From mboxrd@z Thu Jan 1 00:00:00 1970 MIME-Version: 1.0 In-Reply-To: References: <5778ffc079ca4ad7227b37b1158b29f1@proxima.alt.za> Date: Wed, 5 Feb 2014 15:57:33 +0500 Message-ID: From: Muhammad Junaid Muzammil To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net> Content-Type: multipart/alternative; boundary=089e0149cdb2b83b9704f1a69f90 Subject: Re: [9fans] Inferno and the Parallella Topicbox-Message-UUID: b6a26650-ead8-11e9-9d60-3106f5b1d025 --089e0149cdb2b83b9704f1a69f90 Content-Type: text/plain; charset=ISO-8859-1 Well!! I have no idea about the Styx implementation. But certainly this parallela board seems an interesting target. Currently they are only producing with 16 core epiphancy board and not 64 core ones. Definitely hard cores will always be faster than the multi cores. The cost to these faster FPGA design is the longer design time as compared to the multicore counterpart. I think, the generic architecture of multi cores has always been the key to this rapid developemnt. Junaid On Wednesday, February 5, 2014, Shane Morris wrote: > Seems there is some interest in this chip, and the board. As you said > Henry, state machines would run very quickly in parallel - I had some wad > arguing that an FPGA is the only thing you need, but nothing beats a hard > core for hard tasks. Grid it up in parallel, times sixteen, and thats a > fair bit of processing power. > > I have already sent a few people emails concerning this, independently of > the list, to garner their opinions. Out of four people, only one has > replied, I suppose, Australia is on the other side of the world to most of > you guys, time zones and all. But I'm going to make an open call - if a > GSoC mentor in parallelism could be found, willing to advise the project, > could this be a GSoC project? > > I suppose I should tender the idea on the wiki, but I'd rather not. Never > played with wiki's, nor had the interest to try. Too busy designing robots. > Tried a TAFENSW Moodle once, that was bad enough. > > > On Wed, Feb 5, 2014 at 5:32 PM, Henry Millican > > wrote: > >> Parallella seems very cool. I'll probably pick one up when I have free >> time. >> >> I've worked with the Zynq chip on board, which is also great. For $99 >> it's one hell of a dev board, considering you get an FPGA with hard ARM >> cores, as well as the Ephiphany chip. >> >> The Ephiphany processor fills in the gap between CPU and FPGA tasks in my >> opinion. Things that would require complex state machines on an FPGA could >> be done in parallel on the RISC cores very easily (and quickly). I can >> imagine doing some image processing or something (that doesn't lend itself >> well to FPGAs) of the like with this. >> >> I'll be following you guys and may have time to contribute, but I am just >> a hardware guy after all. >> -- >> >> Henry >> >> >> On Tue, Feb 4, 2014 at 10:20 PM, Shane Morris >> > wrote: >> >>> Oh, its ok. I like the GSoC idea. I just don't think I'm GSoC material, >>> I'm hardware type, even if I will be a uni student this year going forward >>> - "If it draws blood, its hardware" as the old maxim goes. >>> >>> The Parallella board is US$99, a far more modest investment in hardware >>> than a GizmoBoard as I had previously suggested, and packs more power for >>> the price, in terms of coding value. Whether it could be accepted as a >>> coding project of the type for GSoC, a mentor for it found, and other >>> logistical concerns are a issue for the GSoC organisers, but I suppose, >>> could it happen? >>> >>> An abstract topic for the time being. >>> >>> >>> On Wed, Feb 5, 2014 at 5:06 PM, >>> > wrote: >>> >>>> > Thoughts? Comments? Critique? Flames? >>>> >>>> I guess this is the real value of efforts like GSOC, if only they >>>> could be extended to a much greater public either with an infinite >>>> budget or by pushing a far more socially-aware ethos. >>>> >>>> I'll refrain from pontificating further. >>>> >>>> ++L >>>> >>>> >>>> >>>> >>> >> > --089e0149cdb2b83b9704f1a69f90 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Well!! I have no idea about the Styx implementation. But certainly=A0this p= arallela board seems an interesting target. Currently they are only produci= ng with 16 core epiphancy board and not 64 core ones. Definitely hard cores= will always be faster than the multi cores. The cost to these faster FPGA = design is the longer design time as compared to the multicore counterpart. = I think, the=A0generic architecture of multi cores has always been the key = to this=A0rapid developemnt.

Junaid

On Wednesday, February 5, 2014, Shane Morris &= lt;edgecomberts@gmail.com>= wrote:
Seems there is some interest in this chip, and the board. = As you said Henry, state machines would run very quickly in parallel - I ha= d some wad arguing that an FPGA is the only thing you need, but nothing bea= ts a hard core for hard tasks. Grid it up in parallel, times sixteen, and t= hats a fair bit of processing power.

I have already sent a few people emails concerning this, ind= ependently of the list, to garner their opinions. Out of four people, only = one has replied, I suppose, Australia is on the other side of the world to = most of you guys, time zones and all. But I'm going to make an open cal= l - if a GSoC mentor in parallelism could be found, willing to advise the p= roject, could this be a GSoC project?

I suppose I should tender the idea on the wiki, but I&#= 39;d rather not. Never played with wiki's, nor had the interest to try.= Too busy designing robots. Tried a TAFENSW Moodle once, that was bad enoug= h.


On Wed,= Feb 5, 2014 at 5:32 PM, Henry Millican <henry@tehserv.net> wrote:
Pa= rallella seems very cool. I'll probably pick one up when I have free ti= me.

I've worked with the Zynq chip on board, which is also great.= For $99 it's one hell of a dev board, considering you get an FPGA with= hard ARM cores, as well as the Ephiphany chip.

The Ephiphany processor fills in the gap between CPU and FPGA tas= ks in my opinion. Things that would require complex state machines on an FP= GA could be done in parallel on the RISC cores very easily (and quickly). I= can imagine doing some image processing or something (that doesn't len= d itself well to FPGAs) of the like with this.

I'll be following you guys and may have time to contribute, b= ut I am just a hardware guy after all.
--

Henry


On Tue, Feb 4, 2014 at 10:20 PM, Shane Morris <<= a href=3D"javascript:_e(%7B%7D,'cvml','edgecomberts@gmail.com&#= 39;);" target=3D"_blank">edgecomberts@gmail.com> wrote:
Oh, its ok. I like the GSoC= idea. I just don't think I'm GSoC material, I'm hardware type,= even if I will be a uni student this year going forward - "If it draw= s blood, its hardware" as the old maxim goes.

The Parallella board is US$99, a far more modest investment = in hardware than a GizmoBoard as I had previously suggested, and packs more= power for the price, in terms of coding value. Whether it could be accepte= d as a coding project of the type for GSoC, a mentor for it found, and othe= r logistical concerns are a issue for the GSoC organisers, but I suppose, c= ould it happen?

An abstract topic for the time being.
<= div>


On Wed, F= eb 5, 2014 at 5:06 PM, <lucio= @proxima.alt.za> wrote:
> Thoughts? Comments? Critique? Flames?
I guess this is the real value of efforts like GSOC, if only they
could be extended to a much greater public either with an infinite
budget or by pushing a far more socially-aware ethos.

I'll refrain from pontificating further.

++L






--089e0149cdb2b83b9704f1a69f90--