This one is looking a lot faster. Normally they take 47 hours, currently they are looking around 31-32 hours. A substantial improvement.

It still hasn't quite finished the 1st one yet (its on 96% after 30 hours). I will provide links once it completes.

Excellent, thanks!

The obvious question is: will it be possible to improve the Android app versions in a similar way? That remains to be seen(I'll need to make a wrapper app first that produces and dumps the FFTW3 'wisdom' on an Android device. Stay tuned.

Cheers
HBE

---

It still hasn't quite finished the 1st one yet (its on 96% after 30 hours). I will provide links once it completes.

Now its finished. 31 hours and 45 minutes. Link to wu is here

It still hasn't quite finished the 1st one yet (its on 96% after 30 hours). I will provide links once it completes.

Now its finished. 31 hours and 45 minutes. Link to wu is here

Cool, validated against an Android (NEON) task that took a bit less than half as long to finish.

Cheers
HB

---

The rest of them have also knocked over one each (I have 5 Pi's on the job). They seem to be a consistent 31 hours and 35 to 40 minutes. Great job guys :-)

This long weekend I made some tests for speed comparison.
My old, retired Intel Atom 2x1.8GHz has seen a new spring. A clean win7 with actual Boinc Manager and a project which is availanble for both platforms and validating against each other.
Pogs.
My Nexus runs these apps in ~8 hrs, using 2 of the 4 cores it finishes ~6 wu's within 24 hrs.
The Atom is running 4 threads, for 16 hrs now. The fastest wu has reached now 64%, the slowest 41%. Even my phone is faster.
My i3 3220 needs ~6000 sec's/wu running 4 threads.

I think ARM, Parelela & Co are the future. In terms of speed, power consumption, amount of devices available and headroom for improvements.
All it needs is some advertising for arm-projects. They can easily replace the older laptops and desktops.

And the Atom? I think, the spring will end soon ...

---

Hi!

I think it's too early for funeral eulogies wrt. Intel Atom. There are (more than) rumors that Samsung will use Atoms in some future tablet and possibly even smart phones. I guess they know what they are doing. We'll see.

Cheers
HB

---

I need to make a correction. The crunching time stabilized in the range of 10 - 11 hrs now; maybe some background tasks after the new installation have slowed down the system (virus check, defragmentation, update-check aso).

My atom is a D525 which is certainly not the one Samsung will choose for the new tablets. New devices have a lot of enhancements on board; my one is 45nm, L1 cache 24/32k, L2 512k, L3 0k.
AFAIK Apple has plans to use ARM devices for Laptops. It looks like we will see a future with a wide variety of cpu's and platforms. Will not make it easier for software developement!

---

I notice the Raspberry Pi app has updated to 1.06 now. Were there problems with 1.05 or some further improvements?

I notice the Raspberry Pi app has updated to 1.06 now. Were there problems with 1.05 or some further improvements?

There was a small bug fix, and the new app was compiled with a different toolchain (cross compiled on an Intel x86 Linux host like the Android (and Windows) app versions). Before that, it was a bit cumbersome to build the app on the Raspberry Pi itself, and it didn't integrate well into our automated continuous integration solution (Jenkins).

I expect no significant performance change.

Cheers
HB

---

I notice the Raspberry Pi app has updated to 1.06 now. Were there problems with 1.05 or some further improvements?

<snipped>

I expect no significant performance change.

Cheers
HB

I'd say slightly faster, around 500 seconds approx from the small sample completed so far.

I've attached my HTC One S to Albert@home, not had any validations yet, eithier the app gets a SIGSEGV: segmentation violation, or if it completes, it then gets a Validate error, running the Official Android 7.0.58 Boinc,
one thing to Note, my Android kernel is newer than anyone else's here, being Android 3.4.10-g8d7cd09, it has some new Power Save Options (which i have switched off, but i don't trust the kernel not to do other power save things unbided)

Computer 7471

I was previously running NativeBoinc on the HTC One S, Both the Setiathome app and the PrimeGrid apps would also regularly get segmentation violations too,
the GridGrid app would sometimes get app error: 9, i think this is when android kills the app to free up memory, also often one of the PrimeGrid Wu's would error if:
a) the HTC switched over to battery power, b) If the Play Store downloaded and then updated any apps, it wasn't the downloading that caused the errors, more the updating,
earlier when i ran the older kernel (I think it was around 3.0.x), i'm sure i only got the app error: 9 condition,

Claggy

I've added an Asus EE-pad two days ago.
For some reason I installed native Boinc on SD-card. The result was: almost all wu's (@ all projects) failed.
[22:43:06][32002][ERROR] Input file on command line ../../projects/albert.phys.uwm.edu/p2030.20120226.G194.26-02.01.S.b6s0g0.00000_1510.bin4 doesn't agree with input file ../../projects/albert.phys.uwm.edu/p2030.20120226.G194.26-02.01.S.b6s0g0.00000_1492.bin4 from checkpoint header.
[22:43:06][32002][ERROR] Demodulation failed (error: 2)!

Yesterday I moved the app back to internal memory; the first wu finished now, waiting for validation.

Will report that to native boinc as well.

---

I've attached my HTC One S to Albert@home, not had any validations yet, eithier the app gets a SIGSEGV: segmentation violation, or if it completes, it then gets a Validate error, running the Official Android 7.0.58 Boinc,
one thing to Note, my Android kernel is newer than anyone else's here, being Android 3.4.10-g8d7cd09, it has some new Power Save Options (which i have switched off, but i don't trust the kernel not to do other power save things unbided)

Computer 7471

I was previously running NativeBoinc on the HTC One S, Both the Setiathome app and the PrimeGrid apps would also regularly get segmentation violations too,
the GridGrid app would sometimes get app error: 9, i think this is when android kills the app to free up memory, also often one of the PrimeGrid Wu's would error if:
a) the HTC switched over to battery power, b) If the Play Store downloaded and then updated any apps, it wasn't the downloading that caused the errors, more the updating,
earlier when i ran the older kernel (I think it was around 3.0.x), i'm sure i only got the app error: 9 condition,

Claggy

Thanks for the report!

I see one, maybe two other devices in the DB that might have similar issues. I'll have to investigate. Thanks for reporting this!!
Cheers
HB

---

I have added an Odroid-u2 yesterday. The first wu finished now, runtime 36774 sec. Will try different projects first and later a 14 day Albert-only session to see tha RAC with 4 cpu's active.
Boinc sees it as an Arm v7 @ 1704MHz.

---

I've attached my HTC One S to Albert@home, not had any validations yet, eithier the app gets a SIGSEGV: segmentation violation, or if it completes, it then gets a Validate error, running the Official Android 7.0.58 Boinc,
one thing to Note, my Android kernel is newer than anyone else's here, being Android 3.4.10-g8d7cd09, it has some new Power Save Options (which i have switched off, but i don't trust the kernel not to do other power save things unbided)

Computer 7471

I was previously running NativeBoinc on the HTC One S, Both the Setiathome app and the PrimeGrid apps would also regularly get segmentation violations too,
the GridGrid app would sometimes get app error: 9, i think this is when android kills the app to free up memory, also often one of the PrimeGrid Wu's would error if:
a) the HTC switched over to battery power, b) If the Play Store downloaded and then updated any apps, it wasn't the downloading that caused the errors, more the updating,
earlier when i ran the older kernel (I think it was around 3.0.x), i'm sure i only got the app error: 9 condition,

Claggy

Thanks for the report!

I see one, maybe two other devices in the DB that might have similar issues. I'll have to investigate. Thanks for reporting this!!
Cheers
HB

Any idea on the reason for the Validate error yet? Is it a Android Boinc Bug, an App Bug, a Server problem, or some other reason?

Only one more segmentation violation since my last report, rest of Wu's all Validate error or pending.

Claggy

Any idea on the reason for the Validate error yet? Is it a Android Boinc Bug, an App Bug, a Server problem, or some other reason?

Only one more segmentation violation since my last report, rest of Wu's all Validate error or pending.

Claggy

No idea yet :-(. Since this seems to affect other projects as well, I guess it's most likely not a science app problem. I also can't think of a way how a BOinc server side problem could trigger this. Strange..

HB

---

My HTC One S has now been sent a v1.06 (VFP) Wu, we'll see how that gets on.

Claggy

My HTC One S has now been sent a v1.06 (VFP) Wu, we'll see how that gets on.

Claggy

That app worked, shame I've only had one of those Wu's so far.

https://albert.phys.uwm.edu/result.php?resultid=903411

Claggy

Installed the Boinc 7.1.10 Alpha client from Google Play, shame it gave me a new host ID and didn't carry on from the last installation.

We'll soon see if the Neon app is any different under this client.

Claggy

Same validate errors with the v1.06 (NEON) app on Boinc 7.1.11:

All tasks for computer 7663

Hopefully i'll get a v1.06 (VFP) Wu soon.

Another way to note, all those validate errors didn't drive my Max tasks per day down, should it?

Claggy