Power 7 CPU - 8 cores

[minator]

There's a big, big difference between massive multicore in special-purpose applications, and massive multicore for general-purpose computing, though.

I know, that why I mentioned single threaded next :-)

Graphics in particular is a task that's pretty much tailor-made for massive parallelism.

Yup.

Word processors, file managers, and other unglamorous productivity software? Not so much.

True, but it's now got to the point that a lot of software doesn't require a high end processor anyway.

On the other hand a lot of the stuff that does require high end processors can be parallelised.  I just bought a new high end laptop with a quad core CPU because I run things that can max it out - video editing, recording music and photo editing.

These can all run across multiple cores.  That said, editing 22 Mpixel images appears will happily use all 8 hardware threads but it seems to be more limited by the hard disk (which is actually a s**t hot fast SDD).

[/QUOTE]web browsers[/QUOTE]

Actually these do quite a lot at the some time so they can take advantage of multiple processors.

[minator]

Indeed; everybody wants to ride the hot new ARM wave that's surely going to last forever, and nobody seems to remember when the new wave was PPC...

Because they're cheap and made in vast quantities - much, much higher than x86.

But the real reason:


Go to eBay and search for "Android Laptop 10", buy it now only.
10" laptop for £80 (including postage) that's faster than any of the Sam boards.

That's why.

Welcome to the world of ultra low cost computing.

[mongo]

I don't know of any processor that supports breaking for an interrupt mid-instruction - that would be overly complex to implement, introduce a highly undesirable degree of non-determinacy, and not get you anything more than slightly lower interrupt latency for the trouble.

Doing it for an interrupt is unnecessary, but a 68010 or above will stop mid-instruction, save the processor state on the stack, and allow you to resume the failed instruction on a bus error.

[ElPolloDiabl]

@matthey
Yes but that's my point. You can't compare AmigaOS from 1980's to Windows 7 from 201X hence my hypothetical example of IF AmigaOS had continued to be updated then it'd be in the same sluggish state.

There are things to be said for the simplicity of AmigaOS but if you like that sort of thing then you should look at HaikuOS and see how to achieve it's minimal feature set still requires some serious CPU performance before you start running any programs on it.
Andy

Try one of the various Linux distros, this runs modern software, runs very fast, has a faster boot time than Windows. Also it has never crashed in my experience, although I haven't flogged the thing as much as Windows.

I have tried HaikuOS it may look like BeOS, but it isn't.

[commodorejohn]

Try one of the various Linux distros, this runs modern software, runs very fast, has a faster boot time than Windows. Also it has never crashed in my experience, although I haven't flogged the thing as much as Windows.

I had more frequent and severe crashes running Linux than I've ever had running XP. Didn't boot any faster, either.

I have tried HaikuOS it may look like BeOS, but it isn't.

Mind elaborating on this? I missed out on the BeOS salad days, so I really don't know what Haiku has yet to achieve on that front (though it certainly could use some newer features, i.e. a proper GUI wireless client.)

[psxphill]

One of my lecturers at uni was pretty high up in CPU design for IBM in the 80s and you can bet your ass they tested every possible scenario for server and desktop OS efficiency as far as parallel processing goes.

One of your lecturers who has no experience in processor design in the last 30 years tells you something and you believe him? Priceless.
 
While some algorithms are inherently parallel and can't be split across cores easily, there are a lot that can be. However it does take intelligence to design your code to work that way & unfortunately most programmers lack intelligence. They only know what they've been taught by people who couldn't hack it, who then become lecturers.
 
In the real world there are people who have written code that is efficient and can take advantage of a large number of cores. Maybe one day you'll meet one.
 

There's a big, big difference between massive multicore in special-purpose applications, and massive multicore for general-purpose computing, though. Graphics in particular is a task that's pretty much tailor-made for massive parallelism. Word processors, file managers, web browsers, and other unglamorous productivity software? Not so much.

Out of your example, web browsing is the only one that really needs a performance boost. With javascript and compressed video, there is definitely a need for more performance & there is no reason that can't be by using more cores.
 
There is a rise in using SQL databases for storage in a lot of applications that you wouldn't have thought, even Android uses SQL. A decent SQL engine will make use of multiple cores.
 
The future is multiple cores for general purpose computing, programmers will need to adapt.

[commodorejohn]

Out of your example, web browsing is the only one that really needs a performance boost. With javascript and compressed video, there is definitely a need for more performance & there is no reason that can't be by using more cores.

Yeah, web browsers are definitely moreso than the other examples. On the other hand, half the reason that's even true is because of the omnipresence of terribly slow (and almost always completely unnecessary) Javascript in modern websites, so even though there is a benefit from a properly multithreaded browser, it's mostly in making bad code less crippling.
 

The future is multiple cores for general purpose computing, programmers will need to adapt.

Again, I'm not arguing that multicore isn't a good thing; it's lovely for what it is. It's just not an Ultimate Solution to the fact that eventually computers are going to reach a point of maximum practically-attainable power, and all the people who have been blithely counting on Moore's Law to cover for their many, many sins of bloat and shoddy coding and poor optimization are going to run straight into that wall at full steam and then start crying about having to learn to be efficient again.

[KimmoK]

I had more frequent and severe crashes running Linux than I've ever had running XP. Didn't boot any faster, either.

On some of my x86s some windows did not even manage to install itself, etc, etc. Linux went in far more smoothly. Booted faster and was stabler.
I think I've had thousand crashes with windows and only a few with linux (+ perhaps hundred crash of the Linux desktop GUI, that really is not linux fault).
And modern linux remembers what you had running when you powered off, it can restore your work's state pretty nicely.

(latest win crash happed just 2 hours ago with my work laptop, when I lift up the lid of the laptop, bluescreen gave me it's warm welcome. and yesterday the laptop just locked up when IE was used, hard power off worked. I know people blame idiotic IT support, but without M$ there would be no need for such a IT support.)

Modern linux main distros seem to require more than 4Ghz of computing power. I've forced to move to lighter desktops, like LXDE and Enlightenment.


Nice to see Linux getting in places also at work. Even Beaglebone HW is in the lab handling various I/O things and routing.

[commodorejohn]

On some of my x86s some windows did not even manage to install itself, etc, etc. Linux went in far more smoothly. Booted faster and was stabler.
I think I've had thousand crashes with windows and only a few with linux (+ perhaps hundred crash of the Linux desktop GUI, that really is not linux fault).

Well that's lovely for you. Doesn't change the fact that I've had XP bluescreen exactly once, and that on account of a crap video driver, while I lost hours of work to sudden core-dumps during my switchover attempts. (And that's not even counting the years during which I was using DSLinux as a portable text editor.)

And modern linux remembers what you had running when you powered off, it can restore your work's state pretty nicely.

Beg pardon, are you telling me that it actually maintains enough stability in a crash to instruct applications to save your work, and then dumps and reboots? Because merely remembering what I had running is peanuts - I can remember what I had running without help.

[ElPolloDiabl]

When I suggested Linux I was being friendly. Now that it has gone hostile:
Your OS choice could depend a lot on experience. Hardware incompatibility is also a problem.
My Windows install usually XP becomes more and more troublesome after a number of resource hungry applications are installed, including the anti-virus (solves one problem, creates another.)
When I start running a lot of 3rd party software on Linux I will get back to you on how stable it is.
I think we should get back to discussing Amiga OS varieties?

[BlackMonk]

Any CPU engineer will tell you instantly we have hit the wall and all this multi-core desktop CPU stuff is just a scam really. The maximum number of cores without losing efficiency of code execution is effectively 3. Not 4 not 8 but 3. You can not utilise much more than this without actually starting to waste cycles of CPU time delaying/setting up use of threads to run on other cores.

And this is why consumer devices are getting the ability to handle more threads?  XBox360 has 3 cores and can handle 6 threads.  Wii U supposedly is based on the Power7 and uses 3 cores and can handle 12 threads.  Why does Tegra and other ARM SoCs add cores and integrate more functionality with each subsequent version, if consumer applications of multithread are worthless?  Yes, I'm extending "desktop" to "consumer" since in my mind it's the same thing.  

Consumers can get applications for their iPhones that allow you to take multiple pictures and stitch them together in one giant panorama--that's a CPU intensive and parallizable workload.  Any video transcoding will also benefit.  Efficiency of code execution?  You're not talking about prefetch units, decode units, schedulers, caches, registers, nothing?  Just, if you have more than X cores, it hurts code efficiency?  I'm not sure any CPU engineer would make a blanket statement about that when not knowing the instruction mix or details of the CPU implementation.

There are many many situations that the AMD 4.2Ghz PC will outgun an Intel i7 3700k fact. One of my lecturers at uni was pretty high up in CPU design for IBM in the 80s and you can bet your ass they tested every possible scenario for server and desktop OS efficiency as far as parallel processing goes. For desktop computers 3 or 4 is about it. So Moore's law is f**ked well and trully unless we start seeing 5 and 6ghz CPUs QUICKLY!

You misunderstand Moore's law.  Here:  http://arstechnica.com/gadgets/2008/09/moore/

We've still got a ways to go, too:

http://arstechnica.com/science/2012/02/we-can-do-no-moore-a-transistor-from-single-atom/
http://arstechnica.com/science/2012/08/researchers-open-door-to-electronics-made-from-single-layers-of-atoms/

I'm leaning towards agreeing with psxphill in that you need to broaden your pool of knowledge--that one guy might talk a good talk but I think it's a minority opinion that he has.

The new Xbox and the new PlayStation are all confirmed to have x86 64bit CPUs already, and this means that those CPUs will be dropped into a much better design of motherboard architecture than any PC for sale in 2013/2014 with the same CPU to compete on price/performance.

Anything you can cite from the vendors on this?  Or is it just rumors as to what CPUs they will use?

I am interested in CPU-related junk and some of your info seems just plain wrong.  Nothin' personal.

[Fats]

I'm leaning towards agreeing with psxphill in that you need to broaden your pool of knowledge--that one guy might talk a good talk but I think it's a minority opinion that he has.

FYI, when I started to work at my current place there were people (including university professors) claiming we would never need submicron technology as it would be too expensive and would never be needed. At the time MACs still contained Motorola processors and hardly could multitask.
Currently we are at 20nm node or more than 10 nodes further and still scaling ...

greets,
Staf.

[BlackMonk]

I suppose I need context for that.  If it's a current university professor expousing such things, yeah, I don't see how he could still believe that in the present day.

If it were university from 10+ years ago, ok, maybe I buy that he wasn't in a minority opinion at the time.

More information needed but it's a minor point of contention, I think.  Anyone alive TODAY should be able to form their own opinions based on current data, not relying on one professor.  Even if it's a present-day professor, it doesn't hurt to question their opinion.  Teachers have their own biases which are sometimes at odds with the real world.

[Iggy]

I had WinXP BSOD on me a couple of months ago while my system was booting and it corrupted the drive rendering it unbootable.
 
Don't think I've had such a drastic crash since the days of DOS based Windows.
 
Since then I've sold most of my Windows hardware and I'm relying on a Win7 equipped netbook.
 
I've got a copy of Snow Leopard and Parallels7. From now on I'm restricting my Windows use to virtualization.
 
I just don't trust Microsoft's software. It blows up too readily.

[vox]

A faster way to run software that simply doesn't exist that would even take advantage of it, at least on the Amiga side.

Give me a 9Ghz PPC CPU on my SAM and I'll still be asking for software that even tasks the 600 mhz chip that's on it now.

It's like overclocking a calculator when things are essentially a "one trick pony" ordeal.

While Linux could use that kind of power we need to fully use existing one and develop some real software first. No need for new hardware yet. But everyone that can buy SAM or X1000 should do so.

Meanwhile, some updated AROS PPC, more support of these 3 boards from Linux PPC Distros and AmigaOS/MorphOS development would be nice.

[Iggy]

But everyone that can buy SAM or X1000 should do so.

Seriously considering a 460, Vox.
Can't justify an X1000, but I'm hoping that whatever Varisys designs as its sucessor is more affordable.