parallela

[Dean Hall]

In case you guys haven't heard, a Kickstarter project to make a massively multi-core computer core is experimenting with p14p: https://www.linux.com/news/enterprise/systems-management/692990-introducing-the-99-linux-supercomputer

!!Dean

[pir...@gmail.com]

Oh my godness, and I was thinking about port it to the barebone
Rapsberry Pi... this is fairly better!!! :-D

2013/1/27 Dean Hall <dwha...@gmail.com>:

> In case you guys haven't heard, a Kickstarter project to make a massively multi-core computer core is experimenting with p14p: https://www.linux.com/news/enterprise/systems-management/692990-introducing-the-99-linux-supercomputer
>
> !!Dean
>

> --
> --
> You are subscribed to the "python-on-a-chip" (or p14p for short) Google Group.
> Site: http://groups.google.com/group/python-on-a-chip
>
>
>



--
"Si quieres viajar alrededor del mundo y ser invitado a hablar en un
monton de sitios diferentes, simplemente escribe un sistema operativo
Unix."
– Linus Tordvals, creador del sistema operativo Linux

[John Griessen]

On 01/26/2013 06:00 PM, Dean Hall wrote:
> In case you guys haven't heard, a Kickstarter project to make a massively multi-core computer core is experimenting with p14p: https://www.linux.com/news/enterprise/systems-management/692990-introducing-the-99-linux-supercomputer

They have a "goal of democratizing access to parallel computing." and they mention having lots of language and
app developers.

Sounds great! Maybe we'll have a python programmable neural net connection machine soon!

Sounds better than depending on neural net hardware functions, since you can rearrange everything in
SW, and keep using the same hardware longer without dead ends, (until it is eclipsed by even better
performance at least).

[Gmail - neonmark]

If I read it right there is a dual core A9 with an epiphany side
processor with 16-64 cores of 32kb each.
If the assumption is to run p14p on those cores then there's not much
left for program. The 1GB of system ram may not be available for them
all at the same time.
of course the new 28nm chip announced in Aug is 800MHz (100GFLOPS) so in
that case maybe I don;t care and I want it anyway...(but only 32Kb each)

Normally I'm all for this kind of thing but it looks a bit messy...

It seems to me that putting p14p on the raspberry pi is a better
solution to general purpose uses.
Also to make it work in parallel will require writing special s/w to
parrallelise(sp?) tasks as python is not internally designed to be
spread around like that...

Thoughts ?

[pir...@gmail.com]

Maybe adding support for the processing module?

Sent from my Android cell phone, please forgive the lack of format on the text, and my fat thumbs :-P

--
--
You are subscribed to the "python-on-a-chip"  (or p14p for short) Google Group.
Site: http://groups.google.com/group/python-on-a-chip

--- You received this message because you are subscribed to the Google Groups "python-on-a-chip" group.
To unsubscribe from this group and stop receiving emails from it, send an email to python-on-a-chip+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[ed2k]

I wish we can have something similar to apple's GCD. that means a piece of context/code is enclosed. it is portable and put into queue. the host can decide send this piece of block to the epiphany co-processor.  

To unsubscribe from this group and stop receiving emails from it, send an email to python-on-a-ch...@googlegroups.com.

[pir...@gmail.com]

In fact it would be possible: a queue of pointers to an structure with a pointer to a function and their required arguments and you are go, and in Python it would be easier: a queue of binded functions :-D The problem would be with contexts, since here we could have race conditions, and also with the order of returned results, but for highly parallel aplications like OpenCL or CUDA or GPGPU it would be feasable.

Sent from my Android cell phone, please forgive the lack of format on the text, and my fat thumbs :-P

[John Griessen]

On 01/31/2013 05:45 AM, Gmail - neonmark wrote:
> If the assumption is to run p14p on those cores then there's not much left for program.

They have other parallel languages in mind for most of the cores. Probably they want python
on just one of them. Here's a parallela feature list from
http://tested-charismatically.blogspot.com/2012/09/insert-coin-parallella-project-dreams.html

"a dual-core ARM A9 CPU running Ubuntu OS as standard, 1GB RAM, a microSD slot,
two USB 2.0 ports, HDMI, Ethernet and a 16- or 64-core accelerator, with
each core housing a 1GHz RISC processor, all
linked "within a single shared memory architecture."

I can see putting p14p on a core so it can do some setup of what the rest on its
row will do, and format data in a simpler language than for parallel processing,
which likes a steady flow of the same kind of data. The p14p might also take inputs
from its row and reformat data that is all different, where each of the cores on its row is handling
a different calculation, reducing data size a lot, then shifting it to the p14p core.

Just a guess. Sounds exciting for image processing, data compression, crypto, sound, sonar,
radar, lidar, etc.

[Jingfeng Liu]

All,

Linksprite is giving away 10 pcduino for python developers. Please visit pcduino.com for more details, and contact sa...@linksprite.com if you want to get one. First come first get.

Thanks!

Sent from my iPhone