Live data from Hacker News

Unless my phone can be a PC, I don't want to keep paying for extra performance

androidauthority.com

211–220 of 430 posts

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#211

FWIW two things have been a game changer for me in terms of mobile UX recently: - ZMK powered super portable wireless Bluetooth split keyboards - portable headset-based displays like the Viture XR pro I am now comfortable leaving my laptop at home with zero anxiety, because I know I can fairly comfortably dive back into anything I need to from anywhere with my split 36 keys, termux, obsidian and earbuds. Yes most of…

> Viture XR pro

What is the experience like? Could you spend time programming in the terminal? Would it be comfortable for several hours?

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#213

> When was the last time you felt your smartphone really couldn’t cope? A couple of weeks ago using Obsidian in an iPhone SE (3rd gen). And I've switched to an iPhone 16 and it's definitely way more fluid (even for browsing). There is clearly some kind of rat race: more raw power available, less time spent by devs in optimizations (and I guess the PRO to that CON would be less time to market). I'm not sure how to fee…

Obsidian is unoptimized. I used it for years on computer, iphone and Android, it always takes longer to open then native app.

At the core it's a capacitor app.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#214

A higher performance chip can do the things I want to do (such as encode video during a video call) while using less battery, than a lower performance chip could. That is: paying for performance, no matter how unusable that performance might seem is the same thing as paying for battery time. And I'm always ready to pay for more battery time.

That's not always the case depending on the different levers that are pulled designing these chips. Focusing on burst performance sometimes/often end up being less efficient as boosting to higher clocks require higher voltages. Process and architectural improvements that yield increases in efficiency is balanced with how much extra burst performance we can wring out of the design, which is a big part of the article's argument.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#215

A higher performance chip can do the things I want to do (such as encode video during a video call) while using less battery, than a lower performance chip could. That is: paying for performance, no matter how unusable that performance might seem is the same thing as paying for battery time. And I'm always ready to pay for more battery time.

That is actually not related to performance but the decreasing node size and related gate switching power consumption decrease. In theory companies could now just keep the heat dissipation constant and increase performance with the more advanced node sizes, but they could just decrease power consumption.

Of course what users see is that their devices are more responsive, also engineers use the new performance to provide more and more cluttered and animated GUI. So you have to have the performance.

Sigh.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#216

I find it highly annoying that these rather powerful handheld smartphone computers don't have decent port access for use with instrumentation, etc. A huge range of features comes to mind from Geiger counters, oscilloscopes, sound level measurement, light intensity, etc , etc. The potential to expand the smartphone capability is enormous yet no manufacturer has tackled it. Why not? Why aren't there multiple USB ports?…

I disagree. My phone supports UVC video input so I can plug my $25 standards compliant borescope into it, I can plug my standards compliant rme sound card into it for field recordings, and I have a bunch of input peripherals that just 'surprise work' with android now too.

You should have a look at the GATT spec for bluetooth LE and the UART service. It has never been easier to build scientific devices that rely on your phone for compute. The thing is, I think we're actually at the point where it's cheaper/ more reliable/more predictable to stick an nrf52 chip into a peripheral than try to support a physical connection to your phone - I guess from a security standpoint as much as anything else.

There are a _ton_ of scopes and stuff that sit in the prosumer space that leverage your phone. They're just not wired, and I think it makes sense for them not to be.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#217
post #210

Until recently, the availability of software updates was a limiting factor, especially for midrange phones. (Before that, phones were so slow and had so little RAM that software updates made them unusable eventually). The mid-range Samsung A52 5G now comes with 7 years of security updates and you can get it with 8GB of RAM for less than 300€. That's a pretty good deal! You just need to replace the battery every three…

You just need to replace the battery every three years or so.

If you used it as a PC, presumably in a dock attached to a monitor, keyboard, etc, then it's be plugged in most of the time and battery wouldn't get much use.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#218
post #215

A higher performance chip can do the things I want to do (such as encode video during a video call) while using less battery, than a lower performance chip could. That is: paying for performance, no matter how unusable that performance might seem is the same thing as paying for battery time. And I'm always ready to pay for more battery time.

That is actually not related to performance but the decreasing node size and related gate switching power consumption decrease. In theory companies could now just keep the heat dissipation constant and increase performance with the more advanced node sizes, but they could just decrease power consumption. Of course what users see is that their devices are more responsive, also engineers use the new performance to prov…

> That is actually not related to performance but the decreasing node size and related gate switching power consumption decrease.

Yes. Of course if a chip isn't actually more power efficient at the same times as being more performant then it wouldn't be the case. But that's almost invariably the case, at least long term (if you upgrade every 3-5 years or so).

Also, the people who make phones are usually not responsible for the most performance hungry GUI's. I'd blame heave web sites for that. App writers are going to do whatever makes enough % of people have an acceptable performance. So you need your own device to be within some margin from the average, or it will be too slow eventually.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#219

Ubuntu Edge [1] was a Linux phone proposed in 2013 that never got built, but today we have the Purism Librem 5 [2] which becomes a full Linux desktop when plugged into a monitor. [1]: https://ubuntu.com/blog/ubuntu-edge [2]: https://puri.sm/posts/a-librem-5-video-made-on-a-librem-5/

The librem 5 looked interesting in 2020, now ... The product page has goals listed and the 20h idle time has yet to be reached :(.

Re: Unless my phone can be a PC, I don't want to keep paying for extra performance

#220
post #210

Until recently, the availability of software updates was a limiting factor, especially for midrange phones. (Before that, phones were so slow and had so little RAM that software updates made them unusable eventually). The mid-range Samsung A52 5G now comes with 7 years of security updates and you can get it with 8GB of RAM for less than 300€. That's a pretty good deal! You just need to replace the battery every three…

Have there been compelling use cases for on device LLMs where the value isn’t based on privacy benefits yet?
Post reply on HN