X-Received: by 10.224.36.66 with SMTP id s2mr1098915qad.6.1369811794692; Wed, 29 May 2013 00:16:34 -0700 (PDT) X-Received: by 10.49.106.169 with SMTP id gv9mr42875qeb.7.1369811794559; Wed, 29 May 2013 00:16:34 -0700 (PDT) Path: csiph.com!newsfeed.hal-mli.net!feeder3.hal-mli.net!newsfeed.hal-mli.net!feeder2.hal-mli.net!npeer01.iad.highwinds-media.com!news.highwinds-media.com!feed-me.highwinds-media.com!ch1no2701997qab.0!news-out.google.com!y6ni57070qax.0!nntp.google.com!ch1no2701995qab.0!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail Newsgroups: comp.lang.forth Date: Wed, 29 May 2013 00:16:34 -0700 (PDT) In-Reply-To: Complaints-To: groups-abuse@google.com Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=62.157.168.251; posting-account=o1zNBwoAAABP2KiH5g6nPQp_sYWidVdt NNTP-Posting-Host: 62.157.168.251 References: <088a35b6-5c19-4197-817b-25d08c7276e1@googlegroups.com> <0a12eb5c-6a59-49d0-a1f7-30053cf0df0f@googlegroups.com> <2013May28.172741@mips.complang.tuwien.ac.at> User-Agent: G2/1.0 MIME-Version: 1.0 Message-ID: <4e9ade2f-b3ce-4e82-9cbd-6747eea8164f@googlegroups.com> Subject: Re: OT (slightly) What is the "best" processor for a new project? From: Howerd Injection-Date: Wed, 29 May 2013 07:16:34 +0000 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-Received-Bytes: 3901 Xref: csiph.com comp.lang.forth:23014 On Wednesday, May 29, 2013 1:05:27 AM UTC+2, rickman wrote: > On 5/28/2013 11:27 AM, Anton Ertl wrote: >=20 > > ric.... writes: >=20 > >> I don't assume the GA144 is low power in all cases. There are any >=20 > >> number of apps that the GA144 will have nodes just spinning, marking o= ff >=20 > >> time, at 5 mW each. >=20 > > >=20 > > Why would a node spin? And what do you mean with "marking off time"? >=20 > > Why would an application do that. >=20 >=20 >=20 > If you need to time something, in a machine with no clocks, how do you=20 >=20 > do that? In the GA144 you would count CPU instructions... that is how=20 >=20 > Chuck prefers to do it when he can get away with it. Read his Video work= . >=20 >=20 >=20 >=20 >=20 > > From what I have read, a core does not consume much power when it >=20 > > waits for input from another core, or when it waits for another core >=20 > > to consume data so it can push some more to it. >=20 >=20 >=20 > Yep, but try writing real world and especially real time apps. A lot of= =20 >=20 > timing has to be done. >=20 >=20 >=20 > --=20 >=20 >=20 >=20 > Rick Hi Rick, I believe there are two issues with power consumption of the GA144 : 1. If you want to measure a time period you have to have a node counting, w= hich uses more power than just a hardware counter 2. If a node needs to act as two "wires" between two pairs of nodes (say up= /down and left/right) at the same time it must run a program loop to do thi= s. Acting as a single wire it can use Chuck's wonderfully clever "running o= pcodes from an input port" technique, which uses only quiescent power. I also believe that Chuck, Greg and the folks at GA know how to solve these= issues, so I await with interest the GA144-2 with the ability to "multi-wi= re", plus a 32KHz crystal oscillator in hardware. Maybe even a PLL on the c= rystal node and a _synchronous_ clock bus running around to a few other nod= es. And how about a crypto node with write-only NV memory, and a few 32 bit= nodes for SHA-256 bitcoin mining...=20 That way I could program as many serial ports as I liked, plus a USB node o= r two, and run a 16 bit polyForth that only consumed power when you press a= key. The cost of a GA144, SRAM, serial FRAM (for program storage) and 1.8V data = buffers would not be too high to put me off. Just my wishlist, but all perfectly possible :-) Disclaimer : the above are all guesses - I have no insider information from= GA ! Best regards, Howerd