Loading...resuming....why?

[Ankit Nawlakha]

This discussion gives birth to yet another question!
Is the "Loading and resuming" the real problem in windows phone?? Or is it just the poor quality of apps??

 

[a5cent]

This discussion gives birth to yet another question!
Is the "Loading and resuming" the real problem in windows phone?? Or is it just the poor quality of apps??

Like I said, it's not an either/or question. It's BOTH (see post #10).

 

[Ankit Nawlakha]

Most answers here are wishful thinking...

Honestly, it's complicated. It's not just one thing, but at least a dozen factors which combined, noticeably contribute to application launch and resume times. Some factors always have a role to play. Some factors depend on how the last app you were using is programmed. Some factors depend on how the app that you launched/resumed is programed. Some factors may or may not apply depending on how you use your phone:

Average seek/read times from NAND storage, RAM bandwidth, CLR initialization time (a big factor), the amount of work done by the apps going through state transitions to and from being suspended, the time it takes to tombstone an app when the RAM requirements for a newly launched app can't be satisfied, whether or not an app supports fast resume, whether the app utilizes the WP8.1 APIs or the 8.0 APIs that influence memory resource management, the size of the app you are launching, etc etc etc.

Most of this stuff won't make much sense to you unless you're a software engineer. Not that it can't be explained, but I don't know how to explain it to non-software-engineers without writing a book.

WP8.1 should have the affect that we see the 'resuming' message less frequently, AFTER all apps are updated to use the WP8.1 APIs, but when we do encounter the 'resuming' message, the device will behave pretty much the same as it does today.

Hmm! Maybe You ar... No! Not Maybe! You are absolutely right!
But What can Microsoft do regarding this?

 

[a5cent]

Hmm! Maybe You ar... No! Not Maybe! You are absolutely right!
But What can Microsoft do regarding this?

I brushed over some issues related to memory management that I don't want to explain, as it would take far too long, but MS has already taken some steps in WP8.1

Simon mentioned .Net native. I think that will represent the single largest step MS can take towards reducing launch/resume times, as it will make WP apps much smaller and completely eliminate CLR initialization time. I just don't know when and to what degree this will start being used in WP apps.

For everything else, it's all about attacking the issue in little steps from all sides. Limiting the amount of time apps may run while suspending (or waking up from being suspended) would be one option. MS would just measure suspend and wake-up times and not publish those apps to the store that take longer than desired. Stuff like that...

 

[Moiz Mian]

I brushed over some issues related to memory management that I don't want to explain, as it would take far too long, but MS has already taken some steps in WP8.1

Simon mentioned .Net native. I think that will represent the single largest step MS can take towards reducing launch/resume times, as it will make WP apps much smaller and completely eliminate CLR initialization time. I just don't know when and to what degree this will start being used in WP apps.

For everything else, it's all about attacking the issue in little steps from all sides. Limiting the amount of time apps may run while suspending (or waking up from being suspended) would be one option. MS would just measure suspend and wake-up times and not publish those apps to the store that take longer than desired. Stuff like that...

So does windows phone 8.1 support .net native. Because if it does, that's big news, and I would expect Microsoft's apps should all be built this way

 

[prune_face]

Can anyone explain why this cannot be fixed so apps load like iOS. Developers...Microsofties hanging around...Anyone at all?
It's really frustrating when even the start screen experiences Loading....and resuming...
Funny anecdote: In central London I saw a sign that said no Loading in this area at any time, I pulled out my phone to load up an app and was met with the dreaded word ...loading...
Tldr WP turned me into a criminal

unpin ur cortana from start screen..

 

[vitorrloureiro]

Ok, but nobody really answered. Why that is not the case with iOS? And as you guys said, its not just about the developers. An iPhone 4S is far more faster than my Lumia 720 on resuming and loading apps. How can it be?

 

[Squachy]

Yes. Yes it will, Firmware not only gives new features but it provides drivers to the OS, instead the OS has to provide emulated drivers which affect battery life and phone performance.

Think of it when you install an OS on a computer something might be buggy with the software or there is a constant error between a piece of hardware and software. Updating the BIOS for the motherboard and getting drivers for your computer is essential to make it run as efficient and cause less head-aches in the future.

When installing an OS for the first time you will see drivers like such (Microsoft Device Association Root Enumerator) that are emulating the driver as best it can or until your install or MS installs the correct associated driver.

Since Windows phone runs the same kernel as its desktop and RT brothers, it shares the same logic.

In short yes, firmware will make the device run like its supposed to when we hit launch, its now a wait and see method.
Don't believe me then wait and see.

Its not going to make it go away, but its likely it can help reduce the issue further because of the firmware is optimized for the OS.

Aside from having fast app resume implemented in the app, and the cyan update, I think at this point throwing more power at the problem is probably the only real solution.

 

[a5cent]

Its not going to make it go away, but its likely it can help reduce the issue further because of the firmware is optimized for the OS.

Aside from having fast app resume implemented in the app, and the cyan update, I think at this point throwing more power at the problem is probably the only real solution.

Nope. Read my earlier posts.

 

[a5cent]

Ok, but nobody really answered. Why that is not the case with iOS? And as you guys said, its not just about the developers. An iPhone 4S is far more faster than my Lumia 720 on resuming and loading apps. How can it be?

I kind of mentioned it....

Because iOS obviously doesn't have to initialize a CLR for each app. That is the biggest difference. IOS has no runtime at all in fact, leading to a number of performance advantages, during startup and afterwards (like the executable being much smaller, not requiring a garbage collector, etc).

 

[Josip Zeljko]

I find this situation we're in funny if not tragic. One of the first things on the list of new features was improved memory management. There was even talk how LUMIA 520 8.1 will feel as fast as LUMIA 920 8.0

 

[a5cent]

I find this situation we're in funny if not tragic. One of the first things on the list of new features was improved memory management. There was even talk how LUMIA 520 8.1 will feel as fast as LUMIA 920 8.0

I think that is a bit overly dramatic. There are advantages to all these things as well... just not performance related.

Also, you shouldn't confuse random talk on blogs or forums with a promise from MS. I'd guess less than 5% of the technical statements made on blogs or forums are accurate. It's mostly rubbish really.

Finally, WP8.1 does offer improved memory management. However, if you aren't a developer that has looked into the specifics of what those changes are, you shouldn't assume to understand the implications. It's complicated, any single scentence explanation is guaranteed to be garbage.

 

[AlastorX50]

I think you are fundamentally misunderstanding something. Looking at a call-stack won't tell you much about performance. I'd agree with Simon.



I'm very sceptical. OEM provided firmware includes display drivers, camera drivers, OS configuration files, sensor drivers... stuff like that. That is all entirely irrelevant in regard to launching/resuming times!

All the drivers that interface with the SoC (CPU, RAM, GPU, etc) are built by MS/qualcomm for all devices, and those drivers are shipped together with the OS. They are already included in the WP8.1 preview for developers. Nothing is being emulated. I think you are taking experiences you made on Windows and expecting them to apply equally to WP, but that isn't true.

Anyway. I see no reason why OEM provided firmware should have any influence on launch/resuming times whatsoever. The issues I listed in post #10 are far more relevant.

The main reason why we should encounter the resuming message less frequently is because WP8.1 will collect more precise information on app memory usage. On WP8.0 the OS just assumes memory usage = maximum memory cap, which is almost always wrong. WP8.1 is expected to tombstone apps less frequently as a result. Under WP8.1 fast resume is also considered the default approach, so apps developed for WP8.1 are assumed to support that, whereas on WP8.0 that wasn't true. But as I said in post #10, many issues are involved.

Simon mentioned .Net native, which will make a huge difference once widely adopted on WP, but I don't know if WP8.1 is the platform on which .Net native will achieve widespread adoption.

http://www.windowscentral.com/lumia-cyan-and-what-it-brings-windows-phone-81

 

[EspHack]

because we got tons of lazy devs who cant update apps accordingly to make them open instantly, just see the facebook's messenger app, it opens instantly exactly were you left it, the music app now does almost the same, then you got the battery saver app which is a tiny core app and it ALWAYS shows a "resuming" screen

 

[Ryken100]

I'm no expert so take this with a grain of salt, but on WP 8.0 at least I believe the biggest contributing factor to an app's resume time had to do with Silverlight restoring the app's UI layout. It never seemed to keep the app's layout in memory, either that or it would always re-measure and re-arrange the layout at resume, which takes time. I believe that's why with some apps you would always see "resuming..." even if you only leave the app for 2 seconds, because Silverlight would discard its layout every time you leave it. It's been like this since 7.5, and I don't think it has anything to do with whether or not devs enabled fast resume ("fast resume" in WP terms doesn't refer to resume time last time I checked).

Of course this would affect apps with more complex layouts. If you have an app that always shows "resuming..." when you return to it, try going to a page with a simpler layout (like the settings) and it would probably resume much faster. Generally, if you tap on an element that's supposed to take you to a new page in an app, and that page takes a second or two to appear instead of appearing instantly, then you will probably see the "resuming..." screen if you try to return to the app while it's on that page.

I haven't tested many WP 8.1 apps besides my own, but I know that behaviour still exists in WP 8.0 apps on 8.1, and it's worse now since as WP 8.1 has been very trigger-happy when it comes to killing app processes after a short period of non-use, even if all you did was lock the screen. This should be less of an issue as more apps are upgraded to WinRT (and maybe Silverlight 8.1?), since it uses much less memory so apps are less likely to be terminated to free up resources. I also haven't noticed my WinRT app struggling to restore its layout on resume, so maybe that's been fixed? Or maybe WinRT's always has to restore the layout at resume as well, except that it's much more efficient at doing it.

Don't forget to take this with a grain of salt!

 

[a5cent]

Lumia Cyan and what it brings to Windows Phone 8.1 and the Lumia 930 | Windows Phone Central

The only thing I found in that article related to this thread was the following:

After spending 24 hours with the Lumia 930 I can say that Windows Phone 8.1 – and specifically launching apps – I feel slightly less lag than devices without Lumia Cyan. The difference is subtle, not in all apps and presumably has to do with optimizing the display and animations for the OS.

It is possible that apps which must initialize certain hardware, like the camera or motion sensor, can now do so more efficiently thanks to Cyan. But that isn't what we discussed here.

The author states that the perceived performance boost has some relationship to launch-animations and display technology. That is a somewhat strange statement to make however, because display technology has absolutely no impact on app launch times. That would be like saying Microsoft Office will launch faster on my PC, if I were to hook up a faster monitor. That is of course rubbish.

Based on the authors description, I think it's most likely that he's just being presented a faster launch-animation, which can provide the illusion of some apps starting up faster (particularly for those apps that have always launched in less time than it took to show that animation). If we were to measure app launch times with a stop watch, particularly those that aren't directly tied to hardware sensors, I'm sure we'd measure no difference at all.

Anyway, alone the fact that the improvement doesn't apply to all apps means that whatever the cause is, it isn't related to any of the things we discussed here. It's not a general, across the board improvement, because firmware doesn't include anything that can achieve that.

 

[AlastorX50]

Yes. Yes it will, Firmware not only gives new features but it provides drivers to the OS, instead the OS has to provide emulated drivers which affect battery life and phone performance.

Think of it when you install an OS on a computer something might be buggy with the software or there is a constant error between a piece of hardware and software. Updating the BIOS for the motherboard and getting drivers for your computer is essential to make it run as efficient and cause less head-aches in the future.

When installing an OS for the first time you will see drivers like such (Microsoft Device Association Root Enumerator) that are emulating the driver as best it can or until your install or MS installs the correct associated driver.

Since Windows phone runs the same kernel as its desktop and RT brothers, it shares the same logic.

In short yes, firmware will make the device run like its supposed to when we hit launch, its now a wait and see method.
Don't believe me then wait and see.

I think you are fundamentally misunderstanding something. Looking at a call-stack won't tell you much about performance. I'd agree with Simon.



I'm very sceptical. OEM provided firmware includes display drivers, camera drivers, OS configuration files, sensor drivers... stuff like that. That is all entirely irrelevant in regard to launching/resuming times!

All the drivers that interface with the SoC (CPU, RAM, GPU, etc) are built by MS/qualcomm for all devices, and those drivers are shipped together with the OS. They are already included in the WP8.1 preview for developers. Nothing is being emulated. I think you are taking experiences you made on Windows and expecting them to apply equally to WP, but that isn't true.

Anyway. I see no reason why OEM provided firmware should have any influence on launch/resuming times whatsoever. The issues I listed in post #10 are far more relevant.

The main reason why we should encounter the resuming message less frequently is because WP8.1 will collect more precise information on app memory usage. On WP8.0 the OS just assumes memory usage = maximum memory cap, which is almost always wrong. WP8.1 is expected to tombstone apps less frequently as a result. Under WP8.1 fast resume is also considered the default approach, so apps developed for WP8.1 are assumed to support that, whereas on WP8.0 that wasn't true. But as I said in post #10, many issues are involved.

Simon mentioned .Net native, which will make a huge difference once widely adopted on WP, but I don't know if WP8.1 is the platform on which .Net native will achieve widespread adoption.

The only thing I found in that article related to this thread was the following:



It is possible that apps which must initialize certain hardware, like the camera or motion sensor, can now do so more efficiently thanks to Cyan. But that isn't what we discussed here.

The author states that the perceived performance boost has some relationship to launch-animations and display technology. That is a somewhat strange statement to make however, because display technology has absolutely no impact on app launch times. That would be like saying Microsoft Office will launch faster on my PC, if I were to hook up a faster monitor. That is of course rubbish.

Based on the authors description, I think it's most likely that he's just being presented a faster launch-animation, which can provide the illusion of some apps starting up faster (particularly for those apps that have always launched in less time than it took to show that animation). If we were to measure app launch times with a stop watch, particularly those that aren't directly tied to hardware sensors, I'm sure we'd measure no difference at all.

Anyway, alone the fact that the improvement doesn't apply to all apps means that whatever the cause is, it isn't related to any of the things we discussed here. It's not a general, across the board improvement, because firmware doesn't include anything that can achieve that.

Straight from the article;

"The only difference is the Lumia Cyan firmware, which is tailored specifically for each model of Windows Phone. Firmware is super important as anyone who has updated a PC knows as it can allow certain drivers to work, optimize performance and enable certain functions."

 

[a5cent]

Straight from the article;

"The only difference is the Lumia Cyan firmware, which is tailored specifically for each model of Windows Phone. Firmware is super important as anyone who has updated a PC knows as it can allow certain drivers to work, optimize performance and enable certain functions."

Yes. Obviously firmware can optimize performance.... graphics performance (GPU driver), photo image quality performance (camera capture post-processing), battery life performance (smarter hardware resource management), and much much more.

However, firmware can't improve performance when it comes to launching apps. These are all different aspects of performance that aren't necessarily related to each other.

I hope that clears things up.

 

[Nathan Sokalski]

I don't mind needing to wait a few seconds, the problem is that when I see the Loading or Resuming screens, I usually end up getting sent back to the Start Screen, preventing me from opening the app at all. These screens would be a good thing if they were true, but they often don't end up doing what they claim to be.