoulders far apart, also I think my overall torso length is above average but never checked that. That chest part is really bone/ribcage structure, not tons of muscles on top of that.
I am not complaining, putting on a bit of muscles and straightening my stance had make me look like second Arnold when wearing t-shirt, at least for chest part.
In the same vein, my legs are relatively thin (both calves and thighs), so my projected contour is far from some average male guy. BMI categories never worked on me.
Overweight (BMI 25~30) is statistically fine, even lower mortality than “normal” (18.5~25) BMI [0]
> Conclusions and Relevance
> Grade 1 obesity overall was not associated with higher mortality, and overweight was associated with significantly lower all-cause mortality.
BMI is a vague measure in the first place, and the ranges are pretty arbitrary, but even accounting for that only the “obesity” range is related to health issues.
> Most 6'2 men are 8" shorter when sitting beside me
I know people don't have perfectly proportional limbs and body parts across the population but a 6'2" man is a 6'2" man no matter what length of torso they sit next to. And 1Kg of wool weighs just as much as 1Kg of iron.
I agree that BMI is very vague. But at 6'2" (~1.88m) being under 100Kg can't be "literally unhealthy" unless you're a special case like a body builder. There's no scientific basis for the statement and no doctor will ever say such a thing. Definitely no doctor will diagnose someone with "long torso" that can justify that weight unless it's a severe deformity. Perpetuating such nonsense is bad because it encourages some people to justify bad eating habits with nonsense pseudo-science.
To illustrate what kind of weight we're talking about, Arnold Schwarzenegger was 6'2" and 107Kg [0] in competitions.
> But at 6'2" (~1.88m) being under 100Kg can't be "literally unhealthy"
Healthiness isn't just a height and weight though, and there is so many other things going on.
People in these discussion have an image of a "standard" body type, with an "average" type of life, but realistically nobody actually fits that average. From my point of view I, I'm not even sure where the parent poster lives, I assume they are male, but have no idea of their ethnicity, nor age, nor what their life looks like. Depending on their exact circumstances, losing more than 10% of the body weight could have more detrimental effects than benefits (it could be "literally unhealthy")
> Arnold Schwarzenegger was 6'2" and 107Kg [0] in competitions.
Imagine he loses some muscle and gets 10kg of fat here and there. That would put him at probably 20% body fat perhaps ? He'd look sloppy as hell, but I don't think it would be a morbid body in medical terms.
What amuses me is that I remember a time where only “powerful” computers could run doom in full screen, whereas others had to settle with playing Wolfenstein 3D, and now Doom is the smallest game we use for these hacks :-)
>The system runs a downsized version of Doom that requires less RAM.
You could even say it's a...light-weight DOOM.
I'll see myself out.
(In all seriousness though, super impressive work! It always amazes me what computing power actually resides in what we deem to be "simple" devices these days - in some ways it feels like computing power gone to waste.)
That's easy, one. But you also have to pay for the scrum master, product owner, bti engineer, test engineer, architect, and 7 other roles that keep the software engineer from rolling out the update ;).
I actually felt a little bad when I electrified my son's play kitchen to add lights and went for an Arduino Nano. That's a bunch of lights along with some programming (microwave timer and state machine, dimming when idle, undimming upon any input, PWM brightness control for the burners) running on a 16 MHz µC, which is most likely overkill for that. On the other hand it was cheap and programming it was accessible.
The interesting thing is that the microcontroller in the Nano - Atmega328P - is really a massive overkill but for a wholy different reason: price.
Even back in 2015 the AVR based Atmega328P was more expensive than quite a few Cortex M0 based STM32 offerings (when comparing price for same quantities of course). And those MCUs had literaly by an order of magnitude bigger SRAM and the ARM core and STM32 peripheries vere obscenly more capable and I cannot stress this enough.
Now the difference is even bigger and AVR based MCU are expensive almost-unobtanium for legacy designs.
Calling AVR MCUs cheap is my trigger since 2015 when I and my collegues were browsing Farnell eshop looking for a cheap MCUs for various gizmos we were working on.
The MCH2022 badge will have a (downsized) version of DOOM running on the ice40 FPGA (implementing a RISK-V core) and a RP2040 (just for reading the buttons).
Yes, and the ESP32 too. Seems this is going to be the first badge with three processors that all can run Doom. On the MCH2022 badge, the RP2040 has no direct connection with display (except for regulating the back-light), so you would have to implement some bridge on the ESP32 or the FPGA. See block diagram on https://bodge.team/docs/badges/mch2022/
Interesting to see the manufacturing similarities between that and another smart bulb I tore apart a few months ago. Same cutout for the antenna, which sticks out the front. Same aluminum backed circuit board for heat dissipation, and same strange through hole socket for connecting the power, data, and lighting boards.
66 comments
[ 3.5 ms ] story [ 143 ms ] threadhttps://news.ycombinator.com/item?id=27498789
Many of the links are dead. The technical details:
https://web.archive.org/web/20210615035229/https://next-hack...
Do you find it odd that most people buy chairs capable of supporting 200kg when they know that no one over 100kg will ever sit on it?
it solved through more materials or technologies per $ input. Chairs that can support more mass require non-trivial production and/or rare materials. You likely can produce a chair out of titanium or carbon fabric that could support more weight. There're no chairs that are both cheap and can support variety of use cases.
I guess you never had sex on a chair.
It's simply that this is "easy" because these chips are plentiful and cheap enough.
The stories of Russia harvesting chips from washing machines and dryers to put into munitions systems[1][2] demonstrates just how much "excess" there is in terms of compute power in the average smart device.
[1]: https://www.techspot.com/news/94589-russian-tanks-using-chip...
[2]: https://www.appropriations.senate.gov/hearings/a-review-of-t...
It is a fairly pricey choice for what it's doing...$5-6 per piece for single unit purchases.
Maybe I've got it all wrong and i should have played doom on them instead.
Now just a moment, I'm 6'2, with a long torso, and 100kg is literally unhealthy for me, unless I decide to lose muscle mass.
(No, other 6'2 people, I don't care what your healthy weight is. I said long torso. Most 6'2 men are 8" shorter when sitting beside me. The torso is the heaviest part of the body...)
Yet now I am unwelcome in your home?!?
It's hard enough buying shirts, and now this.
https://www.medicalnewstoday.com/articles/265215
As I mentioned an unusually long torso, and you ran to a metric which ignores torso size, your statement, and follow up comments, makes no sense.
[0] https://en.wikipedia.org/wiki/Arnold_Schwarzenegger#Bodybuil...
I am 188cm, and I am firmly in overweight part of the scale according to general BMI graphs you can find anywhere. But we talk about a person who has cca 8% body fat according to various measurements taken (which is normal amount for sporty folks, and I am relatively extremely active).
The thing is, I have very wide chest and shoulders far apart, also I think my overall torso length is above average but never checked that. That chest part is really bone/ribcage structure, not tons of muscles on top of that.
I am not complaining, putting on a bit of muscles and straightening my stance had make me look like second Arnold when wearing t-shirt, at least for chest part.
In the same vein, my legs are relatively thin (both calves and thighs), so my projected contour is far from some average male guy. BMI categories never worked on me.
> Conclusions and Relevance
> Grade 1 obesity overall was not associated with higher mortality, and overweight was associated with significantly lower all-cause mortality.
BMI is a vague measure in the first place, and the ranges are pretty arbitrary, but even accounting for that only the “obesity” range is related to health issues.
[0] https://jamanetwork.com/journals/jama/fullarticle/1555137
Height isn't though.
> Most 6'2 men are 8" shorter when sitting beside me
I know people don't have perfectly proportional limbs and body parts across the population but a 6'2" man is a 6'2" man no matter what length of torso they sit next to. And 1Kg of wool weighs just as much as 1Kg of iron.
I agree that BMI is very vague. But at 6'2" (~1.88m) being under 100Kg can't be "literally unhealthy" unless you're a special case like a body builder. There's no scientific basis for the statement and no doctor will ever say such a thing. Definitely no doctor will diagnose someone with "long torso" that can justify that weight unless it's a severe deformity. Perpetuating such nonsense is bad because it encourages some people to justify bad eating habits with nonsense pseudo-science.
To illustrate what kind of weight we're talking about, Arnold Schwarzenegger was 6'2" and 107Kg [0] in competitions.
[0] https://en.wikipedia.org/wiki/Arnold_Schwarzenegger#Bodybuil...
Healthiness isn't just a height and weight though, and there is so many other things going on.
People in these discussion have an image of a "standard" body type, with an "average" type of life, but realistically nobody actually fits that average. From my point of view I, I'm not even sure where the parent poster lives, I assume they are male, but have no idea of their ethnicity, nor age, nor what their life looks like. Depending on their exact circumstances, losing more than 10% of the body weight could have more detrimental effects than benefits (it could be "literally unhealthy")
> Arnold Schwarzenegger was 6'2" and 107Kg [0] in competitions.
Imagine he loses some muscle and gets 10kg of fat here and there. That would put him at probably 20% body fat perhaps ? He'd look sloppy as hell, but I don't think it would be a morbid body in medical terms.
2 is used commonly but only because it used to be easier to compute.
Also, for people with large muscle mass, bmi will show overweight or obese for people that have normal (or even low) fat mass.
One time I had to gently tell a friend that he was exceeding the weight limit of the IKEA kneeling chair he was sitting on.
Really awkward for the both of us.
There are people that weight over 200kg, why draw the line at 100?
:-)
You could even say it's a...light-weight DOOM.
I'll see myself out.
(In all seriousness though, super impressive work! It always amazes me what computing power actually resides in what we deem to be "simple" devices these days - in some ways it feels like computing power gone to waste.)
Even back in 2015 the AVR based Atmega328P was more expensive than quite a few Cortex M0 based STM32 offerings (when comparing price for same quantities of course). And those MCUs had literaly by an order of magnitude bigger SRAM and the ARM core and STM32 peripheries vere obscenly more capable and I cannot stress this enough. Now the difference is even bigger and AVR based MCU are expensive almost-unobtanium for legacy designs.
Calling AVR MCUs cheap is my trigger since 2015 when I and my collegues were browsing Farnell eshop looking for a cheap MCUs for various gizmos we were working on.
RISC-V.