Page 1 of 1

The end of Moores Law?

Posted: Thu Aug 29, 2013 12:43 pm
by dante
http://news.cnet.com/8301-1001_3-576003 ... t-physics/

Might also mean the end of the 'native-is-getting-faster/cheaper-than-DSP' argument.

A flatter ground where, when all silicon power peaks equally at the curve of Moore's law (like a compressor) - then the only competitive ground left may be purpose built processing.

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 1:10 pm
by kensuguro
then we switch to quantum computers I guess. Or grow organic brains and use them as processing units.

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 1:34 pm
by dante
But your only talking about physics there. The article also discusses economics. 'The Matrix' cost a few cool million$$ to make but in reality - the discussion is around billions$$$

Dante's Law: Back to the Future = Back to Reality

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 3:07 pm
by dawman
Sound On Sound Reporter :

Do you accept the challenge from Native synths, U-he and NI.....??

S|C over the backdrop of Mr.T during the movie Rocky III ////

No, I refuse the challenge, as Native synths are no challenge, but I'd be happy to beat up on them some more.

Sound On Sound Reporter:

Do you have any prediction on the audio battle ahead.....?

S|C over the backdrop of Mr.T in Rocky III ///

Prediction....? .......Pain.
untitled.JPG
untitled.JPG (7.36 KiB) Viewed 1676 times

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 3:37 pm
by dante
Prediction ...? .... Pain..... and Purpose
Purpose
Purpose
purpose.JPG (23.47 KiB) Viewed 1671 times

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 7:13 pm
by ShogunSpy
And then there are the wannabes 'sonicxtc'
http://www.kvraudio.com/news/sonicxtc-r ... dows-23170

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 8:04 pm
by dawman
Free stuff is cool as long as you dont mind sounding free... :D

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 8:54 pm
by jksuperstar
Silicon is actually trending further away from purpose built devices. Mostly due to cost. Current designs in 28nm can easily cost over 1.5million, just for lithographic masks (it takes about 40 of them to do a multi-vt design). That doesn't include the 50 engineers, 18 months, CAD tools at $100K per seat, and the farm of servers running 24/7 to do the work. So gone are the custom E-Mu filter designs, the Lexichips, etc. Now it's mostly off the shelf parts...DSPs, Intel processors, microcontrollers with some ARM processor, or a field programmable gate array. Only high volume (or really high overhead) products have custom silicon.

Re: The end of Moores Law?

Posted: Thu Aug 29, 2013 9:12 pm
by kensuguro
I think there's also a problem of "there's only so much processing power the mass market really needs" issue. Of course, previously the OS did a good job of expanding and generally screwing things up as new processors came up.. but it seems like recently it's not so much the case since I'm guessing people got smarter (?) though Apple still locks OS to hardware. And there's web based stuff that don't require tons of hardware.. who needs to upgrade to an i7 when all you do is check FB.. why not do it on your iPad? I think the part of Moores Law that is not written out was that it relied on shitty OS clogging up the CPU whenever it doubled in speed.

So I think breaking that structure generally leads to less people being forced to upgrade, and sort of breaks the economy. I think it's a good thing for the consumer since they're not forced to buy stuff they don't need, or they're not forced to tolerate planned obsolescence.. But sucks for people who really need more processing power since it halts research and makes producing newer faster stuff economically impossible. (since as written, gets more expensive, and demand decreases)

I mean, all this suddenly became relevant to me when the sandy bridge i7s came out. They were just too frickin' fast for normal people. Like, imagine Joe Schmo just out of college on his first job, moved out into the city from somewhere in Philly. He's on FB all the time, posting pictures of random Brooklyn walls and lattes from his phone with every possible filter app he finds, probably doesn't have an iPad, and is probably growing a beard and thinking about going Vegan, as he buys his Vikram Yoga lesson tickets from Groupon. He probably plays a chord or two on his guitar so he records his stuff from time to time. And he's blown his first paycheck on some cheap DSLR so he does some Photoshopping on a pirated copy of CS6. And learning how to tweak his blurry photos of Bedford bands and bars truely brings him joy as he thinks, "man I'm livin' the life in NY!" The thing with sandy i7 is that it can probably handle double his crap no problem and still have enough power left over to render a 3D animation with radiosity at a respectable speed. (all the while, he's on FB telling the world he's livin' the life in NY, and applying more filters to his latte pictures on iPhone and posting them on Tumblr)

At work, I'm on a SB i7 quad (MBpro) developing in Unity just fine, and rearely getting close to maxing out the CPU. ON A FRICKIN' LAPTOP. Great for me, but frankly, I have never ever been in such a situation before. So it's either everyone starts writing unoptimized code at once to slow everything down.. or maybe CPUs are really... fast enough, at least for any normal task a normal person could dream of, using normal software. Games may still max out CPU, but it's more of a GPU thing now... and that upgrade path isn't for everyone.

It's all about forcing function. Imagine if Youtube, FB, Twitter, or God forbid Google search.. all of a sudden required immense processing power to run. Let's say Google moved some of their data center overhead right onto your machine, and your search result became a direct result of how much processing power you have locally. Now that'll get some upgrades. Want to watch Youtube in HD? Upgrade. Want to have more than 100 friends on FB? Upgrade. You shouldn't even be able to install Photoshop on i5. It should be packed with an algorithm so tough to solve that it would take an i7 Quad hours to unpack. Better yet, make the algorithm rely on a specific operation that only i7s can handle. (the old CPU dongle trick) i3 shouldn't even have a GUI. It should just run DOS. If you have a Celeron, it should only allow you to play Pong.

Strangely enough, Apple is doing a very good job on hardware locks and compatibility compartmentalization. Got an iPad 1? I don't even know what you're talking about. Got an iPhone 4? We'll make sure you don't stick around for long. It's cruel, but it keeps things moving along.

I do think though, that if CPUs hit a limit, then perhaps that's when people will start writing more optimized code to make things run faster with less resources. I mean, I LOVE Linux for that. Puppy runs on crappy machines better than Win7 runs on my i7 Quad. My i7 Quad will be able to animate more particles on the screen, but does Joe Schmoe care?

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 7:24 am
by dawman
Great post Ken.. :wink:
I also noticed that using the latest Ivy Bridge CPU.
I was concerned about the on die GPU taking system resources, as we saw in the i5 Clarksdales years back.
Not even an issue. Not only is there no need for a GFX card in my world, I don't even need the CPUs 85-90% overhead.

There is an upside for DSP/Native hybrid rigs here too.
Since I have such a high overhead I have been collaborating with a script writer and trying to take advantage of the excess IPs's I never seem to need.
Check out this brilliant chap who makes great stuff for Kontakt.
I fear his prices will skyrocket as NI might try to get him to become a Licsensed developer, so until this happens this guy is my favorite sample developer, who has great chops btw, and insane visions of feats in Kontakt.
Check out his little fix that even NI couldn't implement called Mind Control.
For 20 bucks Kontakt becomes a truly live performer.
But the stuff below is truly a step forward, and I got cycles to spare for this kind of Native development.

http://www.vi-control.net/forum/viewtop ... highlight=

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 1:14 pm
by dante
Great idea if it can do sectional balancing stuff, maybe a poor mans version of DVZ http://www.audioimpressions.com/ which hits the pocket at $999

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 4:24 pm
by dawman
This guy already brought us Mind Control for 20 bucks, and I already use the Poor Mans Pro Tools with Scope.
I kind of like the poor mans approach.

But the SE-1s, EML-101 and Solaris cannot have poor mans versions, which makes the poor mans software/DSP stuff even more relevant and wise.

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 6:49 pm
by kensuguro
heh, never heard of DVZ but it looks impressive! I use Synful partially because it internally does something very similar. Synful does a bit more, but the problem is that its base sample set has a very particular sound, and not too good quality.. May look into DVZ since strings arrangement is one of my strong points.

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 8:04 pm
by jksuperstar
kensuguro wrote:At work, I'm on a SB i7 quad (MBpro) developing in Unity just fine, and rearely getting close to maxing out the CPU. ON A FRICKIN' LAPTOP. Great for me, but frankly, I have never ever been in such a situation before. So it's either everyone starts writing unoptimized code at once to slow everything down.. or maybe CPUs are really... fast enough, at least for any normal task a normal person could dream of, using normal software. Games may still max out CPU, but it's more of a GPU thing now... and that upgrade path isn't for everyone.
The great news is I have a farm of 16 core 250gb RAM loaded machines, that I DO max out at work :). And the irony here, is the people designing the silicon (or whatever), will always want and need more powerful machines, just to design new machines with. So hopefully, there is never a need to slow down. Or I'd be out of a job. And my music chops are not involved with my day job for a reason.

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 8:12 pm
by kensuguro
true, that's the thing.. there is always a reason to move forward, but it needs to be funded. And if that requires a bit of "forceful" contribution from the mass market, then perhaps it's a small sacrifice for a worthy cause.

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 11:03 pm
by wayne
kensuguro wrote: I mean, I LOVE Linux for that. Puppy runs on crappy machines better than Win7 runs on my i7 Quad.
Just bought my son a Raspberry Pi for his 11th birthday - awesome little thing - just starting to look at some Linux fun. Luckily my older bro can code!

Re: The end of Moores Law?

Posted: Fri Aug 30, 2013 11:20 pm
by dante
dawman wrote:Solaris cannot have poor mans versions
Wanna bet ?

http://futuremusic.com/blog/2012/02/02/ ... on-review/

I'll be trying this with Solaris Core.... :lol: :lol: