iPad (IOS) & Android Benchmarks to avoid the dreaded CPU Overload Error

[passedpawn]

What benchmark number runs the DJI Go 4 app without problems, in the hot summer?

For anybody looking to get an phone or tablet (of ANY make), it's difficult to know if the combined processor / on-chip graphics / cache memory / etc will run the DJI Go 4 app. If the processor is not capable enough, you'll get the dreaded "cpu overload" message in the app, the video will lag, and you'll lose confidence in controlling the drone. \

Apple constantly updates it's processors. The newest one gets integrated in a iPhone, then later makes it into the latest tablet.

The Android world is even wilder, with many manufacturers offering a wide range of processors and related capabilities.

The charts in the following posts boil the performance down to a single number. There must be a specific benchmark ABOVE WHICH the app runs reliably, even in hot weather.

All data comes from here: iOS Benchmarks - Geekbench Browser

 

[passedpawn]

IOS Devices and Benchmarks

 

[passedpawn]

Android Devices and Benchmarks (note: this was only a small percentage of all the devices - see the actual site for full list: Android Benchmarks - Geekbench Browser )

 

[gnirtS]

Ive never seen a CPU overloaded error on any android device ive used (roughly 6 of them with the drones). Im not sure its even coded into android Go4 app.
Some devices work better than others, some for sure get bogged down CPU wise but i dont think that overloaded message ever shows.
My devices all seem fine in the heat - i live in the tropics so every flight is in direct, hot 32c or more sunshine. The devices are sometimes too hot to comfortably touch but still work.
My Pixel 2 XL will turn itself off in its black case if i leave it in the sun for half an hour though (not flying a drone). Needs 20 mins in the fridge to recover.

Benchmarks really aren't much use at they test things that may not necessarily be needed or optimum for Go4 so you're not going to find a minimum magic number at all.
Android has so many different architectures for CPU, GPU, internal bus, interface speeds etc its just not going to produce anything.
Its also going to depend how much the manufacturer has butchered the OS with power, battery or other random optimisations. A nice clean pixel could easily give a different performance to what you'd expect comparing to a horrifically altered OS shipping on a Samsung for example.

 

[passedpawn]

My experience in Florida heat:

iPad Mini 2 (7.9") would eventually overheat and stutter on the app. Apparently it's processor core identical to Mini 3 internally.

1245 = BAD!

iPad 6th generation 9.7" runs great.

3488 = GOOD!

 

[passedpawn]

Benchmarks really aren't much use at they test things that may not necessarily be needed or optimum for Go4 so you're not going to find a minimum magic number at all

There is no better way to assess. If you want to know what goes into the score, you can look here, but I can tell you that it assesses the hardware AND firmware drivers, as it should. It also takes into account floating point efficiency, GPU efficiency, etc. Maybe there' a better benchmark - I don't know. But benchmark testing is what's important because it allows us to assess the device as a fully system, instead of just comparing processor cores and expecting to extrapolate some result from that alone.

 

[gnirtS]

There's no better way but that doesn't matter if the method is simply not going to work anyway. Thats the problem with benchmarks - they're generally not specific to certain applications.
Its not using the hardware and all its functions in a way that mirrors the app. The app is going to depend on a lot more specific bottlenecks such as memory read/write blocks, caches, floating point, multithreaded ability and so on.

The benchmarks will tell you which program runs the benchmarks fastest but it may have no real world bearing on Go4. Especially if looking for a magic number. They dont help on real computers for specific apps either - that's why most games have their own built it.

Its perfectly possible to have a specific platform that runs go4 well with a significantly lower benchmark score than a platform that runs it terribly, stutters and ranks highly. Especially when you get into battery and temperature optimisation modes that most manufacturers write themselves, dont document and implement with a huge amount of variability.

Its also going to be app dependent to a huge degree. My old Nexus 7 2013 used to run Go4 and the mavic perfectly. Then over a series of progressive updates to go 4 got more and more bogged down and slower and hasn't been acceptably usable on Go4 for about 18 months now. The benchmark score wouldn't have changed, the magic number wouldn't have changed but the app itself had progressed (bloatier, less optimised, lazier coding etc).

 

[passedpawn]

There's no better way but that doesn't matter if the method is simply not going to work anyway. Thats the problem with benchmarks - they're generally not specific to certain applications. Its not using the hardware and all its functions in a way that mirrors the app. The app is going to depend on a lot more specific bottlenecks such as memory read/write blocks, caches, floating point, multithreaded ability and so on.

Obviously, it would be ideal if the DJI app had its own benchmark. Many of the benchmarks that gamers rely on actually use the game engines - good for them for sure. I don't think the drone market is as big as the Fortnite market yet.

I'll just have to rely on benchmarks such as the one I posted above. The key is to ensure that the benchmark workload approximately simulates the app you're running. I'd guess it's not bad. If you have a closer look, you'll see that indeed this benchmark is exercising all cores simultaneously. It's also running floating point math. Hard to assess cache hit efficiency, depends on the actual code running, but it has to be part of the test. Read/write blocks that you mention is related to cache and MMU efficiency also, and of course dependent on the coders' use of threads to get the benefit out of multiple processor cores.

 

[Johnmcl7]

In general benchmarks can have questionable relevance to mobile applications due to differing levels of optimisation but they're even less applicable to DJI Go which doesn't appear to correlate directly with hardware performance as there's cases where high end devices are not working as well as much older devices. For a long time I was using an old Snapdragon 800 based device which was running DJI Go without issue while many much higher performing devices were having problems. I'm not saying lower performance devices are guaranteed to be better and there's certainly low end device with insufficient performance to run DJI Go but it's certainly not the case that a device that scores a number in an arbitrary benchmark has any relevance to its DJI Go performance.

Furthermore I assume you're thinking that the cpu loaded message means the device has insufficient performance but although I've not seen it myself, I thought it was related to thermal performance issues especially as devices with high end hardware appear to suffer the issue. Apologies if I'm misunderstanding your post.

There's a useful list of user feedback on phones and tablets for DJI Go below and noticeably some of the highest recommended tablets are lower performance models:

Best Tablets and Phones for DJI GO & DJI Fly (Updated Daily) | Phantom Help

Phantom Help is a site for help with DJI Phantom drones. Find the latest fixes, how-tos, guides, accessories, modifications, and more.

www.phantomhelp.com

That's what I would use (and have used) to choose suitable devices for DJI Go.

 

[passedpawn]

There's a useful list of user feedback on phones and tablets for DJI Go below and noticeably some of the highest recommended tablets are lower performance models:

Best Tablets and Phones for DJI GO & DJI Fly (Updated Daily) | Phantom Help

Phantom Help is a site for help with DJI Phantom drones. Find the latest fixes, how-tos, guides, accessories, modifications, and more.

www.phantomhelp.com

That's what I would use (and have used) to choose suitable devices for DJI Go.

That's pretty sweet list there. I assume that only the ones with green dots are deemed suitable for DJI Go 4?

 

[Johnmcl7]

That's pretty sweet list there. I assume that only the ones with green dots are deemed suitable for DJI Go 4?

It's not my site but I believe a green dot means DJI officially support it and it's on the compatibility list here:

Mavic 2 - Product Information - DJI

The Mavic 2 is DJI's flagship consumer drone built for pros and enthusiasts. Discover more about the Mavic 2 Pro and Mavic 2 Zoom with specs, FAQs, video tutorials, guides, and user manuals. Learn more at DJI.com.

www.dji.com

"
iOS V 4.3.20
Requires iOS 10.0.0 or later (Mavic 2 Pro/Zoom requires iOS 10.0.2 or later).
Compatible with iPhone X, iPhone 8 Plus, iPhone 8, iPhone 7 Plus, iPhone 7, iPhone 6s Plus, iPhone 6s, iPhone 6 Plus, iPhone 6, iPhone SE, iPad Pro, iPad, iPad Air 2, iPad mini 4.
Optimized for iPhone X.

Android V4.3.20
Requires Android 5.0 or later.
Compatible with Samsung S9+, Samsung S9, Samsung S8+, Samsung S7, Samsung S7 Edge, Samsung S6, Samsung S6 Edge, Samsung Note 8, Huawei P20 Pro, Huawei P20, Huawei P10 Plus, Huawei P10, Huawei Mate 10 Pro, Huawei Mate 10, Huawei Mate 9 Pro, Huawei Mate 9, Huawei Mate 8, Honor 10, Honor 9, Vivo X20, Vivo X9, OPPO Find X, OPPO R15, OPPO R11, Mi Mix 2S, Mi Mix 2, Mi 8, Mi 6, Redmi Note 5, Google Pixel 2XL, OnePlus 6, OnePlus 5T. "

I don't this list is updated that often as the iPhone Xs has been out nearly a year now and it's not on there nor is the iPad Mini 5 both of which I believe work well with DJI Go 4.