Parallella Community

Hi,

this is my first post in the parallella forums so i want to say HELLO to all people interested in new hardware like parallella

Since the sucessfull end of the kickstarter project i sit here and wait for my board
Till now i only had spare time, so i only read the project news and some of the threads within this forum but since i finished my studies and was able to gain some ground in the "real" working work i would spend some time supporting the opensource world (hard- and software )

I am a boinc user for years and fotunately a c++ coder too . I am looking for projects to get my hands dirty, and think porting one or another boinc-client sounds really interesting.
So if someone really starts some work in that direction, please keep me informed => i'm in

i'll stay tuned and hope the board will arrive soon

kind regards
christian

Hi Andrew,

yeah it sounds great. But i have to admit that i am new to the whole opencl/multicore-programming thing...
thats one of the reasons why i wanted to have this board, to start making some experiences in that field, so please don't expect too much.

So if there are some people willing to talk about what they've done i can learn faster.. but i'll try it anyway so yeah, it would be great if you could get me in touch

kind regards
christian

Hi

I'm a software developer with the Einstein@Home project (on astrophysics, relativity etc). Together with apps for Android, we provide a Raspberry Pi app and are currently testing (on our test project, Albert@Home: http://albert.phys.uwm.edu ) an ARMv7 Linux app with support for NEON. Unfortunately I don't have my Parallella yet (backer # 5000ish, might take a while....), so anybody with a prototype board is welcome to try it.

Of course the ARMv7-NEON version we have now does not use the Epiphany. The source code (including an OpenCL version we use for NVIDIA, ATI/AMD and (integrated) Intel GPUs) is open source, tho.

As much as I would like to see the Einstein@Home app optimized for the Epiphany chip, I'm really skeptic this would speed up the app: it spends most of the time doing really big FFTs (on the order of 2^22, so about length= 4 million, real to complex, single precision FFT). With the combination of a not-so-fast (wrt throughput) interface between the RAM and the Epiphany, small internal RAM of Epiphany and low (O(n log n) ) complexity of FFT, my gut feeling is that it will be very hard to implement FFT faster in Epiphany than in the ARM CPU. Right??

Any comments welcome.

Cheers
HBE

Hi Mike

Good to see you here (as expected ).

As for FFT lengths, the one app that is currently ported to ARM ("BRP4") actually uses transform length of 3*2^n . The original data (a time series of radio telescope samples data) has length 2^n but to improve frequency resolution , it is padded at the end with twice as much data for a total of 3*2^n before being FFTed. Our CPU versions use FFTW which can deal with this sort of length natively, CUDA's cuFFT can also smallow it, and for OpenCL we break up the task into three length 2^n transforms and then we twiddle together the results in custom kernels on the GPU. So if we had a way to do general length power of two transforms on the Epiphany, we would be fine anyway, the rest is simple.

Other parts of BRP4 are perhaps better suited for the Epiphany (e.g. the "resampling" stage where we de-modulate the Doppler effect from the time series [due to the radio source being in an assumed close binary system]), but unfortunately this step is less computing intensive compared to the FFT step.


If I had to put my money on a single BOINC project where one might get the best benefit from Epiphany with the least effort I would go for Enigma@Home (w/o having had a look at the source code, I must admit). The project deals with cryptanalysis of WW II German U-boat radio traffic. The messages are very short and there will be some brute force parallelism involved that should map well to the Epiphany (unless it makes use of huge precomputed look up tables, perhaps).

Cheers
HB

bikeman and hewsmike,

I crunch for einstein and became interested in the parallela board for that purpose...so I'm here to encourage your efforts! If you manage to get anything up and running at least one other person will be executing that code

jon