5 ms·

I'm not sure I see the "revolutionary" aspect of this. With displays, you don't need to make power consumption arbitrarily small. By Amdahl's law, you only need

I'm not sure I see the "revolutionary" aspect of this. With displays, you don't need to make power consumption arbitrarily small. By Amdahl's law, you only need to make it a small percentage of the cpu (and wireless) consumption. In this respect, Sharp's transflective displays (that were found, for example, in the pebble watches,) were already resolving the power issue in the context of smartwatches.

kridsdale1 9y ago
All energy savings have value. You may argue relative significance, but a joule saved is a joule saved.
- utopcell 9y ago
  Agreed on that, but it's hard to justify the "revolutionary" tag if the end-product only increases its battery life by 10 or 20%.
  - bdamm 9y ago
    Combine this with 10 or 20% savings in CPU energy, 10 or 20% savings in radio power, 10 or 20% in RAM energy for example, and you start getting into revolutionary use cases. We're now looking seriously at embedded small devices that can sit in one place, take up the space of a pack of cards, and transmit useful data over the course of 20 years. Now we start getting into revolutionary because this opens up applications like building moisture detectors into roof construction, embedding a screen into the back side of your front door that shows a log of when it was opened or closed, or maybe something surprising like color changing shoes.
djrogers 9y ago
The revolution is in the amount of time the display can be on, and the amount of data it can display 24/7. That is currently the big knock on most smartwatches - you have to move your wrist to activate the display and see the time. If that barrier is removed, it is revolutionary.
- utopcell 9y ago
  The Sharp displays on pebbles were always on though.
- nradov 9y ago
  The displays on Garmin smartwatches are always on. However they're reflective displays so you have to activate the backlight to see them in the dark.
  - ISL 9y ago
    And the battery lasts for a week or more. Source: last charged my watch (3HR) more than a week ago; 47% remaining.
- grahamburger 9y ago
  Both of the Android (WearOS I guess now) smart watches that I've personally used have had always on displays and 2-3 day battery life. Maybe this is just an iWatch or whatever it's called problem? It does seem like whenever I see one of those mini iPhones strapped to someone's wrist it's blank.
- mikestew 9y ago
  "Most smartwatches"? I can think of only one, and that's the primary reason I have a Garmin instead of an Apple Watch. Other than Apple, I can't think of another smart watch that requires one to make a wrist motion in order for the time to be displayed.
  - brokenmachine 9y ago
    My original Samsung Galaxy Gear (that I used literally once, lol) had this "feature". You had to do a douchey motion to see the time, and it often missed your motion so you'd have to do it again. I knew I'd never actually use it but was hoping it'd be relatively hackable and I'd think of a use case for it. That never eventuated.
yjftsjthsd-h 9y ago
In phones, at least, I was rather under the impression that display did dwarf the entire SoC+radio?
Someone 9y ago
Firstly, these displays promise brighter colors and a wider gamut. Secondly, you only get long battery life in a small package one tiny step at a time, and you want them, either to increase battery life or to make smaller watches. The latter is particularly hard. A typical smart watch is around 10 mm thick, nowadays, 2 or so of which are taken by the display and the back panel. So, if you want to make it 8 mm thick, you lose 25% of inner volume and, ballpark, the same on battery volume (taking into account that the battery is shielded, likely closer to 30%) Finally, I’m not sure that transflective display is that low power if you want to frequently update the display. Looking at the data sheet of https://eu.mouser.com/new/Sharpsma/sharpmemorylcd/ https://eu.mouser.com/new/Sharpsma/sharpmemorylcd/, page 14 lists power use of 15 μW on a static display, 50 μW on a display that updates at 1 Hz. Extrapolating, I get 2115 μW at 60 Hz, or about 2 mW. The capacity of a typical smartwatch battery is less than 1 Wh, so that display could display video for 500 hours or 20 days (rounded up a few times)
- utopcell 9y ago
  Wider gamut is what might be the differentiator here. With respect to power consumption, the video codec would probably draw more than 100mW [1], and at that point, the 2mW display consumption would not matter. [1] a 640x480@15fps hardware decoder draws 71mW. https://cseweb.ucsd.edu/classes/fa11/cse291-a/pdf/02/JSSC_stacked_mobile.pdf https://cseweb.ucsd.edu/classes/fa11/cse291-a/pdf/02/JSSC_st...
  - Someone 9y ago
    The display i mentioned is 144x168. 640x480 is over 12 times as many pixels. _if_ decoding scales by #pixels, that would move that 71 mW down to around 6 mW. Still larger, more so if you go to 30 Hz or higher, but there is no silver bullet if you want a smaller watch or longer battery life.
    - utopcell 9y ago
      Even assuming a CPU that would only consume 6 mW when decoding video, and nothing else (DRAM, wireless) draining power, then (at 15fps) the display is only drawing 0.5 mW, or 1/12-th of the CPU power consumption. Even if you were to assume a display that draws no power, that would only increase battery life by 8%. A more realistic estimate would be 1% to 2%. Nothing revolutionary about that.
elihu 9y ago
Somewhere, there's an engineer saying to him/herself "we could use a low-energy CPU in this design, but there's no point because its power consumption is insignificant compared to the display".

kridsdale1 9y ago

All energy savings have value. You may argue relative significance, but a joule saved is a joule saved.

utopcell 9y ago
Agreed on that, but it's hard to justify the "revolutionary" tag if the end-product only increases its battery life by 10 or 20%.
- bdamm 9y ago
  Combine this with 10 or 20% savings in CPU energy, 10 or 20% savings in radio power, 10 or 20% in RAM energy for example, and you start getting into revolutionary use cases. We're now looking seriously at embedded small devices that can sit in one place, take up the space of a pack of cards, and transmit useful data over the course of 20 years. Now we start getting into revolutionary because this opens up applications like building moisture detectors into roof construction, embedding a screen into the back side of your front door that shows a log of when it was opened or closed, or maybe something surprising like color changing shoes.

djrogers 9y ago

The revolution is in the amount of time the display can be on, and the amount of data it can display 24/7. That is currently the big knock on most smartwatches - you have to move your wrist to activate the display and see the time. If that barrier is removed, it is revolutionary.

utopcell 9y ago
The Sharp displays on pebbles were always on though.
nradov 9y ago
The displays on Garmin smartwatches are always on. However they're reflective displays so you have to activate the backlight to see them in the dark.
- ISL 9y ago
  And the battery lasts for a week or more. Source: last charged my watch (3HR) more than a week ago; 47% remaining.
grahamburger 9y ago
Both of the Android (WearOS I guess now) smart watches that I've personally used have had always on displays and 2-3 day battery life. Maybe this is just an iWatch or whatever it's called problem? It does seem like whenever I see one of those mini iPhones strapped to someone's wrist it's blank.
mikestew 9y ago
"Most smartwatches"? I can think of only one, and that's the primary reason I have a Garmin instead of an Apple Watch. Other than Apple, I can't think of another smart watch that requires one to make a wrist motion in order for the time to be displayed.
- brokenmachine 9y ago
  My original Samsung Galaxy Gear (that I used literally once, lol) had this "feature". You had to do a douchey motion to see the time, and it often missed your motion so you'd have to do it again. I knew I'd never actually use it but was hoping it'd be relatively hackable and I'd think of a use case for it. That never eventuated.

yjftsjthsd-h 9y ago

In phones, at least, I was rather under the impression that display did dwarf the entire SoC+radio?

Someone 9y ago

Firstly, these displays promise brighter colors and a wider gamut. Secondly, you only get long battery life in a small package one tiny step at a time, and you want them, either to increase battery life or to make smaller watches. The latter is particularly hard. A typical smart watch is around 10 mm thick, nowadays, 2 or so of which are taken by the display and the back panel. So, if you want to make it 8 mm thick, you lose 25% of inner volume and, ballpark, the same on battery volume (taking into account that the battery is shielded, likely closer to 30%) Finally, I’m not sure that transflective display is that low power if you want to frequently update the display. Looking at the data sheet of https://eu.mouser.com/new/Sharpsma/sharpmemorylcd/ https://eu.mouser.com/new/Sharpsma/sharpmemorylcd/, page 14 lists power use of 15 μW on a static display, 50 μW on a display that updates at 1 Hz. Extrapolating, I get 2115 μW at 60 Hz, or about 2 mW. The capacity of a typical smartwatch battery is less than 1 Wh, so that display could display video for 500 hours or 20 days (rounded up a few times)

utopcell 9y ago
Wider gamut is what might be the differentiator here. With respect to power consumption, the video codec would probably draw more than 100mW [1], and at that point, the 2mW display consumption would not matter. [1] a 640x480@15fps hardware decoder draws 71mW. https://cseweb.ucsd.edu/classes/fa11/cse291-a/pdf/02/JSSC_stacked_mobile.pdf https://cseweb.ucsd.edu/classes/fa11/cse291-a/pdf/02/JSSC_st...
- Someone 9y ago
  The display i mentioned is 144x168. 640x480 is over 12 times as many pixels. _if_ decoding scales by #pixels, that would move that 71 mW down to around 6 mW. Still larger, more so if you go to 30 Hz or higher, but there is no silver bullet if you want a smaller watch or longer battery life.
  - utopcell 9y ago
    Even assuming a CPU that would only consume 6 mW when decoding video, and nothing else (DRAM, wireless) draining power, then (at 15fps) the display is only drawing 0.5 mW, or 1/12-th of the CPU power consumption. Even if you were to assume a display that draws no power, that would only increase battery life by 8%. A more realistic estimate would be 1% to 2%. Nothing revolutionary about that.

elihu 9y ago

Somewhere, there's an engineer saying to him/herself "we could use a low-energy CPU in this design, but there's no point because its power consumption is insignificant compared to the display".