request: stronger CPU version?


21 posts
by ranma1988 » Sun Dec 16, 2012 1:00 am
Hi im happy with my pi, but im in big need of more computing power * cpu intensive jobs *
are there any plans for pi with better CPU?
Posts: 10
Joined: Wed Nov 21, 2012 7:02 pm
by plugwash » Sun Dec 16, 2012 1:13 am
The problem is that a new CPU is unlikely to be a drop in replacement so it would involve considerable effort and money redesigning and recertifying the board. It would also run the risk of fragmenting the community.

I suspect it will happen eventually but not in the immediate future.
Forum Moderator
Forum Moderator
Posts: 2239
Joined: Wed Dec 28, 2011 11:45 pm
by ranma1988 » Sun Dec 16, 2012 1:15 am
thank You!
Posts: 10
Joined: Wed Nov 21, 2012 7:02 pm
by mahjongg » Sun Dec 16, 2012 3:28 am
Do you now that there is a safe way to overclock your PI so it can potentially run at speeds as high as 1GHz? Search for overclock in this forum for more info.
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am
by W. H. Heydt » Sun Dec 16, 2012 5:48 pm
*IF* the base clock for the Pi goes up in a future revision, I would expect it to be as a result of a process shrink.

I haven't seen anything on what process size the BCM2835 is made on, but I'd suspect that it's not out on the bleeding edge.
Posts: 1483
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)
by lmcc » Sun Dec 16, 2012 6:29 pm
The RPi is not expensive. Have you considered partitioning your project across more than one unit?

-L
Posts: 25
Joined: Wed Jul 25, 2012 11:25 pm
Location: Winnipeg, Canada
by jamesh » Sun Dec 16, 2012 7:57 pm
40nm IIRC.
Soon to be unemployed software engineer currently specialising in camera drivers and frameworks, but can put mind to most embedded tasks. Got a job in N.Cambridge or surroundings? I'm interested!
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 11660
Joined: Sat Jul 30, 2011 7:41 pm
by mahjongg » Mon Dec 17, 2012 12:11 am
Indeed not exactly bleeding edge, (2008 technology I think) 28nm is a bit more standard now, not that I would have expected 28nm for this chip. :roll:

Its an older chip as we all know, that is the reason the price is so affordable.

The current chip is perfectly fine for its intended market, but that other markets exist is a given.
But its a non sequitur that the PI should follow these markets.

In any case many of these other products wouldn't have existed in the first place if it wasn't for the Raspberry PI to have "lead the way", and showed it could be done, and that there was a market.
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am
by W. H. Heydt » Mon Dec 17, 2012 12:18 am
mahjongg wrote:Indeed not exactly bleeding edge, (2008 technology I think) 28nm is a bit more standard now, not that I would have expected 28nm for this chip. :roll:

Its an older chip as we all know, that is the reason the price is so affordable.


I I remember what I've read correctly, current chips are mostly in the 28-35nm range. Intel's current production is (again, IIRC) 22nm, with 14nm projected in a couple of years and roadmaps extending to 7nm.
Posts: 1483
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)
by mahjongg » Mon Dec 17, 2012 12:22 am
W. H. Heydt wrote:
mahjongg wrote:Indeed not exactly bleeding edge, (2008 technology I think) 28nm is a bit more standard now, not that I would have expected 28nm for this chip. :roll:

Its an older chip as we all know, that is the reason the price is so affordable.


I I remember what I've read correctly, current chips are mostly in the 28-35nm range. Intel's current production is (again, IIRC) 22nm, with 14nm projected in a couple of years and roadmaps extending to 7nm.

as always, Wikipedia is an excellent source of information:
http://en.wikipedia.org/wiki/Die_shrink
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am
by gritz » Mon Dec 17, 2012 1:11 am
mahjongg wrote:as always, Wikipedia is a source of information copy / pasted from other sources that may or may not be true


Fixed that for you! ;)
Posts: 449
Joined: Sat Jan 28, 2012 2:33 am
by mahjongg » Mon Dec 17, 2012 1:41 am
In the end all information is created by fallible humans, at least Wikipedia has a mechanism in place to correct fallacies, but believe what you want, perhaps some "authoritative" sources are more to your liking?
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am
by gritz » Mon Dec 17, 2012 2:15 am
mahjongg wrote:In the end all information is created by fallible humans, at least Wikipedia has a mechanism in place to correct fallacies, but believe what you want, perhaps some "authoritative" sources are more to your liking?


I'm sure that all Wikipedia contributors are just trying to help their fellow humans, but there's a lot of chaff out there. I was doing some research a couple of weeks ago in order to resolve a thorny DSP problem and while reading a particular W.P. entry (can't remember what exactly, but it was related to stats and pattern recognition) I realised that whole paragraphs had been lifted from a paper that I'd just read - and without citation. Not a big deal, but the person that posted the article didn't have the faintest idea about the subject, so instead of posting a real-world example he / she just reproduced a stack of partial differential equations, with no context whatsoever.

What a waste of everybody's time.

Sure, we could all dedicate the rest of our lives to editing this garbage, but I can't help feeling that our time would be better spent attacking the guilty parties with cattle prods instead, while shouting "You may think you're helping, but you're not..." Zap.

Capttalists the world over are trying to think of ways of monetarising the free web. How's this for an idea...

Time is not an infinite resource. We are each allotted a finite number of heartbeats and they are precious. If somebody wastes our time with erroneous information, spam or duff product then they have to pay. I haven't figured out the finer detail yet, but everyone will have some kind of online account.

I'll stop now, before I end up owing everyone money... ;)
Posts: 449
Joined: Sat Jan 28, 2012 2:33 am
by jamesh » Mon Dec 17, 2012 7:54 am
W. H. Heydt wrote:
mahjongg wrote:Indeed not exactly bleeding edge, (2008 technology I think) 28nm is a bit more standard now, not that I would have expected 28nm for this chip. :roll:

Its an older chip as we all know, that is the reason the price is so affordable.


I I remember what I've read correctly, current chips are mostly in the 28-35nm range. Intel's current production is (again, IIRC) 22nm, with 14nm projected in a couple of years and roadmaps extending to 7nm.


Current Leading edge chips. Still many many chips out there on larger processes.
Soon to be unemployed software engineer currently specialising in camera drivers and frameworks, but can put mind to most embedded tasks. Got a job in N.Cambridge or surroundings? I'm interested!
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 11660
Joined: Sat Jul 30, 2011 7:41 pm
by Jessie » Mon Dec 17, 2012 3:27 pm
Actually dispite the fact that many colleges don't trust or allow the use of Wikipedia research has shown that in most cases it is as accurate as a printed encyclopedia and in the cases where the topic evolves constantly (such as technology) Wikipedia is more accurate.

Here is one, but you must register to read it. http://www.nature.com/nature/journal/v4 ... 8900a.html

45nm may not be the smallest node out there but it is still fairly common, and citing that as the reason for the SOC being sub-par is bs. There are plenty of fast processors that were produced at 45nm. Producing at older mature nodes allows for lower defect rates. Broadcom could have chosen a newer ARM generation to base the SOC on which would have resulted in higher licensing cost and that would increase chip costs. Broadcom could could have added another core to this SOC which would have increased die size and increased defect rates and decreased the number of dies per wafer and all of that would also increased cost. Nobody is holding out just because they want to competition in the ARM space is fierce. It isn't like Star Trek... Broadcom dosnt just say "computer make me a 22nm chip." They have to engineer the die then pay a fab to produce and package it. Queues at smaller nodes are longer (more demand), defect rates are higher, and therefore you need a reason to go smaller.

I'm sure when a good faster alternative is avalible the foundation will make use of it.
User avatar
Forum Moderator
Forum Moderator
Posts: 1484
Joined: Fri Nov 04, 2011 7:40 pm
by jamesh » Mon Dec 17, 2012 4:18 pm
Worth remembering that the chip is 3 years old...back then, 40nm wasn't that old hat....
Soon to be unemployed software engineer currently specialising in camera drivers and frameworks, but can put mind to most embedded tasks. Got a job in N.Cambridge or surroundings? I'm interested!
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 11660
Joined: Sat Jul 30, 2011 7:41 pm
by W. H. Heydt » Mon Dec 17, 2012 5:36 pm
As regards the chip feature size...my remarks are not a criticism of the process choice, merely a note that--as time goes on, yields improve, and costs come down--a smaller process size is likely to become economical and may very well result in a faster chip with no new design work involved. (It may also result in lower power consumption as well...though I don't think that's a particularly large issue for the Pi.)

As regards Wikipedia... I consider it a reasonable place to start, so long as either the subject is non-controversial (the facts are known and no one disputes them) and/or one is aware of any controversies that exist and is willing to cross-check against other sources, primary sources where available.

Saying that "X chip uses process size Y on Z date" should be pretty non-controversial.
Posts: 1483
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)
by mahjongg » Mon Dec 17, 2012 11:50 pm
when reading wikipedia on -any- contentious fact, I have one tip, start with reading the talk page of the article, you will pretty soon get a very good idea exactly which parts of the article are contended! After that you will have a much better understanding of the good and weak points of the article.
In fact I would recommend this for almost all articles you have more than a passing interest in!
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am
by gritz » Tue Dec 18, 2012 12:10 am
mahjongg wrote:when reading wikipedia on -any- contentious fact, I have one tip, start with reading the talk page of the article, you will pretty soon get a very good idea exactly which parts of the article are contended! After that you will have a much better understanding of the good and weak points of the article.
In fact I would recommend this for almost all articles you have more than a passing interest in!


My criticism of Wikipedia has nothing to do with possibly contentious / contoversial articles. My objection stems from the fact that technical entries are often unhelpful because they are not written with the intention of imparting knowledge - and at least some are merely copy / pasted from other (possibly overly opaque) sources without any additional context.

I note with some glee that there's a short news article in this week's New Scientist (15 Dec pp27: "Wikipedia is a tough muddle") regarding a study that questions Wikipedia's legibility by comparing it to Encyclopaedia Britannica.

There's a suitably scholarly report on the matter at http://www.dl.kuis.kyoto-u.ac.jp/~adam/ ("Is Wikipedia Too Difficult? Comparative Analysis of Readability of Wikipedia, Simple Wikipedia and Britannica")

Being free is no excuse for being clunky and opaque.
Posts: 449
Joined: Sat Jan 28, 2012 2:33 am
by liz » Tue Dec 18, 2012 12:13 am
Hearty agree: I wish someone would give the Raspberry Pi article a good old shake-down. I used to be a very enthusiastic Wikipedia editor, but stopped when I started working on Pi (largely because now all I do is Pi and sleeping, with occasional breaks for washing and eating). The current article pains me, but because of NPOV there is bugger all I can do to correct the problem.
--
Head of Comms, Raspberry Pi Foundation
User avatar
Raspberry Pi Foundation Employee & Forum Moderator
Raspberry Pi Foundation Employee & Forum Moderator
Posts: 4083
Joined: Thu Jul 28, 2011 7:22 pm
by mahjongg » Tue Dec 18, 2012 12:18 am
well i have more than 10.000 edits to my name on en.wikipedia.org, and have been editing the Raspeberry PI article quite heavily, so im not quite neutral on the subject (LOL), but If there is anything specific in the article that bothers you, maybe I can help. I know it has a header that says its a bit out of date, but only some parts of it might be. Basically the article is quite OK, me thinks. But then I know I sometimes write a bit convoluted, not that I'm to blame this time.... :D

p.s. the anonymous poster who made the above rubbish remark has been reverted as a vandal, and has had a reprimand on his talk page by one of the more major editors. Things like that happen now and then, but are usually gone before you can blink.
User avatar
Forum Moderator
Forum Moderator
Posts: 5297
Joined: Sun Mar 11, 2012 12:19 am