New Mersenne Software For Test Mersenne Prime Numbers On Android

[ewmayer]

Originally Posted by axn

How difficult will it be to compile MLucas for android ?

I would love to see the performance of an actual state-of-the-art code running on one of these.

What kind of hardware CPU hardware are recent-model Android phones based on - Arm Cortex?

Again, no one in their right mind uses their actual phone to test M-numbers, I'm thinking more along the lines of a "buy a bunch of cheap used phone mainboards, hook to some kind of barebones power/comms harness, slap a little heatsink on each CPU, crunch away" kind of setup. No idea if that could potentially provide more bang-for-the-buck than, say, an Odroid micro-PC cluster, but surely there are many millions of such phones which get retired due to obsolescence, screen-damage, etc every year, but for which the mainboard is still functional. These things could potentially offer a cost-competitive LL-crunching solution, but one needs a lot of them (perhaps 20 or so Cortex A53 quads to equal a high-end Intel quad CPU) to match the crunching power, and so they need to be cheap. In the wilder of my fever dreams I picture 50-100 retired-phone mainboards bought on the cheap and assembled into a single chassis with similar footprint to a rackmount server blade a few inches high. :)

[kladner]

Originally Posted by ewmayer

What kind of hardware CPU hardware are recent-model Android phones based on - Arm Cortex?

https://www.phonearena.com/news/Sams...hmarks_id93192

Like has happened so many times before, Samsung's built two main versions of the Galaxy S8 (and correspondingly, GS8+), one running one of the company's own Exynos chips, and one featuring a top-of-the-line Qualcomm Snapdragon processor. And what makes that pairing especially exciting this time around is that this is the very first time we've been able to interact with either chip on a commercial handset.

Samsung Galaxy S8 processor showdown: Exynos 8895 vs Snapdragon 835
In most international markets, the Galaxy S8 is built around the new Exynos 8895, featuring new custom CPU cores, new Mali GPU, and with the whole thing fabricated via a cutting-edge 10nm process intended to maximize performance and power efficiency.

Based on the model number reported by my phone, a Galaxy S8, the zip file should be the correct specs. Model # is SM-G950U1. The CPU is a Samsung 8 core, with fast and slow quad units.

[ewmayer]

Originally Posted by kladner

https://www.phonearena.com/news/Sams...hmarks_id93192

Thanks - that glosses over the processor details, but that info is easily found - Here is AnandTech:

Our initial look at Snapdragon 835 revealed that its Kryo 280 performance cores are loosely based on ARM’s Cortex-A73 while the efficiency cores are loosely based on the Cortex-A53. Samsung's Exynos 8895 also has an octa-core big.LITTLE CPU configuration, but uses four of its own custom M2 cores paired with four A53 cores. Samsung introduced its first custom CPU core, the M1, last year. Compared to ARM’s A72, integer IPC was similar but the M1 trailed the A72 in efficiency. The M2 does not appear to be a radical redesign, but rather a tweaked M1 that offers the usual promises of improved performance and efficiency. Are the changes enough to top Qualcomm’s flagship SoC?

That doesn't say precisely what "loosely based on" means - I'm hoping that at least the basic instruction set is identical.

By way of experimentation, we really need just one phone containing such an Arm-quad-based CPU and a suitable developer-tools interface of the kind app-writers use. In our case, said dev-tools could actually be just barebones gcc+libs.

[thorken]

Originally Posted by Batalov

Great job, Gord!

The app its for mobile phone I make it for mobile phones and for mobile phones I dont show a app more faster than this only for mobile.

[science_man_88]

Originally Posted by thorken

The app its for mobile phone I make it for mobile phones and for mobile phones I dont show a app more faster than this only for mobile.

which algorithm of LL testing do you use ?

[thorken]

Originally Posted by LaurV

@OP: can you give us some info about the multiplication algorithm used? Is it school grade, is it karatsuba? FFT?

As per firejuggler's suggestion, we tried to run this toy in an android emulator - our kingdom is Cortex M, but we have some android emus laying around, as we have colleagues who develop applications with them, for our customers. We wanted to see how the tests scale in time - if the application is bogus, like reading exponents from a list and saying prime/composite, and doing other things meantime, then the time-scaling would be also odd. We have some idea and could guess the multiplication algorithm used, if any tuning was done or it is just blind school-grade high-level stuff. And we consider this to be a good test, because it would not be easy for the guy to fake the process (which involves exponentiations and multiplications related not only to the size of the exponents, but also to the number of bits which are 1 involved). And our toys can also debug/disassemble code. But unluckily, we can not run it. Our emulator does not like it, and our colleagues who could help didn't come back from their NY holidays yet...

Our advice for now: avoid it.

P.S. Carlos, you seem to be the only one liking it? Can you detail why? Or was that a voting mistake?

Let's see colleague who is very good at insulting you without even being able to execute the application given your extensive experience. You say my program reads numbers from somewhere? Finally apprentice magician is very good insult without having much idea of ​​programming so it seems, make your an app for Movil that calculates numbers of Mersenne, here you are making comparisons with i7 and the cpu of a mobile. This program is for mobile and is the fastest on Android, the only one I've seen is a Google Play apk and it's very slow, this one is much faster and I'm talking about Android and mobile cpus, prime95 in AtoM n270 It is slower than this app in a GT-I9195. Greetings.

[kladner]

Originally Posted by ewmayer

Thanks - that glosses over the processor details, but that info is easily found - Here is AnandTech:

That doesn't say precisely what "loosely based on" means - I'm hoping that at least the basic instruction set is identical.

By way of experimentation, we really need just one phone containing such an Arm-quad-based CPU and a suitable developer-tools interface of the kind app-writers use. In our case, said dev-tools could actually be just barebones gcc+libs.

Now I have to disavow my chip identification. I was going through apps, or things integrated with the OS, and came upon one with 'qualcomm' in its name. This shoots down the Samsung chip supposition, making it a Snapdragon octo, with lower clocks than the Samsung. (WRONG. The Snapdragon has higher clocks.) I am not sure it matters to me, as I don't do heavy compute on the phone. I mostly use it to take pictures, text, and set reminders. I rarely talk on it.

The only regular web use is the Chicago Transit Authority Bus Tracker. It is generally quite accurate. It gives me a sense of whether I should jog to the corner, or just take my time.

[thorken]

Here you have an installable apk of an application to calculate decimals of pi and e is fast but not the most and it is for Android who wants to try it

The app save the results in a txt file

https://drive.google.com/file/d/1yLu...w?usp=drivesdk

[Batalov]

When I was 14, I calculated 100,000 digits of Pi with my own (obviously very simple) program. Now, let's see... this was 40 years ago and every 1.5 years the computational ability of modern hardware doubles, so that was done on a computer that was 100,000,000 times slower than modern computers. (BESM-6)

Now if you compute 100,000 digits of Pi today, it will not even get you laid. And you computed, ... 2,000?

[Batalov]

Originally Posted by thorken

Here you have an installable apk of an application to calculate decimals of pi and e is fast ...

...and of course, you should have first googled just for a minute or so.
You cannot easily compete with the likes of y-cruncher and Mini-Pi for Pi and e.

On the Pixel 3XL (snapdragon 845), computing 1,000,000 digits took around 5.8 seconds.

You can definitely try - but for starters, you need to know what the state of the art is before beginning.

[chalsall]

Originally Posted by Batalov

Now if you compute 100,000 digits of Pi today, it will not even get you laid. And you computed, ... 2,000?

LOL... Back in the day the girls who were will willing to trade compute for sex looked a bit like Romulans or Cardassians.

Now-a-days even a serious Amazon instance won't impress most human females....