Is there a mobile device which computing strength beats the Air 5's M1?

[SJZL 2.0]

I'm wondering if there is a mobile device like the ones that aren't included in the public or are experimental. If so, could you list them for me? Thanks.

 

[Count von Schwalbe]

What exactly do you mean by mobile? Depending on who you ask, that could be anything from a ITX tower to a smartwatch. If you are looking for a direct competitor to the iPad Air M1 chip, you will not find any Android tablets with that kind of performance. I cannot, however, speak for Chromebooks, as I have had no real interest in them. I would recommend a 2-in-1 laptop with a Ryzen 5 5600u or better, if you are not opposed to Windows. Link to Example 1 and Example 2.

 

[SJZL 2.0]

What exactly do you mean by mobile? Depending on who you ask, that could be anything from a ITX tower to a smartwatch. If you are looking for a direct competitor to the iPad Air M1 chip, you will not find any Android tablets with that kind of performance. I cannot, however, speak for Chromebooks, as I have had no real interest in them. I would recommend a 2-in-1 laptop with a Ryzen 5 5600u or better, if you are not opposed to Windows. Link to Example 1 and Example 2.

Phones and Tablets

 

[R0H1T]

What exactly do you want it for? Like running benchmarks, work or anything else in particular?

 

[SJZL 2.0]

What exactly do you want it for? Like running benchmarks, work or anything else in particular?

Nothing other than acknowledging its existence. It is a part of my curiosity.

 

[Frick]

No.

 

[GreiverBlade]

No.

not totally right ... for day to day real life use the M1 is not too much above a SD860 based device ... based on feeling and blind test between an iPad Air M1 and a Xiaomi Pad 5
tho if tied (shackled?) to Apple ecosystem without android alternatives, yeah the iPad Air M1 has no rival

but if not tied to it, the Xaomi Pad 5 is a top dog that can give the former a run for the money at around 350$ for a 128gb (aka "~half the price, double the storage"

it's the same as saying a SD 8 Gen 1 is always better than a SD870 device, when in day to day there is little to no differences (and the former has more heat issues than the later which more than often make the SD 8 Gen 1 slower than the SD870 )

 

[SL2]

not totally right ... for day to day real life use the M1 is not too much above a SD860 based device ... based on feeling and blind test between an iPad Air M1 and a Xiaomi Pad 5
tho if tied (shackled?) to Apple ecosystem without android alternatives, yeah the iPad Air M1 has no rival

but if not tied to it, the Xaomi Pad 5 is a top dog that can give the former a run for the money at around 350$ for a 128gb (aka "~half the price, double the storage"

The first you mention is almost as fast, and the second is very price competitive. Neither contradicts fricks "no", you just confirmed what he said.

 

[bug]

I'm wondering if there is a mobile device like the ones that aren't included in the public or are experimental. If so, could you list them for me? Thanks.

It really depends on what your definition of "computing strength" is.
M1 is a good chip, but it's not magical. It packs a lot of punch because it uses a lot of die area. However, perf per sq mm is about in line with the better ARM designs. Being so big, it must be expensive to make, so it probably doesn't win in the perf/$ department either. This cannot be easily confirmed, because Apple doesn't disclose M1's price, but big chips have never been cheap. M1's big win is that because somehow (I suspect due to its size) it doesn't push the silicon so hard, so it says in the efficient territory. More battery life is probably the biggest win you can have in the mobile space.

 

[LabRat 891]

None that I know of.
Apple spent A LOT of money (and probably near-equal amounts of engineering manhours) on the M1. AFAIK, they've been working on the uArch at least since the first rumblings of Apple dropping Intel. (Don't get me wrong, all new architectures and lithography start R&D years before commercial/industrial availibility.) I do not like Apple, at all. However, I have to respect the engineering might that got the M1 on the market, when they did.

The Snapdragon 860 is a tweaked and upclocked revision of an older chip; it is generationally behind. The fact it performs as close as it does is indeed impressive. The Snapdragon 870 5G and Snapdragon 8 Gen 1 I'm less familiar with, but by benchmarks, they're *very* close to the A1. Neither appears to 'beat' it, at least in benchmarks, however.
For the time being, I would concede that Apple is a generation ahead with their M1 (in the mobile space).

 

[GreiverBlade]

The first you mention is almost as fast, and the second is very price competitive. Neither contradicts fricks "no", you just confirmed what he said.

yet it does contradict it ... unless the op is very specific about raw computing strength which only apply in very specific scenario ...

thus for real life application : no, no confirmation , for synthetic/computational bench : yes, i might confirm indeed, baseline is : yes that alternative compete with it with ease.

with years on ... i learned to ditch synthetic/computational benchmarks as they are somewhat irrelevant.

@LabRat 891 yep the SD860 is a tweaked 855, as for the SD870, it is the most competitive high end SOC around, the SD 8 Gen 1 is the latest flagship from Qualcomm but as i mentioned it suffer from poor thermal and performances usually are close, or even bellow in some case, to the 870 in day to day scenario (not in synthetic bench or "Samsung controversial boosted" bench although these bench usually favor their Exynos SOC while in pure "impartial" bench the Exynos is just a touch behind instead of being "abyssal gap" ahead and in unboosted it actually perform under, for both SOC, meaning day to day users have lower result than the bench the brand and some reviewer posted)

@bug yet strangely enough the biggest complaint i hear from friends about their iPhones (12/13) and even my mother about her iPad Mini devices is battery, having to plug them every end of days (middays even sometime), on my side i rarely have to plug my F3 more than once every 2 days (my mom also have 2 days out of her iPhone 7 but that's without using it more than checking mails and browsing a bit)

on Apple i do not dislike their hardware ... rather their software based on how closed it is and the company itself based on how they behave.

 

[r9]

I'm not Apple fan in general but they played the M1 amazingly.
Having the first line up relatively cheap for what it is and being an Apple product so people would adopt it.
Also it's not just the chip it's how great Rosetta works for non ARM apps. The Windows for ARM was pure garbage you couldn't even use Chrome properly let alone anything more demanding.
If you into video/photo editing/rendering they just destroy Intel/AMD.
It reminds me of PowerPC days of Apple.

 

[Darmok N Jalad]

Another component that makes the M1 more effective is the use of specialized hardware, and that is tied to Apple’s ability to move developers along more quickly since there are fewer hardware configurations to be concerned with.

A perfect example is DxO’s PureRAW, a RAW AI noise reduction software. Version 1 used the M1s GPU, and a 20MP RAW processed in about 20s, which might be comparable to a Polaris-era GPU. Version 2 of DxO’s software uses the Neural Engine in the M1, which reduces the processing time down to around 7 seconds. If one were using Windows for this software, a relatively powerful GPU is required, or you will be waiting maybe 2 minutes per file to run this via CPU, where the base M1 is plenty good enough on a Mac. The NE gets overlooked, but it actually does quite a bit of heavy lifting on both Mac and iOS devices for photo and video tasks.

 

[SJZL 2.0]

yet it does contradict it ... unless the op is very specific about raw computing strength which only apply in very specific scenario ...

thus for real life application : no, no confirmation , for synthetic/computational bench : yes, i might confirm indeed, baseline is : yes that alternative compete with it with ease.

with years on ... i learned to ditch synthetic/computational benchmarks as they are somewhat irrelevant.

@LabRat 891 yep the SD860 is a tweaked 855, as for the SD870, it is the most competitive high end SOC around, the SD 8 Gen 1 is the latest flagship from Qualcomm but as i mentioned it suffer from poor thermal and performances usually are close, or even bellow in some case, to the 870 in day to day scenario (not in synthetic bench or "Samsung controversial boosted" bench although these bench usually favor their Exynos SOC while in pure "impartial" bench the Exynos is just a touch behind instead of being "abyssal gap" ahead and in unboosted it actually perform under, for both SOC, meaning day to day users have lower result than the bench the brand and some reviewer posted)

@bug yet strangely enough the biggest complaint i hear from friends about their iPhones (12/13) and even my mother about her iPad Mini devices is battery, having to plug them every end of days (middays even sometime), on my side i rarely have to plus my F3 more than once every 2 days (my mom also have 2 days out of her iPhone 7 but that's without using it more than checking mails and browsing a bit)

on Apple i do not dislike their hardware ... rather their software based on how closed it is and the company itself based on how they behave.

8Gen 1 is 2.5 Teraflops while M1 is 2.6

Using android/windows felt better and sturdier to use than the MacOS and the iPhone/iPad software. Although I've used an iPad for a long time.

 

[bug]

Another component that makes the M1 more effective is the use of specialized hardware, and that is tied to Apple’s ability to move developers along more quickly since there are fewer hardware configurations to be concerned with.

A perfect example is DxO’s PureRAW, a RAW AI noise reduction software. Version 1 used the M1s GPU, and a 20MP RAW processed in about 20s, which might be comparable to a Polaris-era GPU. Version 2 of DxO’s software uses the Neural Engine in the M1, which reduces the processing time down to around 7 seconds. If one were using Windows for this software, a relatively powerful GPU is required, or you will be waiting maybe 2 minutes per file to run this via CPU, where the base M1 is plenty good enough on a Mac. The NE gets overlooked, but it actually does quite a bit of heavy lifting on both Mac and iOS devices for photo and video tasks.

If only specialized hardware wasn't the double-edged sword that it is...

 

[SL2]

yet it does contradict it

Look here:
"for day to day real life use the M1 is not too much above a SD860 based device"

A contradiction to what frick said would have been to say that the 860 is faster than M1.

unless the op is very specific about raw computing strength which only apply in very specific scenario ...

That has very little to do with what the OP asked for, he never mentioned real life scenarios, in fact, he even asked to include non public products.

No one here have showed any benchmarks where a chip beats M1 in a tablet, and I'm starting to realize why. Preferably something else than geekbench.

 

[bug]

No one here have showed any benchmarks where a chip beats M1 in a tablet, and I'm starting to realize why. Preferably something else than geekbench.

Keep in mind fair benchmarking is hard when both the silicon and the APIs are different.

 

[Darmok N Jalad]

If only specialized hardware wasn't the double-edged sword that it is...

True, but I think you could make the case that this is where the industry needs to head anyway, provided it can be standardized. In my case, the NE replaces the need for an expensive GPU, as I don’t really use a GPU for anything else intensive. I don’t even need a more expensive version of the Mac either, the base M1 has the same NE as even the M1 Max. It isn’t till you get to the Ultra that it is faster, and in that case it is doubled up.

 

[GreiverBlade]

Look here:
"for day to day real life use the M1 is not too much above a SD860 based device"

A contradiction to what frick said would have been to say that the 860 is faster than M1.

technically not, since "no." is a generalization thus imprecise and not specifically true in all scenario, for me "is not too much" mean "you don't feel the difference in day to day usage, thus: it's equal in that scenario ... Synthetic/computational scenario : yes indeed the M1 has no equivalent, outside of that, i maintain that anything above a SD860 or equivalent match it with ease, granted if not being tied to Apple ecosystem. (plus i only mentioned a 860 which is generation behind if the Xiaomi pad 5 came with a 870 or a 8 gen 1 or even a Dimensity 9000 it would be another slice of bread)

No one here have showed any benchmarks where a chip beats M1 in a tablet, and I'm starting to realize why. Preferably something else than geekbench.

not really needed Bench do not replace side by side (preferably blind) comparison and that's what i did (why show a benchmark that would show controlled environment superiority, imho: pointless)

That has very little to do with what the OP asked for, he never mentioned real life scenarios, in fact, he even asked to include non public products.

indeed hence my first answer mentioning real usage and pricing, since the OP did not precise "i want the ultimate E-peen competitive SOC that can beat the M1 on synth/computational"
and on my second post i mentioned "unless the op is very specific about raw computing strength which only apply in very specific scenario ..."
innit?

well the OP just mentioned: "SD 8 Gen is 2.5 Teraflops while M1 is 2.6" thus yes on that side we got it covered (then the Dimensity 9000 would also qualify since it's on par or slightly above the SD 8 Gen 1 )

nevermind i re read the title then yes : computational side : almost no equals to the M1, some close call like the 8 Gen 1 and Dimensity 9000 tho ...
although computational is not what you would ask from a tablet nonetheless (thus my initial divergence, because i tend to pay more attention to the content of the 1st post that the title my bad "teehee" )

@SJZL 2.0 i share your point of view on Android/win versus iPhone/Mac/iPad and i also used for a long time an iPad as a tablet (a shame, or a half shame ... some games did support cross platform accounts although for other app than games well yes... some equivalents but i prefer those found on Android)

 

[bug]

True, but I think you could make the case that this is where the industry needs to head anyway, provided it can be standardized. In my case, the NE replaces the need for an expensive GPU, as I don’t really use a GPU for anything else intensive. I don’t even need a more expensive version of the Mac either, the base M1 has the same NE as even the M1 Max. It isn’t till you get to the Ultra that it is faster, and in that case it is doubled up.

Well, the FPU was specialized hardware in its day. Let's hope neural accelerators follow the same path. Because for the moment everyone has their own version and it's pretty hard to support everything at the same time.

 

[cvaldes]

No one here have showed any benchmarks where a chip beats M1 in a tablet, and I'm starting to realize why. Preferably something else than geekbench.

Geekbench runs are too limited in duration.

The bigger problem with benchmarks is that they typically use maximum loads in an extremely limited subset of situations which is not how people use devices in the real world.

Worse there are recent technological improvements that aren't being captured accurately by current benchmarking utilities. A lot of these relate to specialized functions like hardware-based video encoding or machine learning.

It's not about how your mobile device performs in a 10 minute benchmark, it's how it performs until the battery is dead through a variety of tasks and situations.