Post a Comment
"There are no multi-core systems on the PPC"
There are many multi-core PPC systems. There are even quad-core PowerMac G5s.
The PPC970MP series had 2 cores and Apple put 2 processors in each machine, so yes, effectively quad-core. Those were the last of the series before the Mac Pro
I don't see why you need to make that distinction. If it supports multiple cores, it will support any multi processing. It doesn't matter if it's running on PowerPC, AMD or Intel.
This is very different from vector processing that is dependent on the instruction set (i.e. Altivec vs SSE vs 3DNow).
I don't think that there is any Intel specific instruction set that makes parallel processing any easier. Preventing deadlocks and breaking down an algorithm into "chunks" that can execute in parallel are the main problems in writing multiprocessor aware code. I can't really see an instruction set helping out with that.
Even so, the experience (remember 64 bit Carbon?) tells me that PowerPC support could be dropped for purely political reasons.
Edited 2008-06-18 17:41 UTC
Not enough to stop me from making my next purchase from apple, but still an annoyance
Maybe they didn't go for ZFS due to the overhead that is required for it; ZFS isn't a 'slim file system', it loves to let it all hang out, so a machine with atleast 1gig memory might have persuaded many off within Apple making it the boot file system.
Then again, this is an early preview, what appears there, and what might appear in the final version can be worlds apart. If there are enough demands placed on them by customers, I'm sure they'll make ZFS atleast an option.
Edit: Btw, I run Solaris on two machines, and use ZFS on all my drives; its a great file system, but with ever good, there is a down side.
Edited 2008-06-18 16:57 UTC
Not enough to stop me from making my next purchase from apple, but still an annoyance
What makes you think that Case Insensitivity is going to go away...
http://www.opensolaris.org/os/community/arc/caselog/2007/244/;jsess...
Personally, I hope that ZFS makes it to client machines and not just server...
IMHO: That site is a joke.
Basically 10.6 will break a lot of old apps and screw old PPC customers, but it will be good for the enterprise.
Okay .. once you start catering to enterprises breakling compability is not really appreciated anymore .. so maybe it is a good choice to have a fairly "future-proof" base.
That said: I don't think OSX has a lead on competitors _overall_ .. so waiting to release new features for 2 more years might be not such a good idea.
Who knows what Windows7 or (K)Ubuntu 10.10 will be like?
Does this link seem better then Thom if that website is so bad?
http://www.anandtech.com/weblog/showpost.aspx?i=461
Edited 2008-06-18 19:21 UTC
Apple does not cater to UNIX "purists". Dropping PowerPC support makes sense eventually, seeing as PowerPC Macs have not been manufactured in the last 3 years or so.
This is a good thing, as I have a good chuckle everytime I load up Xcode and look at the project settings. To make everyone happy, I have to compile for PPC, PPC64, x86 and EM64T (and the options are there in Xcode). Now that's a real fat binary. It makes sense at some point to fully go EM64T as all Macs shipped within the last 2 years support 64 bit. You get the performance benefit of EM64T (think 8 more GPRs) and you get slimmer binaries.
Edited 2008-06-18 17:28 UTC
That right there is the problem. If Apple drop PPC support on the OS level, developers will too stop making universal "fat" binaries. Right now, because the vast majority of the software is universal, Apple can switch back to PPC in a snap, if they so fancy. This is an advantage that no other OEM has - Windows is officially x86 (and ia64?) only, just about all the software for Windows is x86 only, and Linux is not really an option form most PC buyers. If you consider that ten years ago, when Apple were still scant on cash, they could afford to develop and build Mac OS X for both PPC and x86, secretly but still, I don't think they'll save (or need to save) all that much by dropping PPC.
Also, neither Grand Central, nor OpenCL necessitate dropping PPC. Auto-parallelization (or whatever Grand Central is supposed to be) is not dependent on instruction set (certainly there is pretty much no difference in that regards between PPC and Intel 64). OpenCL is also supposed to transparently decide between GPGPU and CPU vector units, so SSE vs. AltiVec should not matter either.
In short, Apple has quite a lot to lose and nearly nothing to gain from dropping PPC.
There is no quicker way to destroy OSN as a news source than by linking to Roughly Drafted. There is a reason why it is curiously neglected, and its not because of anti-Apple conspiracies. But you will have to find that out for yourself. What will happen is this. As soon as you select yourself as one of the only places on the net prepared to regularly link to it, you will be bombarded with a flood of stories from a horde of apparently different people.
They will all, these stories, say one of the same things. Apple computers are cheaper than PCs (after you add in various ridiculous costs to the PCs, Apple has larger market share than reported (after you double Apple sales by separately counting hardware and software), Apple security issues are malicious black propaganda by an anti-Apple news establishment. There are various convoluted reasons why every decision Apple has ever made was right. Apple has a brilliant master plan which is exemplified by its latest move, which has been universally misunderstood, but which will result in its total domination of [pick one: the Web, mobile phones, laptops, desktops, music downloads, TVs....] And so on.
This view of RD is generally accepted in the Mac world. Here for example is Technovia, quoting in support TUAW:
Posts like this that use underhanded techniques and shoddy math to prove a biased point aren't helping the Mac community. In fact, they're making it look even worse because, once found out, they are (rightfully) transformed into key evidence for clueless Apple fanaticism, which can easily harm the reputation of almost anyone with something genuinely educated and relevant to say about Apple or their products, whether it's a good or *gasp* unpleasant statement.
Unsurprisingly, the post that TUAW is referring to is from Roughly Drafted, which can only be described as the lunatic fringe of Mac fandom. While it's perhaps understandable that some Mac users will want to "strike back" against what they see as media bias against the platform, as TUAW puts it:
"These wholly inaccurate and misleading articles need to stop if computer users are going to learn the genuine advantages (and even disadvantages) of Apple and their products. Lying to grab attention for Apple may briefly taste sweet, but helping consumers make an educated choice (no matter what the outcome) based on pure, true merit takes the cake."
When your own Party starts to describe your agitation on behalf of the cause as 'lunatic', you're in deep trouble.
But, you will have to find all this out for yourself. Meanwhile, if you are going to start linking to RD, it would be nice if you can introduce some blocking feature into the site so we can filter it out. Its the Slashdot/Roland Picquepaille problem. Please help those of us who simply do not want to be reminded of its existence.
Edited 2008-06-19 07:37 UTC



