Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Number crunching :
Android
| Author |
Message |
|---|
|
|
|
with cellphone's now having gpu's and old dying systems being replaced with apu bearing motherboards the rate of gpu supporting systems is going exponential.
Are these time frames realy going to holed? | |
|
 Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
                         
|
with cellphone's now having gpu's and old dying systems being replaced with apu bearing motherboards the rate of gpu supporting systems is going exponential.
Are these time frames realy going to holed?
Of course not. :)
The estimates are the best bet for what we know now.
Certainly, computers will get faster as time goes on, but not always, and it's impossible to predict how much improvement there will be. For example, the Kepler series (GTX 6xx) of GPUs are a huge improvement for gamers, but are very disappointing for us number crunchers and actually do many of our kind of computations slower than the previous generation of GPUs.
As for mobile device (i.e., phones and tablets), I don't expect those to ever be a huge factor when compared to desktops or laptops. The power and cooling budgets on those devices just do not support hard core crunching. Yes, it can be (and has been) made to work, but these are, by definition, ultra low power devices and can't crunch very fast.
Far worse, however, is the fact that they're not designed for crunching and have absolutely no cooling system at all. Cell phones overheat running normal apps (my iPhone occasionally shuts down to protect itself when running a navigation app, for example.) Running a full time 100% crunching program on a mobile device will likely overheat it, and on a device that isn't smart enough to shut down because of excessive temperature, destroying the device is a very real possibility.
(I know the iPhone 4 has thermal protection, but I'm fairly sure many Android phones do not. Crunching on a device without over-temp protection may destroy it.)
For both reasons, I don't expect phones and tablets to ever play a significant role here.
____________
My lucky number is 75898524288+1 | |
|
|
|
For both reasons, I don't expect phones and tablets to ever play a significant role here.
Consider the following:-
1. There will be around 1 billion active Android devices in the near future.
2. The speed of a current Android device is only around 2-3 times slower than the speed of a current average (desktop) CPU-core (for a TRP-sieve task)
Now calculate what percentage of all those phones and pads it would take to, say, double PrimeGrid's current TRP-Sieve thruput, while running just 8 hours per day. (i.e. while plugged-in and charging overnight.)
And then it only remains to ask yourself.. "Why Not?"
While overheating is obviously a concern, it would be child's play to add code to ensure (serious) crunching only takes place while charging, and I'm pretty sure EBay would quickly be awash with combination charger/heatsink/fan devices for yr favourite Android gizmo as soon as the need (and market!) arose.
We seem to be missing the whole point of what BOINC is all about here.. Its NOT about how fast (or efficient) any particular device is, at doing the job... its ALL about how many (different!) devices we can get to work together to get the job done as quickly as possible!
I personally feel it would be an absolute tragedy if we don't go all out to persuade those many millions of potential BOINC'ers to get their phone/pad to do something Really Useful for a change, while charging overnight!
Anyone care to do the math?
____________
 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
And then it only remains to ask yourself.. "Why Not?"
While overheating is obviously a concern, it would be child's play to add code to ensure (serious) crunching only takes place while charging, and I'm pretty sure EBay would quickly be awash with combination charger/heatsink/fan devices for yr favourite Android gizmo as soon as the need (and market!) arose.
You hear about how the FAA just grounded the entire 787 Dreamliner fleet because of battery overheating problems?
Ever notice how hot your phone gets while charging sometimes?
From a thermal perspective, charging is the worst time to be crunching because the battery heats up a LOT. Sometimes the device will even stop charging temporarily if the battery exceeds thermal limits.
They're not designed for continuous full power use of the CPU, and if your device doesn't have built in thermal protection, it can destroy itself.
I know people who have destroyed their smartphones this way.
From a project perspective, sure, it's a great untapped source of computing potential that is growing exponentially.
But it would be irresponsible for us (or anyone) to recommend that people crunch on their phones or tablets. We don't want to be in the position of recommending actions that have significant potential to destroy your property. (I'd go so far as to speculate that if your device lacks thermal protection, eventual destruction is either very likely or guaranteed.)
The prime numbers can wait a bit longer to be discovered. I don't want to be responsible for naive BOINC crunchers (and many are not very computer savvy) destroying their phones and tablets.
Things may change in the future, of course. But right now, anyone crunching on their hand held device is being foolish unless they don't care whether the device survives.
People wreck their laptops, and even desktops by crunching. Smartphones don't even have fans, and are often enclosed in protective cases which do a great job of sealing in the heat.
If Dr. Evil (or Dr. Apple???) wanted to destroy all the world's Android phones and tablets, releasing a BOINC crunching app on Android would be a darn good way of doing it. Call it "Angry BOINCs" and there will be 100 million downloads in a week. :)
Running BOINC on Android is fun, it's cool -- and it's very, very misguided in my opinion.
I'm not posting this to argue with you -- I'm posting this as a warning to readers who might be tempted to try this without understanding the implications and dangers.
I personally feel it would be an absolute tragedy if we don't go all out to persuade those many millions of potential BOINC'ers to get their phone/pad to do something Really Useful for a change, while charging overnight!
We'll just have to agree to disagree on this. I think it's a terrible idea.
____________
My lucky number is 75898524288+1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
                  
|
|
If I may contribute myself & stir things further...
Yes the massive capability is out there, but BOINC has a setting to allow only certain cores, therefore reducing heat. Surely there could be a port of BOINC that will only do work under a limited throttling, & if there are heat sensors then that's a bonus & BOINC-throttling could be dependant on heat. This I would hope would steer clear of device destruction. | |
|
|
|
|
I really don't see the difference between encouraging your average Joe with a desktop gaming rig, and the same guy with an Android device, from dabbling in this pastime. Neither device was custom built for the extreme stresses of this kind of code, and both carry a very real risk of meltdown if Joe is not careful!
I myself have cooked a few CPUs and GPUs when cheap & nasty fans fail in mid-summer! - But I don't recall either rushing out to file a lawsuit, or looking up to see how many Boeings I had brought down!
We clearly need to make all potential users aware of the risks, especially in the early experimental stages (as indeed we do already) - but as our Android-Device-Database grows, we will soon be able to release 'tweaked' code (or core% settings!) that are guaranteed to be no more dangerous than that good ol' 'Angry-Birds' app for each of the more popular Android devices.
From a thermal perspective, charging is the worst time to be crunching because the battery heats up a LOT. Sometimes the device will even stop charging temporarily if the battery exceeds thermal limits.
Yes, batteries can get very hot if charged quickly.. but.. as you point out - your average Android device will stop charging to protect the battery! - I just Googled "Android Temperature Monitor" and there are oodles of apps out there to monitor and protect yr baby! - I downloaded a couple to my little cheapie Samsung GT-B5510 and they seem to do a fairly good job of monitoring CPU and Battery, and I'm pretty sure that most current and future devices will have this kind of protection built in!
Anyway, running a power-intensive app while charging will actually be beneficial in this regard, as it will be using up a large chunk of the charging-current, forcing the battery to charge more slowly!
I'm pretty sure there must be plenty of nefarious or simply badly-written apps out there that are just as capable of melting-down your baby - without that warning!
So I would consider us to be duty-bound to investigate the processing-limits of all those Android devices out there, not just for our own sake, but for the sake of expanding the current knowledge-base, for the benefit of Manufacturers and End-Users alike.
____________
| |
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2102
ID: 1178
Credit: 8,172,728,223
RAC: 4,520,595
                                    
|
|
My own opinion on the Android crunching issue falls somewhere more "middle-of-the-road." Mike's points are very valid, and PG would do well not to be seen as responsible for tons of devices melting down regardless of the likelihood of lawsuits (e.g., bad press might be even more damaging).
On the other hand, the tremendous computational potential is out there and should be investigated. Furthermore, there are already a number of tablet coolers out on the market now (with more likely to come as that market continues to expand) that answer at least a portion of the heat issue for some devices (much as laptop coolers help with heat).
As I see it, there are several issues that would need to be taken very seriously before PG actively pursued any Android (or other portable OS) applications:
1) throttling (as noted above) would be a necessity for such apps
2) Tablets and phones are very different devices with very different capabilities/issues that would need to be understood as apps were developed (e.g., any cell coolers on the market yet? Do those make sense given how phones are carried/used? etc.)
3) What applications would make sense to begin with? That is, Android devices are low power devices (both in the sense of power saving for battery use and computational power per device compared to CPUs and GPUs used now). As we push more deeply into sieves and prime searches, the size of our work units is growing larger. This begs the question of whether Android devices could crunch much of our work in a timely fashion or if special projects/adaptations would be needed.
4) Android (and other portable OSes) are a rapidly changing beast (e.g., Android has moved from Gingerbread to ICS to Jelly Bean very rapidly compared to Windows/Mac/Linux). What are the costs inherent in supporting such rapid changes for PG? Would those resources be better placed in other tasks?
I am sure there are others that I am missing at the moment...
For these reasons (at least), I think any significant effort into Android applications at PG is at least a year away (probably longer) if ever.
| |
|
|
|
|
I just found the NativeBoinc Android client (http://nativeboinc.org/site/uncat/start) and I'm trying it out on my phone. I have a Razr Maxx HD, which has a dual core processor, with each one clocking in at 790 MIPS floating and 2856 MIPS integer. So far, PG seems to run fine...
____________
 | |
|
|
|
|
Some phones need to be rooted to get full effect of both processors. Some Telco's lock it down (stupidly). I still have not had success rooting mine, but I've not tried really hard.
So if tasks run twice (or more) as long as CPU time, this is the reason.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
Get System Tuner app, set BOINC to 1 core and monitor the temps. Don't go above 70C. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
|
I moved this discussion to its own thread since it has little or nothing to do with "sub-project "life" expectancy".
____________
My lucky number is 75898524288+1 | |
|
|
|
|
I have been running TRP Sieve on my single core HTC Thunderbolt and my quad core HTC DNA.
There are a number of projects available to crunch on NativeBoinc. The projects I participate in outside of NativeBoinc are Asteroids/MW/PG/SETI/WUProp. So my experience is only with those projects on the phones.
So far, I have not gotten any SETI WU's on either phone, so it is out of the discussion for me at the moment.
Of the projects running MW/Asteroids/PG (& WUProp), PG's TRP Sieve, by a noticeable margin, hotter then the other projects. Credit wise, due to WU run time and obvious processing power, along with only running while charging on a strong charger, the credit return is not much.
But for the WUProp badge people, and those who can monitor the heat, it is not bad for a little extra credit. It would be nice if more sub-projects were available, such as the smaller ones that don't need huge power or mega run times to complete.[/b]
____________
 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
It would be nice if more sub-projects were available, such as the smaller ones that don't need huge power or mega run times to complete.[/b][/b]
Except for the two sieves, every other PrimeGrid BOINC project is going to A) REQUIRE double-precision floating point hardware, which I understand is not present in all ARM processors, B) will likely run significantly HOTTER than the sieves, and C) in many cases will be longer tasks than the sieves.
Both LLR and Genefer are almost entirely dependent on double precision floating point for the entire calculation, so all of those projects can't run on a processor that doesn't have it.
The Sieves are the easiest tasks here, and run much cooler than either LLR or Genefer, which are essentially CPU torture tests. Very few apps run a CPU as hard as those programs.
Furthermore, while the length of an LLR or GFN calculation is dependent on the number being tested, and can't be changed, with a sieve, the tasks can be made as long or as short as we want.
For those reasons, sieves are clearly the best PG application to run on a phone or tablet. Since the PPS-Sieve is primarily crunched on GPUs, those tasks are much longer than the TRP Sieve tasks, which are only run on CPUs. This make the TRP sieve the better choice for a phone or other mobile device.
In other words, if you're running TRP Sieve on your phone, that's absolutely the best PrimeGrid application for your device.
____________
My lucky number is 75898524288+1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
Again just to point out I wouldn't have all cores going for heat reasons, unless you can maintain your phone being in a cool environment e.g by a cool window. Also do not process while charging - this is 'twice' the load on the battery + will shorten the lives of everything even more. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Also do not process while charging - this is 'twice' the load on the battery + will shorten the lives of everything even more.
How long can you crunch if you're *not* charging? Isn't charging a pre-requisite to doing any sort of crunching?
This isn't entirely a joke: perhaps you could try charging/crunching with the phone inside one of those little cube refrigerators? The phone should stay warm enough to avoid condensation, and the fridge will keep the phone from overheating. I won't guarantee nothing bad will happen, but it might work. The wifi signal might suffer, however. One of the bad things I could imagine is the outside of the touchscreen glass being a lot colder than the inside of the glass, which could cause it to crack. That doesn't usually happen with thin glass, however.
(Several years ago someone built a heavily overclocked desktop computer inside a cube fridge, essentially using the fridge as the computer case. If you google it, you'll get lots of hits.)
Mandatory disclaimer: We don't recommend you use your phone or tablet for crunching. It's not built for it, and damaging (or destruction) of the device is within the realm of possibility.
____________
My lucky number is 75898524288+1 | |
|
|
|
It would be nice if more sub-projects were available, such as the smaller ones that don't need huge power or mega run times to complete.[/b][/b]
Except for the two sieves, every other PrimeGrid BOINC project is going to A) REQUIRE double-precision floating point hardware, which I understand is not present in all ARM processors, B) will likely run significantly HOTTER than the sieves, and C) in many cases will be longer tasks than the sieves.
Both LLR and Genefer are almost entirely dependent on double precision floating point for the entire calculation, so all of those projects can't run on a processor that doesn't have it.
The Sieves are the easiest tasks here, and run much cooler than either LLR or Genefer, which are essentially CPU torture tests. Very few apps run a CPU as hard as those programs.
Furthermore, while the length of an LLR or GFN calculation is dependent on the number being tested, and can't be changed, with a sieve, the tasks can be made as long or as short as we want.
For those reasons, sieves are clearly the best PG application to run on a phone or tablet. Since the PPS-Sieve is primarily crunched on GPUs, those tasks are much longer than the TRP Sieve tasks, which are only run on CPUs. This make the TRP sieve the better choice for a phone or other mobile device.
In other words, if you're running TRP Sieve on your phone, that's absolutely the best PrimeGrid application for your device.
Thanks for the reply. Are there other Sieve sub-projects similar to TRP that could run on the phones?
____________
| |
|
|
|
Again just to point out I wouldn't have all cores going for heat reasons, unless you can maintain your phone being in a cool environment e.g by a cool window. Also do not process while charging - this is 'twice' the load on the battery + will shorten the lives of everything even more.
If processing on just battery, you are looking at VERY short crunching time before you have to plug in the charger. Means a great deal of baby sitting and even less actual credit return. I will try to back off one core, just downloaded the app suggested above.
____________
| |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Thanks for the reply. Are there other Sieve sub-projects similar to TRP that could run on the phones?
At this time, PPS-Sieve is the only other sieve, but the tasks are about 10 times longer.
____________
My lucky number is 75898524288+1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
Also do not process while charging - this is 'twice' the load on the battery + will shorten the lives of everything even more.
How long can you crunch if you're *not* charging? Isn't charging a pre-requisite to doing any sort of crunching?
This isn't entirely a joke: perhaps you could try charging/crunching with the phone inside one of those little cube refrigerators? The phone should stay warm enough to avoid condensation, and the fridge will keep the phone from overheating. I won't guarantee nothing bad will happen, but it might work. The wifi signal might suffer, however. One of the bad things I could imagine is the outside of the touchscreen glass being a lot colder than the inside of the glass, which could cause it to crack. That doesn't usually happen with thin glass, however.
(Several years ago someone built a heavily overclocked desktop computer inside a cube fridge, essentially using the fridge as the computer case. If you google it, you'll get lots of hits.)
Mandatory disclaimer: We don't recommend you use your phone or tablet for crunching. It's not built for it, and damaging (or destruction) of the device is within the realm of possibility.
You sound like what-if.xkcd :D!
To answer the question: on 2 out of 4 cores, 6 hours. Then a 1.5 hour charge. I need to push with PPS sv at this time anyway so it's serving as a boost. | |
|
|
|
|
When I try to run TRP sieve tasks on this host all I get is errors. Most of them almost immediately error, but one processed about 35% and then crashed. All tasks crash with the same error. Anyone have an idea what an app error: 9 is? How can I solve this?
Here is also some part of the log from the client,, if it could be helpful:
08-Jul-2013 21:12:19 [---] No config file found - using defaults
08-Jul-2013 21:12:19 [---] Starting BOINC client version 7.0.36 for arm-android-linux-gnu
08-Jul-2013 21:12:19 [---] log flags: file_xfer, sched_ops, task
08-Jul-2013 21:12:19 [---] Libraries: libcurl/7.28.0 OpenSSL/1.0.1c zlib/1.2.5.f-codeaurora-mods-v1
08-Jul-2013 21:12:19 [---] Data directory: /mnt/sdcard/Android/data/sk.boinc.nativeboinc/files/boinc
08-Jul-2013 21:12:19 [---] Processor: 1 ARM ARMv6-compatible processor rev 5 (v6l) @832MHz
08-Jul-2013 21:12:19 [---] Processor features: swp half thumb fastmult vfp edsp java
08-Jul-2013 21:12:19 [---] OS: Android: 2.6.35.7
08-Jul-2013 21:12:19 [---] Memory: 289.69 MB physical, 0 bytes virtual
08-Jul-2013 21:12:19 [---] Disk: 1.85 GB total, 294.88 MB free
08-Jul-2013 21:12:19 [---] Local time is UTC +3 hours
08-Jul-2013 21:12:19 [---] No usable GPUs found
08-Jul-2013 21:12:19 [PrimeGrid] Found app_info.xml; using anonymous platform
08-Jul-2013 21:12:19 [SETI@home] Found app_info.xml; using anonymous platform
08-Jul-2013 21:12:19 [---] Config: GUI RPC allowed from any host
08-Jul-2013 21:12:19 [PrimeGrid] URL http://www.primegrid.com/; Computer ID 403367; resource share 50
08-Jul-2013 21:12:19 [SETI@home] URL http://setiathome.berkeley.edu/; Computer ID 7017728; resource share 100
08-Jul-2013 21:12:19 [SETI@home] General prefs: from SETI@home (last modified 21-Jun-2013 14:44:10)
08-Jul-2013 21:12:19 [SETI@home] Computer location: school
08-Jul-2013 21:12:19 [---] General prefs: using separate prefs for school
08-Jul-2013 21:12:19 [---] Reading preferences override file
08-Jul-2013 21:12:19 [---] Preferences:
08-Jul-2013 21:12:19 [---] max memory usage when active: 144.84MB
08-Jul-2013 21:12:19 [---] max memory usage when idle: 260.72MB
08-Jul-2013 21:12:19 [---] max disk usage: 0.36GB
08-Jul-2013 21:12:19 [---] don't use GPU while active
08-Jul-2013 21:12:19 [---] (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
08-Jul-2013 21:12:19 [---] Not using a proxy
08-Jul-2013 21:12:19 [---] [battery detect] I cant detect battery!
08-Jul-2013 21:12:20 [---] Authorize monitor access
08-Jul-2013 21:12:35 [PrimeGrid] project resumed by user
08-Jul-2013 21:12:36 [PrimeGrid] Restarting task TRP_sieve_2736375_1 using trp_sr2sieve version 105 in slot 0
08-Jul-2013 21:15:29 [PrimeGrid] Computation for task TRP_sieve_2736375_1 finished
08-Jul-2013 21:15:29 [PrimeGrid] Output file TRP_sieve_2736375_1_0 for task TRP_sieve_2736375_1 absent
08-Jul-2013 21:16:53 [PrimeGrid] Sending scheduler request: To fetch work.
08-Jul-2013 21:16:53 [PrimeGrid] Reporting 1 completed tasks
08-Jul-2013 21:16:53 [PrimeGrid] Requesting new tasks for CPU
08-Jul-2013 21:16:55 [PrimeGrid] Scheduler request completed: got 1 new tasks
08-Jul-2013 21:16:58 [PrimeGrid] Started download of TRP_sieve_2728269_cmd
08-Jul-2013 21:17:00 [PrimeGrid] Finished download of TRP_sieve_2728269_cmd
08-Jul-2013 21:17:00 [PrimeGrid] Starting task TRP_sieve_2728269_2 using trp_sr2sieve version 105 in slot 0
08-Jul-2013 21:20:39 [PrimeGrid] Computation for task TRP_sieve_2728269_2 finished
08-Jul-2013 21:20:39 [PrimeGrid] Output file TRP_sieve_2728269_2_0 for task TRP_sieve_2728269_2 absent | |
|
|
|
|
I would suggest writing to the programmers of the Android software, NativeBOINC, because they would probably be better at debugging the issue.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1 | |
|
|
|
(Several years ago someone built a heavily overclocked desktop computer inside a cube fridge, essentially using the fridge as the computer case. If you google it, you'll get lots of hits.)
Many, many years ago (early 1980s?) I worked on the EMC testing of an Inmos multi processor device using about a dozen TRANSPUTERS. It was built in the alloy ice box from a fridge, with the cooling pipes still working and connected to the heat fins.
I redesigned the outer case to stop it wiping out half the radio spectrum and later presented a case study on it at Olympia in London. (No pun intended).
If that is what you were refering to then you have a good memory!!
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 | |
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1875
ID: 352
Credit: 2,990,621,692
RAC: 1,205,242
                         
|
|
There should be Android client 7.2 available sometimes next week released by UCB.
In general, there should be 7.2 clients for Win, Linux, Mac and Android.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
There should be Android client 7.2 available sometimes next week released by UCB.
In general, there should be 7.2 clients for Win, Linux, Mac and Android.
The Android TRP-Sieve client can now be downloaded automatically from the PrimeGrid server. Just tell the BOINC client on your Android device to connect to PrimeGrid and select TRP-Sieve in the project preferences on the website. It's no longer necessary to set it up manually using app_info on your Android device.
____________
My lucky number is 75898524288+1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
Looking at 4m40 per unit on 2 cores on a Note II here. | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
4h40 I mean...! | |
|
|
|
The Android TRP-Sieve client can now be downloaded automatically from the PrimeGrid server. ...
Any idea how much RAM is required as a minimum for the application to run? | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
The Android TRP-Sieve client can now be downloaded automatically from the PrimeGrid server. ...
Any idea how much RAM is required as a minimum for the application to run?
According to WUProp (http://wuprop.boinc-af.org/results/ram.py), it takes 18.5 MB.
WUProp, for those who are unfamiliar with it, is another BOINC project that gathers statistics about other BOINC projects by running Non-CPU-Intensive (NCI) tasks along side your other BOINC tasks. It's a great place for getting answers to questions like this.
____________
My lucky number is 75898524288+1 | |
|
|
|
The Android TRP-Sieve client can now be downloaded automatically from the PrimeGrid server. ...
Any idea how much RAM is required as a minimum for the application to run?
According to WUProp (http://wuprop.boinc-af.org/results/ram.py), it takes 18.5 MB.
WUProp, for those who are unfamiliar with it, is another BOINC project that gathers statistics about other BOINC projects by running Non-CPU-Intensive (NCI) tasks along side your other BOINC tasks. It's a great place for getting answers to questions like this.
I am not quite sure about this info. Actually, I notice jumps of RAM ussage from about 150MB to 260MB on this ARM device when I run the sieve application. I think it happens after the initialization period and seems to be the reason for the errors the device is producing (not enough memory). | |
|
|
|
|
Having come out, all guns blazing, in support of Android crunching earlier in this thread, I thought I'd better 'put my money where my mouth is', and give my (brand new!) Sony Xperia-Z tablet, and my (older) Samsung Galaxy-III-mini phone a whirl..
Pleased to say I was pleasantly surprised at how easy it all was.
The Xperia-z crunched a single long TRP-sieve task in around 23 hours, and is now crunching the shorter ones on 3 (out of 4) cores in around 4hours, 40mins.
The Samsung phone crunched a long task in around 37 hours, and takes around 7 hours for the short ones.
Looks to me like all those 'smoking-golf-ball' 'melting-battery' fears are largely unfounded!
My devices are not getting hot at all, as far as I can tell... Even with 3 cores crunching, the tablet struggles to heat up even a small patch of the rear-plate to anything higher than a 'pleasant warmth' (around 30-35degC, I reckon)
I was REALLY pleased to see the 'default' settings that disable crunching when battery-temp goes above 40degC , or when battery-charge goes below 90%, as well as the default of only crunching on 50% of the available cores.
These are pretty-much guaranteed to completely protect both the processor and the battery from meltdown under any conditions. (-but available for 'tweaking' by the brave!)
I would love to see some comparative "Boinc-Credit per Kilowatt" stats for Android vs Intel vs nVidia.. - I have a funny feeling the more modern Android ARMs are going to knock everything else out of the ballpark!
I challenge all those out there with an Android device - Especially if u have last-year's model gathering dust in a drawer! - to plug it into a charger, give it a go, and post your results here.
Just one small suggestion.. It might be a good idea to make TRP-Sieve one of those 'default' sub-projects that enable automatically when nothing else is selected - as there is no obvious way to directly select (the only) valid Android sub-project from the Manager on an Android device, which might confuse newbies who have never run PrimeGrid on a PC before..
____________
| |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Looks to me like all those 'smoking-golf-ball' 'melting-battery' fears are largely unfounded!
Hardly unfounded, they were serious enough that safeguards were built right into the BOINC client, as you noted:
I was REALLY pleased to see the 'default' settings that disable crunching when battery-temp goes above 40degC , or when battery-charge goes below 90%, as well as the default of only crunching on 50% of the available cores.
They did a really nice job of setting it up so it would be unobtrusive on a mobile device.
Just one small suggestion.. It might be a good idea to make TRP-Sieve one of those 'default' sub-projects that enable automatically when nothing else is selected - as there is no obvious way to directly select (the only) valid Android sub-project from the Manager on an Android device, which might confuse newbies who have never run PrimeGrid on a PC before..
Noted. We need to think about this.
____________
My lucky number is 75898524288+1 | |
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 900
ID: 3110
Credit: 174,849,845
RAC: 74,994
                
|
Just one small suggestion.. It might be a good idea to make TRP-Sieve one of those 'default' sub-projects that enable automatically when nothing else is selected - as there is no obvious way to directly select (the only) valid Android sub-project from the Manager on an Android device, which might confuse newbies who have never run PrimeGrid on a PC before..
Noted. We need to think about this.
I imagine that setting up the usual projects, but also setting "Send work from any subproject if selected projects have no work", should have the same result. I don't know if it will - I haven't tried it - but it should.
When I try to run TRP sieve tasks on this host all I get is errors. Most of them almost immediately error, but one processed about 35% and then crashed. All tasks crash with the same error. Anyone have an idea what an app error: 9 is? How can I solve this?
Yeah, I got some of these errors during the race. Anybody know the cause?
____________
| |
|
|
|
|
Thought I'd share this:
CPU-Z for Android
...it's not perfect at the minute but it doesn't cause any problems, it's quick and easy to install, and doesn't require much space. Could be very useful for anyone who's unsure of exactly what hardware they're running. | |
|
|
|
|
Hello All,
Couple of questions on TRP_Sieve for Android.
1. Has anyone actually found a factor or double checked one successfully?
2. Why is it when there is a factor, my device(s) say no factors file found?
Here's two examples.
Samsung Galaxy S4:http://www.primegrid.com/workunit.php?wuid=375936097
Droid Bionic: Workunit has yet to be completed by a 3rd host but I think the
results will be the same. Have a look.http://www.primegrid.com/workunit.php?wuid=376548708
Any help appreciated! | |
|
|
|
|
Seeing as this thread has been bumped, I might as well chime in with my experiences over six months of crunching on Android devices
As for mobile device (i.e., phones and tablets), I don't expect those to ever be a huge factor when compared to desktops or laptops. The power and cooling budgets on those devices just do not support hard core crunching. Yes, it can be (and has been) made to work, but these are, by definition, ultra low power devices and can't crunch very fast.
My Nexus 5 is significantly more powerful than my netbook, amusingly, so high-end phones can definitely be worthwhile. I'd love to see a GPU app for Android devices from some project, because the GPUs in high-end mobile devices can outperform a lot of CPUs, so they could be somewhat worthwhile.
Obviously they're not even comparable to high-end GPUs, but it's better than nothing.
Far worse, however, is the fact that they're not designed for crunching and have absolutely no cooling system at all. Cell phones overheat running normal apps (my iPhone occasionally shuts down to protect itself when running a navigation app, for example.) Running a full time 100% crunching program on a mobile device will likely overheat it, and on a device that isn't smart enough to shut down because of excessive temperature, destroying the device is a very real possibility.
(I know the iPhone 4 has thermal protection, but I'm fairly sure many Android phones do not. Crunching on a device without over-temp protection may destroy it.)
I live in Australia, where ambient temperature is rarely under 40°C in the summer. Despite this, even with CPU at 100% on my Nexus 4 and Nexus 5 (and, for a short time, a Samsung Galaxy S2 mini), I've never seen the battery exceed 55°C under natural circumstances, and they need to go far above 60° to suffer serious damage.
Obviously phones vary massively in SKU, but it seems that, at the least, Nexus devices are perfectly fine for BOINC, especially if you live in a country where it doesn't get up to 50 degrees in the shade. | |
|
AlezSend message
Joined: 30 Mar 13
Posts: 5
ID: 209838
Credit: 749,176,044
RAC: 17,387
              
|
|
Lot of scare mongering about melted phones etc and remember Android devices are a lot more common than just phones and tablets.
A lot of us crunch on Ordroid devices which do have a heat sink and a cooling fan. I personally use quad core android TV boxes. I strip the cases and leave the heatsink sitting on top of my desktop cruncher and they never get more than warm to the touch with boinc running on all 4 cores 100%. Also these devices don't have batteries and will run off a powered 5v hub all day, everyday. Mine run quite happily 4 to a hub with no problems.
Most Android devices are designed to run on low power and most automatically throttle the cpu load back as you push them harder, so protecting themselves from overheating. My boxes run at 1700 MHZ on 1 core which scales back as you use more cores to all 4 crunching at 1100 MHZ.
Far more phones die due to overcharging or alcohol / toilet drowning incidents than will ever be killed by Boinc :)
Currently on the native Boinc stats http://nativeboinc.org/site/host_stats
there are 39899 android devices, 15811 users and 1533 teams all using or have used Android devices to crunch. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
|
Don't forget that these forums cover a significant number of years, and things change. Technology improves. Earlier phones weren't always protected from thermal runaway, but I don't see that as a big problem anymore. Additionally, the BOINC client for android monitors temperature and even if nothing else did, that would protect the device.
I'm the one who was warning people at the beginning about the potential dangers. I'm also the one who installed the Android app at PrimeGrid. Times change.
____________
My lucky number is 75898524288+1 | |
|
|
|
|
Don't trust the nativeBOINC stats page. It misrepresents the true number of devices. It says I have 19 devices. I have only ever had 5 and only run 4 currently. All of my devices had been connected to multiple projects so cross id should have identified them. I even merged them when they became duplicates at various projects. Even after doing all of this, the stats still say I have almost 4 times the devices that I do.
I still run a couple of the older phones still running ARMv6 processors. That is all they do is run nativeBOINC (Berkeleys wont work on the older devices). No special heat sink or fan setups to keep them cool. (Vortex & Optimus V and previously the Ally before I gave it to my kid) I still run a Bionic (ARMv7) full time and my Galaxy S3 runs probably ~12 hours a day. It may be different for each device, but I think the scares were all based on skepticism rather than through demonstration.
I will chime in that though they do work and that there are billions of these devices, they wont make a huge dent until someone comes out with an affordable multi-socket rig for the hard core crunchers. Or the GPUs are finally usable with the energy efficiency making them a valid choice. Current devices are just too expensive to make it worth building a farm out of them unless it is just cycled out equipment or hobby gear. Billions of people still don't know these devices can be used and DC volunteers are a very small percentage of the population.
____________
 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
I'd love to see a rack full of phones :D! Could they all sort off 1 PSU just via the USB? | |
|
|
|
I'd love to see a rack full of phones :D! Could they all sort off 1 PSU just via the USB?
I have found that most phones can't keep a good enough charge off of the USB while crunching. All chargers are different and the phones will have different power needs as well. Mine stay plugged into a power strip. Most of these devices are more power efficient per their results and are worth using from an efficiency stand point. They just don't crank them out very fast. It certainly will take a lot of them to get most crunchers interested.
____________
| |
|
|
|
|
I loaded boinc on my new nexus 5. Three and a half to four hours per work unit. For some reason I thought it would take day or so. A pleasant surprise.
____________
| |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
I loaded boinc on my new nexus 5. Three and a half to four hours per work unit. For some reason I thought it would take day or so. A pleasant surprise.
We made the TRP_Sieve workunits one fifth as long as they used to be so they would run in a reasonable time on Android devices.
____________
My lucky number is 75898524288+1 | |
|
|
|
This isn't entirely a joke: perhaps you could try charging/crunching with the phone inside one of those little cube refrigerators? The phone should stay warm enough to avoid condensation, and the fridge will keep the phone from overheating.
Condensation can easily be handled by wrapping the phone in a single layer of plastic wrap. If really paranoid stick a desiccant pack inside the plastic with the phone - but thats the easiest way to handle it. If condensation forms, it forms on the plastic, not your device! It's also a great tip for photographers. When winter comes, the first thing I do is stuff my camera bag with ziplock bags and all my lenses and my camera go in them. That way if I take my camera out and then come inside, I'm automatically protected. There's nothing worse than getting a permanent spot on your lense that ends up with you having to ship it off for a couple of hundred bucks to be repaired :( | |
|
|
|
|
Please let me know if (and when) you find a factor (validated) with your Nexus.
I was running them on my S4 and Motorola Droid Bionic a while back until I saw the "validation inconclusive" message when I actually found a factor. I think TRP_Sieve for android is broken, or the validation process, and a waste of time personally. Had no response to my post earlier so it leads me to believe that it is compromised.
Good luck!
John | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Please let me know if (and when) you find a factor (validated) with your Nexus.
I was running them on my S4 and Motorola Droid Bionic a while back until I saw the "validation inconclusive" message when I actually found a factor. I think TRP_Sieve for android is broken, or the validation process, and a waste of time personally. Had no response to my post earlier so it leads me to believe that it is compromised.
Good luck!
John
Sorry about that. Sometimes, nobody thinks they're the best person to answer a question, and nobody ends up answering it. It's not a terrible thing to ask again if you don't get a response.
1. Has anyone actually found a factor or double checked one successfully?
Certainly many factors have been found. They're far from uncommon. But I think you're asking if any Android devices have found factors. That's not as easy a question to answer as it might seem, unfortunately. It's not something I can just pull out of the database.
What I can tell you is that at this moment, there's 6877 Android tasks in the database that have been returned, validated, and received credit. That's not the total of all Android work, of course; the tasks are purged after 6 days, so roughly there's about 1000 returned each day.
Your questions will take a bit of analysis.
2. Why is it when there is a factor, my device(s) say no factors file found?
I'm not familiar with running BOINC on Android. Are you saying that the app tells you it found no factors? And how do you know there was a factor? Can you explain?
____________
My lucky number is 75898524288+1 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
|
I have turned off the Android app while we investigate this problem. Tasks already sent to your device may continue to be processed, but no new tasks will be sent out.
____________
My lucky number is 75898524288+1 | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
Surely with a fridge the door would have to remain ajar due to recharging cable. Or cables - I'm thinking a farm is possible here :D! | |
|
|
|
I have turned off the Android app while we investigate this problem. Tasks already sent to your device may continue to be processed, but no new tasks will be sent out.
Perhaps you should post a notification as a news item.
BOINC for Android has just been updated with all sorts of flashy stuff (including news alerts!) and when all my Android devices stopped getting tasks, (after trashing a whole bunch of them!) I assumed the new BOINC update was broken!
Any idea when things might be up and running again? - No Pressure! - but if its going to be more than a day or two, I'll get them working on a different project while we wait..
____________
| |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
I have turned off the Android app while we investigate this problem. Tasks already sent to your device may continue to be processed, but no new tasks will be sent out.
Perhaps you should post a notification as a news item.
I was hoping to find out a bit more about the nature of the problem and perhaps how long it will take to get fixed, but I may need to make an announcement without having a whole lot to say.
Any idea when things might be up and running again? - No Pressure! - but if its going to be more than a day or two, I'll get them working on a different project while we wait..
At this point, I have no information at all on how long it will take.
My advice would be to put them on something else. I don't have an ETA, but even in the most optimistic of scenarios it will take another day or two.
____________
My lucky number is 75898524288+1 | |
|
|
|
|
Thank you Michael for taking a look at this issue. My S4 spent many nights sleeping on an icepack for trp_sieve! (Although it didn't need it, with the thermal protection, the 4 cores ran at max on ice). It must be something with the validation process, just a guess though.
John
| |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Thank you Michael for taking a look at this issue. My S4 spent many nights sleeping on an icepack for trp_sieve! (Although it didn't need it, with the thermal protection, the 4 cores ran at max on ice). It must be something with the validation process, just a guess though.
John
The validation process is fine. The app is either never finding factors, or if it does, the output it sends back to the server says it didn't find any factors. Either way, it's never sending factors to the server.
____________
My lucky number is 75898524288+1 | |
|
|
|
|
Definitely an app issue then, hopefully solved soon. With phones and other devices running faster than computers 3 yrs ago, is LLR possible on android if an app is created? (and android devices get a Huge cache upgrade via 64bit cpu's)
John | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Definitely an app issue then, hopefully solved soon. With phones and other devices running faster than computers 3 yrs ago, is LLR possible on android if an app is created? (and android devices get a Huge cache upgrade via 64bit cpu's)
John
LLR is all double precision floating point, which a lot of ARM processors don't support at all, and if they do, they'll be much slower than modern desktop CPUs. AVX is a set of vector processing instructions that allow the CPU to do 4 floating point operations simultaneous.
That's why only the sieves are good candidates for Android, and the PPS Sieve is sized for GPUs. That leaves just the TRP sieve.
____________
My lucky number is 75898524288+1 | |
|
|
|
Definitely an app issue then, hopefully solved soon. With phones and other devices running faster than computers 3 yrs ago, is LLR possible on android if an app is created? (and android devices get a Huge cache upgrade via 64bit cpu's)
John
LLR is all double precision floating point, which a lot of ARM processors don't support at all, and if they do, they'll be much slower than modern desktop CPUs. AVX is a set of vector processing instructions that allow the CPU to do 4 floating point operations simultaneous.
That's why only the sieves are good candidates for Android, and the PPS Sieve is sized for GPUs. That leaves just the TRP sieve.
Good point, I forgot about that. | |
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2765
ID: 130544
Credit: 919,293,227
RAC: 24,213
|
|
Any update? Thanks. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Any update? Thanks.
There is no estimate at this time for when, or even if, this app will be fixed. The original developer can no longer work on the application.
____________
My lucky number is 75898524288+1 | |
|
|
|
|
Since the Android App is indefinitely disabled, can TRP be put back to the larger size sieve it was?
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Since the Android App is indefinitely disabled, can TRP be put back to the larger size sieve it was?
It's a bit more complicated than simply flipping a switch, so I'd prefer to keep it where it is. I would like to see it fixed, and don't want to switch it back and forth.
____________
My lucky number is 75898524288+1 | |
|
|
|
The original developer can no longer work on the application.
How can I help?
____________
 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
The original developer can no longer work on the application.
How can I help?
We have a broken Android app, and nobody to fix it. If you're able to fix it, that would help. The original source code for the x86 sieve is pointed to in the "Source Code Repositories" sticky thread in this topic. The broken source code for the android version I'd have to dig up, but I'll do so if you want to take a stab at fixing it.
____________
My lucky number is 75898524288+1 | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
|
Here's the link for the Android source code:
https://github.com/matszpk/native-boinc-for-android
____________
My lucky number is 75898524288+1 | |
|
|
|
with cellphone's now having gpu's and old dying systems being replaced with apu bearing motherboards the rate of gpu supporting systems is going exponential.
Are these time frames realy going to holed?
Of course not. :)
The estimates are the best bet for what we know now.
Certainly, computers will get faster as time goes on, but not always, and it's impossible to predict how much improvement there will be. For example, the Kepler series (GTX 6xx) of GPUs are a huge improvement for gamers, but are very disappointing for us number crunchers and actually do many of our kind of computations slower than the previous generation of GPUs.
As for mobile device (i.e., phones and tablets), I don't expect those to ever be a huge factor when compared to desktops or laptops. The power and cooling budgets on those devices just do not support hard core crunching. Yes, it can be (and has been) made to work, but these are, by definition, ultra low power devices and can't crunch very fast.
Far worse, however, is the fact that they're not designed for crunching and have absolutely no cooling system at all. Cell phones overheat running normal apps (my iPhone occasionally shuts down to protect itself when running a navigation app, for example.) Running a full time 100% crunching program on a mobile device will likely overheat it, and on a device that isn't smart enough to shut down because of excessive temperature, destroying the device is a very real possibility.
(I know the iPhone 4 has thermal protection, but I'm fairly sure many Android phones do not. Crunching on a device without over-temp protection may destroy it.)
For both reasons, I don't expect phones and tablets to ever play a significant role here.
Couldn't Android phones ever be used for basic sieving, and then let the desktops do all the prime number calculations? Also, a number of android devices slow down the CPU when it gets too hot. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
with cellphone's now having gpu's and old dying systems being replaced with apu bearing motherboards the rate of gpu supporting systems is going exponential.
Are these time frames realy going to holed?
Of course not. :)
...
Couldn't Android phones ever be used for basic sieving, and then let the desktops do all the prime number calculations? Also, a number of android devices slow down the CPU when it gets too hot.
You're quoting me from a post I wrote nearly three years ago. In the intervening years we tried an Android sieving app. If it actually worked we'd probably still be using it. :)
____________
My lucky number is 75898524288+1 | |
|
|
|
with cellphone's now having gpu's and old dying systems being replaced with apu bearing motherboards the rate of gpu supporting systems is going exponential.
Are these time frames realy going to holed?
Of course not. :)
...
Couldn't Android phones ever be used for basic sieving, and then let the desktops do all the prime number calculations? Also, a number of android devices slow down the CPU when it gets too hot.
You're quoting me from a post I wrote nearly three years ago. In the intervening years we tried an Android sieving app. If it actually worked we'd probably still be using it. :)
Whoops. Well, ok then. | |
|
|
|
|
Hello all:
I hate to revive a dead post, but I was wondering - has Android been given up on? If so, it should probably be removed from the list of BOINC projects that support it in the "choosing BOINC projects" on the platform's homepage, and the server set to notify Android users that these devices are not supported.
If still working on it / refining etc, carry on! My apologies if I've missed a newer conversation elsewhere. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13299
ID: 53948
Credit: 224,092,485
RAC: 32,608
|
Hello all:
I hate to revive a dead post, but I was wondering - has Android been given up on? If so, it should probably be removed from the list of BOINC projects that support it in the "choosing BOINC projects" on the platform's homepage, and the server set to notify Android users that these devices are not supported.
If still working on it / refining etc, carry on! My apologies if I've missed a newer conversation elsewhere.
Please read this thread from August: http://www.primegrid.com/forum_thread.php?id=6960
Bottom line: There's a mechanism that detects which sites support Android. We have that turned off. On the official BOINC site, we do not show up as supporting Android.
And yet, on Android devices we appear as an site that offers Android apps. So does at least one other BOINC project that never has offered Android apps, and likely never will.
I have no idea why PrimeGrid shows up on there. It shouldn't.
____________
My lucky number is 75898524288+1 | |
|
Message boards :
Number crunching :
Android |
