WARNING: This website is obsolete! Please follow this link to get to the new Albert@Home website!
BRP4U (Raspberry Pi , single DM tasks) feedback thread |
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
Patrick Send message Joined: 13 Mar 13 Posts: 18 Credit: 5,642 RAC: 0 |
Ok thanks for that info I think then it´s not especially good for live crunching but for testing it´s ok. I´ve send you my log of nativeboinc to MPI... i thought there´s something wrong with my device or firmware :) delete it whatever |
Alex Send message Joined: 1 Mar 05 Posts: 88 Credit: 398,734 RAC: 0 |
Is NEON a standard feature of the arm7 devices? My task list shows that they are crunching 1.03 NEON apps, but computer details don't say anything about NEON. edit: wikipedia says NEON is standard on cortexA8 and optional on cortexA9 but no word about arm7. |
Bernd Machenschalk Volunteer moderator Project administrator Project developer Send message Joined: 15 Oct 04 Posts: 1956 Credit: 6,218,130 RAC: 0 |
computer details don't say anything about NEON It is not visible in "Computer Details", but when you click on the "last contact" date of your device, in the scheduler log you see: [send] CPU features: swp half thumb fastmult vfp edsp thumbee neon vfpv3 tls vfpv4 So it's correctly detected by the Client and communicated to the Server. BM |
Alex Send message Joined: 1 Mar 05 Posts: 88 Credit: 398,734 RAC: 0 |
THX for the info. I took a look into it and found an entry there: 2013-04-23 17:21:43.1669 [PID=15276] [CRITICAL] No filename found in [WU#197956 p2030.20120226.G194.26-02.01.S.b6s0g0.00000_868] I found an similar entry also on my other arm device. |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
Is NEON a standard feature of the arm7 devices? Hi One has to be careful about the ARM nomenclature: ARMvn (as in ARMv6, ARMv7) designates a generation of CPU architecture. ARMn (as in ARM7 or ARM11) designates a family of ARM processors belonging to the same architecure, but an architecture can comprise more than one family (it's more like individual models or lines of similar models). So for example the processor of the Raspberry Pi is actually an ARMv6 (architecture) and at the same time an ARM11 (family) CPU.... This is further complicated by the fact that ARM (the company) is not really a chip manufacturer: they do not produce chips themselves (like Intel does), they just license the design to others (Apple, Samsung, NVIDIA, ...whoever) who then build their own CPUs. The ARM designs are modular, so you can add certain optional modules from the ARM "design portfolio" (like NEON) to your CPU or not, so NEON will be on some (most) ARMv7 CPUs but not on others. Cheers HB |
Alex Send message Joined: 1 Mar 05 Posts: 88 Credit: 398,734 RAC: 0 |
Is NEON a standard feature of the arm7 devices? Oh, so they found a way to keep it 'simple' ... Thanks a lot for the info! |
Saenger Send message Joined: 15 Feb 05 Posts: 361 Credit: 600,993 RAC: 0 |
My Galaxy S2 needed 102h for the first one, a v1.02, for some meagre 63 credits. But it gave me the required 100h for bronze in WUProp in a single WU, that's fine ;) I just reseted the project and got a v1.03 (NEON), and it's now, that's after 1:36h, at 13%. So this one will take about 12h, an more than eightfold decrease in crunch-time! Grüße vom Sänger |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
My Galaxy S2 needed 102h for the first one, a v1.02, for some meagre 63 credits. But it gave me the required 100h for bronze in WUProp in a single WU, that's fine ;) cool!! To be fair, the newest non-NEON app version, "1.03 (VFP)" seems to be twice as fast as the old 1.02 , so that would mean a factor of 4 between NEON and non-NEON on the same hardware. Still impressive. I've even seen a Galaxy S2 do a task in about 9hrs when crunching on one core only. When crunching on 2 cores, the phone needed some additional cooling (I think Bernd took a photo of the cooling solution :-) ) to keep it below 45 deg C (the level configured in the test to trigger app suspension for thermal protection). Personally I think a (CPU) runtime of up to ca 16 hrs per task would be ok. This would mean that if the device is allowed to crunch for only a few nights per week while connected to a charger, it would easily make the deadline of 14 days for BRP4(U) tasks. Thanks for the feedback Cheers HB |
Bernd Machenschalk Volunteer moderator Project administrator Project developer Send message Joined: 15 Oct 04 Posts: 1956 Credit: 6,218,130 RAC: 0 |
cool!! BM |
Patrick Send message Joined: 13 Mar 13 Posts: 18 Credit: 5,642 RAC: 0 |
That´s the funniest cooling solution i´ve ever seen. :) But for those who haven´t so a small fan at home be warned... Don´t forget your Smartphone in the Refrigerator if you go to work! :) |
Patrick Send message Joined: 13 Mar 13 Posts: 18 Credit: 5,642 RAC: 0 |
First Valid result on my Galaxy S 1 mini with app 1.03(VFP) http://albert.phys.uwm.edu/results.php?hostid=6955&offset=0&show_names=0&state=3&appid= long runtime but valid :) |
Alex Send message Joined: 1 Mar 05 Posts: 88 Credit: 398,734 RAC: 0 |
Which brings up the question whether it is worthwhile to support non-NEON Android devices at all. I found this today in the Hardkernel Forum: 1,000,000 BOINC credits on ARM devices http://forum.odroid.com/viewtopic.php?f=7&t=43 And NativeBoinc News says New SETI@Home app version Added by: matszpk , 2013-04-29 21:45 This version (0.1.1) of SETI@Home application brings new support for VFP based on LibFFTS (instead PFFFT) and improved support for NEON. You can expects 20%-30% speedup on a ARM processors with VFP. Could this help our app as well or is it already in use? |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
It's interesting to read, in the discussion on SETI@Home, that libFFTS is considerably faster than FFTW for ARM. So yes, this could be interesting. FFTW was our natural choice for the initial app version on ARM because we use it for all CPU versions so far. But that doesn't mean it has to be that way, e.g. the BRP4 GPU versions use different FFT libs, obviously, for CUDA and OpenCL. And all those versions validate against each other quite happily. OTOH the libFFTS seems to be quite new and maybe it needs to settle and mature a bit longer. Thanks for the pointer Cheers HB |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
First Valid result on my Galaxy S 1 mini with app 1.03(VFP) Congrats :-) Good to have some ARMv6 testers here! Given the speed of the devices, it takes some time to get enough validated results (say, around 200) that you will want to see before being confident that cross validation with ARMv6 works. I have one of the few other ARMv6 Android devices on Albert and the run-time is roughly the same. Cheers HB |
Patrick Send message Joined: 13 Mar 13 Posts: 18 Credit: 5,642 RAC: 0 |
Ok thank you I´ve buyed me an S3 before yesterday but not for crunching it´s such a great fast Smartphone. But everyone has to decide this by himself. I wish you many fast mobile devices if the project is released to Einstein. You all are doing good work. |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
Hi fellow Raspberry Pi fans ;-) I have released a new test version (1.04) which should be quite a bit faster than the previous one. In general the performance should be somewhere between that of the Android ARMv6 (VFP) 1.03 version and the ARMv7 (NEON) 1.03 version (scaled by clock rate). There were no fundamental changes, just compilation option settings. If I find the time to write API wrappers, I will later try the FFTS library which is now used on Android over at SETI@Home, but for now, the version sticks to the good old FFTW. Cheers HB |
MarkJ Send message Joined: 28 Feb 08 Posts: 26 Credit: 28,160 RAC: 0 |
Hi fellow Raspberry Pi fans ;-) Err no. First one completed in about the same time as the 1.01 versions (168,000 seconds). I know its a bit early to tell seeing as its only the 1st one, will keep an eye on the others as they complete. Link to task is here |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
Hi! yeah, I think you are right. Doesn't seem faster at all. Maybe I didn't turn off the FFT wisdom file when I tested this one....(including a wisdom file is another thing I'm working on and this WOULD speed up things indeed). So stay tuned. Thanks for testing Cheers HB |
Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1483 Credit: 1,864,017 RAC: 0 |
OK, next attempt .... I just released version 1.05 for Raspberry Pi . If all goes well this should be 20 - 25 % faster than the previous version. Feel free to cancel task that are/will be running under version 1.04, this will help to get 1.05 results in soon. Thanks for testing!! Cheers HB |
MarkJ Send message Joined: 28 Feb 08 Posts: 26 Credit: 28,160 RAC: 0 |
OK, next attempt .... I just released version 1.05 for Raspberry Pi . If all goes well this should be 20 - 25 % faster than the previous version. 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. |