Expected runtime of Pyra and clock rate


SneHebNor

Active Member
Joined
May 15, 2016
Messages
81
Are there any ideas, how long the battery of the Pyra will last when using it in a "light" way, i.e. not playing games, just editing text and low bandwidth wifi, i.e. not watching video streams, but doing text chatting?

1 hour? 10 hours?

Will it be possible to reduce the clock rate to improve battery life time?
(I assume that 500 MHz is enough for the things I plan to do on the Pyra.)
 
10 hours? Probably. I'm expecting power management to be subpar on launch, so while runtime at light use will probably reach double-digits, standby time will probably not be that great.
 
Short answer: It depends.

Long answer: It is a big battery. I think 2-3+ times the capacity of most cell phones? It is as large as could fit the case and the case is as large as practical for the device. The actual run time will then be dependent on usage. The only viable benchmark I can think of for estimating that is the Pandora. The Pyra is -capable- of using more power per minute, but can also do MUCH more per minute than the Pandora. So, doing the same tasks as the Pandora, it should use -less- power overall for accomplishing the same task. Still, if you're doing tasks outside of the Pandora's capabilities on a Pyra, then you should expect less run time per quantity of power. But, the Pyra's battery is about 150% of the capacity of the Pandora's so it might actually run longer still even with the greater load. At release the Pyra won't have much optimization, but that is expected to improve as the community starts kicking in input. Nobody actually has a Pyra in the wild yet, so there isn't any real world experience to reference. Phone module use will impact battery life depending on how the module gets used - always monitoring for a call will require some OS interaction so it won't be in deep standby at that point, but maybe the two smaller cores could be used for that, but nobody has built that yet but maybe someone will someday...

I understand that it can be under clocked, BUT under clocking may actually decrease battery life as the rest of the system could then be using power longer while pending on the SoC to catch up vs leaving it on full cycle and letting it accomplish it's task earlier. However, if using an application that utilizes all available processing power continuously, that could be void for your usage case. But, cycling it down manually shouldn't be necessary as the SoC should go into lower power modes between it's execution cycles if it isn't maxed out, but that may be dependent on software that may or may not exist yet.

So, I'm going to say about 1 hour full tilt with something pulling power from the USB ports, 4 hours of heavy use, 6-10 hours of surfing the web or playing a video, maybe 15-20 hours of audio playback with the screen off, 16 hours of standby with the 4G module monitoring for calls, maybe 1-2+ weeks of standby without the 4G running. Of course none of that has any basis beyond the rule of thumb and everyone's thumb is different unless you don't have thumbs but then using a Pyra without thumbs could be difficult and I'm sorry but there simply isn't a good way to make it more usable without thumbs.

And yes, there was meant to be humor in that reply.
 
I'm wondering how much power 4G LTE, 3G, Edge or GPRS connection will use when using full bandwidth for data, and how much in stand-by. And would radio data use more or less power than Wi-Fi and/or bluetooth?
If for instance GPRS or Edge use way less power, it might be worth it to switch to that network while in stand-by as those are fast enough for notifications and such.
 
I understand that it can be under clocked, BUT under clocking may actually decrease battery life as the rest of the system could then be using power longer while pending on the SoC to catch up vs leaving it on full cycle and letting it accomplish it's task earlier. However, if using an application that utilizes all available processing power continuously, that could be void for your usage case. But, cycling it down manually shouldn't be necessary as the SoC should go into lower power modes between it's execution cycles if it isn't maxed out, but that may be dependent on software that may or may not exist yet.
I did believe that was the case with the Pandora, at least since we learned about/enabled all of the CPU power modes. But my own personal experience states that since the latest round of idle power improvements, my battery lasts just under 48 hours with an hours or so audio playback per day at 1100MHz, or just over 48 hours if I underclock to 600MHz. It's not a massive difference, but for my uses lasting a complete two days then conking out at around the same time of day is a big win. I reckon it's at most a 5% difference between 600MHz and the highest reliable speed of my GPU though.
 
Will it be possible to reduce the clock rate to improve battery life time?
Sure, it should be the same as on any linux PC, e.g. you can use cpupower to set the maximum frequency.
But if the CPU is mostly idle, then (maximum) frequency hardly matters.

I'm wondering how much power 4G LTE, 3G, Edge or GPRS connection will use when using full bandwidth for data, and how much in stand-by. And would radio data use more or less power than Wi-Fi and/or bluetooth?
Generally, the newer mobile standards spend less energy per bit, but use more power at full blast because of their greater throughput. Wifi and bluetooth benefit from the shorter distances involved. For bulk data, wifi is the clear winner.

I'm not sure about standby though. I think it's going to look something like bt < mobile << wifi in terms of power expended to maintain a connection. For the different mobile standards, no idea.
 
The Pyra is -capable- of using more power per minute, but can also do MUCH more per minute than the Pandora. So, doing the same tasks as the Pandora

I thought especially about interactive usage, like IM chat, where the bottleneck is the user. A faster CPU doesn't make me think and type faster :~)

But let's see, how the Pyra will behave, when it's ready. I just hope, that one will be able to down-cycle the CPU for use cases, where CPU does not matter much (IO bound, user interaction bound).

Anyway, for mobile phones, the critical component is screen. Reducing screen brightness is probably more important than reducing clock cycles.
 
But let's see, how the Pyra will behave, when it's ready. I just hope, that one will be able to down-cycle the CPU for use cases, where CPU does not matter much (IO bound, user interaction bound).
The flip side to that is if you leave it running 'wide open', then it can process a given chunk of data then drop to low power while waiting for the rest to catch up. It shouldn't draw 'full power' at 'wide open' if there isn't anything for it to do.

Of course, we won't know much until some ship out and real world experience starts trickling back in. Then, as the system software matures, those initial experiences will loose relevance as the software develops further.
 
Back
Top