What is Digg?76 Comments
- geminitojanus, on 10/12/2007, -4/+30"For one, thread optimization is not a secret thing; I am sure that any professional code writer can work with such issues as optimization. If they don't do it, you can't really blame the hardware manufacturer."
First of all, I spoke in generals so that we could avoid such a conversation; I didn't say "Apple needs help with thread generation and scheduling, so they went to IBM looking for help with it", I said "Apple needed help with the processors, IBM didn't offer it". Specifically, with the IBM processors, there were a whole different class of problems that don't translate directly to Intel processors; they're two different architecures, and code generation on either platform is tough business, and the only way to know for sure you're getting it right is to talk to the manufacturer.
"Secondly, IBM has produced, and still produces, a Power PC optimized compiler"
And IBM has poured extensive work into Linux and the GCC compiler to make it a world class PPC compiler. This, however, doesn't make it an issue in this conversation. Apple looked to IBM for a mobile G5, and they looked to them to make modifications in the current G5 to make them more hospitable to higher clockrates, and they looked to them for code which would aide in code generation from GCC. Some of these modifications happened. Others didn't. Apple wasn't happy that IBM wouldn't pay them enough attention, as the chief purchaser of these chips, to make the modifications, or work with them on the code for those modifications to go into software. And so, doing as any business would do, they moved on to a business that was bright eyes, bushy tailed and ready and rearing to help in every way they possibly could.
"I could never understand why Apple software developers - or third party software developers - would not use such an IBM compiler but instead refer to gcc."
Because GCC is FREE SOFTWARE, it's redistributable, everyone knows how to work with it, everyone knows how to optimize for it and around it, because it's not exceedingly quirky in the way it generates code, and unlike IBM's own internally developed compiler, it works on x86 as well as PPC, by changing a few lines around on the back-end rather than by having to shift all of your front end tools to a new compiler suite (See Adobe/CodeWarrior debacle). Furthermore, your whole argument about Macs not being powerful should be shoved up your own ass as seeing they've made supercomputers out of them that were cheaper than all other presented alternatives, and that GCC was used in writing the applications for (at least one) of those Supercomputers (one of the kids I just hired wrote part of it). Would it have been more powerful if they bought 1/3rd of the number of PCs and threw Itaniums in it? Yes, it would have been a hell of a lot more powerful. It would have also costed nearly three times as much, would have needed way more in the way of energy, and would have netted a slightly higher amount of airconditioning (which means more cost).
"However, since 2 single core processors (most Macs are that way) don't provide an advantage over 1 single core unless the load exceeds a certain amount (with smaller jobs, you'd lose time just by distributing the job!), the need for special multithreading was not as big for most dual processor Macs. So it was enough to have some jobs distributed among the CPUs in a rough way for any dual processor Mac."
You obviously think you know what you're talking about. You also obviously don't. Most Macs being sold today are dual core, single chip Macs. In the yesteryears of Apple, it would have been the other way. However, even in these older Macs, they had one of the best schedulers in the business (I think the only operating system that can truly claim to have a better scheduler is Linux 2.6 and its incredibly experimental and successful O(1) scheduler). With smaller tasks, only ONE processor is used because only ONE processor is needed. Things like Word Processing never pegged the second processor because it never needed to pass instructions to a second processor. With load, the kernel scheduled the tasks between the two processors, and something called "scheduling affinity" helped keep applications from ping-pogging between processors (which was actually a huge problem for other kernels leaping from single core to dual core or single processor to multiple processor). You should also know that there was a reason Apple didn't sell dual core, general desktop machines until just very recently; even as their OS is incredibly good at dealing with it, they're nowhere as good as needed to run the next generation applications and keep affinity, and have the kernel scale well with the number of tasks and processes. This is why they called in Intel when they moved platforms; when you've worked on someone else's processor for 30 years and you suddenly decide to move to another company's processors, you can't know how the scheduler and your compiler's code generation will be affected.
"With hyperthreading and multiple-core processors, the minimum workload required in order to yield a significant advantage using hyperthreading (you can try that out on Pentiums if you want) is much lower - so it makes sense to commit some more decision power on what to multithread and what not."
Good thing Hyperthreading was never supported by Apple in any generation or future generation of their computers; Intel has killed Hyperthreading because they are ready to go with symmetrical core computing (multi-cores). Multi-core computing still needs a huge amount of data to be processed in order to be advantageous. However, since we now have computers capable of managing these huge amounts of data, it's now advantageous to build computers with multiple cores. HD video, HD audio, super powerful video editing and audio editing, Applications that blend well and lend to a central theme, multitask well, and help the user manage their lives in a meaningful way, these things are all being ushered in by multiple core computing, and Apple is taking one of the leaping first steps into this area. This kind of work cannot be done alone, and IBM and Freescale said they were not willing to help Apple do it. Intel practically begged them to.
"It's interesting to hear that rumors indicate a performance obsession at Apple."
Apple has been a premiere institution for computer performance for years. They've fought for years to squeeze out every drop of performance from their computers as absolutely possible, caches everywhere, hand optimizing applications before they shipped to make sure every ounce of performance they could milk out of an application was there. They can't do it anymore on IBM and Freescale processors; the company's won't help them do it, and it's not something you can do in a vacuum. If you don't believe Apple does all of this work, go talk to an Apple engineer and listen to the war stories they get into with their managers when they say something like "there's simply no more performance we can get out of this application". Hell, I work for another company altogether, and I've been called by Apple employees trying to squeak performance out of their application because I've worked with the same processors they work with every day and know some tricks about them that they might not. And just because you don't like Apple, doesn't mean you can't appreciate that about them. - geminitojanus, on 10/12/2007, -3/+26Because, you know, if the company that helps you build the processors also helps you write the code, it's going to be hell of a lot better code.
That being said, where was IBM when Apple tried to work exclusively with them on the G5? That's right, they were too busy working on gaming processors like the CELL, the Xbox 360's "Xenon", or the Revolution's "Broadway". Where was Freescale when Apple said they wanted to work with them on dual-core G4s? IBM and Freescale dropped the ball, Intel picked up the ball and passed it back to Apple with a lucrative offer.
And no, it shouldn't run any more ***** on the older PPC processors than it will run on an Intel processor (it's all still going through GCC, and GCC is still a world-class PPC compiler thanks to the effort put into it by IBM). - pzzq, on 10/12/2007, -7/+24ah, macosrumors, then we know it can't be true.
- Yarnage, on 10/12/2007, -8/+21Wow.... 100% speculation presented as if they watched it in action and saw the code.
While this sounds really cool and it would be nice if it's true, but you gotta remember. It's macosrumors.com. They have a horrible track record.
Wait for it to be on ThinkSecret.com. Then you know it has a 95%+ chance of happening.
Marked lame - JohnBooty, on 10/12/2007, -8/+18Why does Apple get all the press? Is it because Digg readers think Apple is perfect? No, it's because most of the interesting things in desktop computing are coming from Apple these days.
What's the last interesting thing to come out of Microsoft as far as desktop operating systems are concerned? Vista is a year away and I can't really think of anything I'm looking forward to from it.
By the way, I'm a professional Windows developer. I use both OSs and spend more of my time on Windows, if truth be told. - Angostura, on 10/12/2007, -11/+19As an Apple fanboy, I say no digg. Breathless hype and no technical detail should not a front page story make.
- SwampYankee, on 10/12/2007, -5/+12More Apple Polishing. Call rewrite! What it says is that Apple is PLANNING to do amazing things with Intel. As for actualling doing it, well, it hasn't happened. I'm planning to do amazing things with Angilina Jolie & Shakeria, but I would not quite write about it yet.
- inactive, on 10/12/2007, -2/+8There should be a law against websites with more than 10 ads on a single page.
- geminitojanus, on 10/12/2007, -1/+7A real programmer asks for help. All programmers need it. Especially manufacturers of operating systems.
Working in a vacuum on a computer is not unlike working in a vacuum on the design of an airplane. No one person is an expert in avionics, fluid dynamics, jet propulsion, guidance systems and platforms, and possibly weapons systems, and yet an aircraft needs (most of) all of these things to be considered an aircraft in modern engineering. No one person is an expert on any given language, code generation for that language, operating system scheduling of tasks on the platform your application was coded, the actual hardware executing the instruction, and making sure that this operation happens in the most effecient way possible, yet all of these steps should be taken to make a good program (even if you don't believe in aggressive optimization, or are a "code-and-can" programmer [one who codes an application and then "cans" the software, either by deleting the source, or archiving the source on their private CVS server, never to be viewed by human eyes again], as many programmers for other companies I know are.
I'm not an expert on threading and scheduling. I work with an expert on task scheduling though. I'm no expert on code generation, but the guy who works in the cube next to me knows more about hand rolling assembly than anyone else I will ever know, and can compile some C code in his head for multiple different architecures (ARM, x86; but to his credit he helped design a few ARMs in his day). What I am is an expert on making streaming data applications scream. And when people need that help, I get a phone call. Ask me how many times I've recieved a call from Microsoft in 11 years (read: once, and it was a wrong number). Ask me how many calls I've recieved from Sun in 11 years (none). Ask me how many times I've recieved calls from Apple, even in the more recent two years of my experience? I have a conversation from someone working for them every few days anymore, asking how or why an ARM does something nutso, or asking if I'll review a piece of code for them, or asking why their application crashes if a certain MPEG-4 frame is broken in a certain way. I can't answer half of the questions I get calls from them about, but I can damned well try, and I bet their applications are better for it.
Next time you're at a conference, pick up some business cards and give everyone a courtesy call, tell them your specialty, and tell them you're friendly for business. It'll make you a better engineer. - LaughingMan11, on 10/12/2007, -1/+6The site seems to have only a fleeting understanding of what Intel's microprocessor roadmap is going to be for the next couple of years...
Conroe is not a family of microprocessors. It is one member in the "Core" microarchitecture. Conroe itself is the upcoming desktop chip. It will not have 16 or 32 cores like they claim... it will be decidedly dual core only. In early 2007, we can probably expect quad-core chips in both the Destkop and Server spaces from Intel based on the same core mircoarchitecture, but do not expect 8 or 16 cores immediately thereafter...
Further scaling from 4 cores to higher numbers for Intel will require significant changes to process technology and bus technology. Do not expect 8 or higher until 2009 by the earliest, when 45 nm process technology should be in full rollout. - devwal, on 10/12/2007, -1/+5Heh, don't get angry. We stole your Intel processor and our desktop is better, and now we're going to hit on your girlfriend. Have a nice day.
- geminitojanus, on 10/12/2007, -1/+4But there is one part of the equation you're not looking at; Multiple core, multiple processor. 16 cores is achievable with 4, 4-core processors, 32 with 8, 4-core processors (or visa versa, 4, 8-core processors). Any way around it, it's very smart to be mindful that multiple cores are coming, and that your kernel will scale well when you throw N processors into the mix. Hell, think of what it'll do for Xserves and Apple Supercomputing ("Big Mac Attack", if you will).
- mark1372, on 10/12/2007, -3/+6I stopped reading MOSR long ago because of the made-up stories, for one, but mainly because of the unbelievable amount of ads and the way they write a paragraph of an article and then force you to click through to another page to read the next paragraph, and then another page for the next...ad nauseam.
It's a gross pit of ad-whoring, and what's even more pathetic is that they're the Windows "warning"-type ads that link to spyware and malware payloads (oh, but I get smileys for my email!) that only work on Windows machines. The fact that they can't get valid Mac sponsors or credible advertisers is a testament to their site's value. - geminitojanus, on 10/12/2007, -0/+3Wow, there's a man on digg I can talk to about Chip Technology. I'd buy you a beer if you were anywhere around here, but since I can't I'll add you as a friend. But, let's talk through it, because I'm competitive like that and love my chip technology.
"The first round of Intel Core Microarchitecture procs will still use traditional P4-style front side busses, and will inherent many of the problems with that bus in lack of scalability."
This is the roughest point about the new Core architecture, how do you feed the beast the amount of data that it's designed to deal with. Intel does have an exit strategy, as it were, but it's too late in development for this round and the next round of Core chips, and it looks like it's going to be right on schedule in the 2008 late generation of chips which will be looking at 4 and 8 cores, which by Intel's clock, is right on schedule (even though AMD will be wiping the floor with them with 4 and 8-way configurations, they will have gone through two sockets to get there, and as every systems manager knows, going through sockets like there is no tomorrow is not a lot of fun, Intel on the other hand will have gone through one socket, if we don't see them adding a memory controller, which has been hinted at by people I know).
"The article came close to the truth on one point: multiprocessor performance will not scale well past 16 cores... but this has more to do with Intel's choice in multiprocessor architecture than with Apple and their software... any optimization that is to be done in software is secondary to Intel's responsibility of improving their multiprocessor architecture."
And this is why it's so important for Apple to get their software right now, while it's something they can put a leash on and walk around ("cute little toy with bright eyes and a bushy tail" vs "big, ugly truth-beast that's hard to code around and not break applications"). One of my very first professors in engineering school said "This is Von Neumann, and this is his architecure. We talk about him as history because very soon his ways will be. The ones in my class that can wrap their heads around [multi-threaded] processing will be the ones with jobs in 10 years. Everyone else here will not." To this day, it feels like that has been the turning point in my career, if not life.
And who knows, maybe Apple will talk Intel into bumping the schedule on their next generation interconnect technology. Okay I know you got a giggle with that one..
"The bottom line is that yes, there are scalability problems, but I don't believe at all that they have as much to do with the operating system than it does with the architecture. This article is BS."
The bottom line is, I totally agree with you, and it's the reason I didn't digg the article myself. But, I like the idea that Intel and Apple are getting along, whereas Apple and IBM had almost a love-necessity relationship. - inactive, on 10/12/2007, -1/+4'Firefox prevented this site from opening 4 popup windows.'
Macosrumors has gotten really lame lately. - nurriz, on 10/12/2007, -2/+5By all that is binary, what an awful site full of blinking crap. The credibility factor absolutely plummeted alongside the loading of "Warning" gifs and screensaver.com popups. If you have a site that reports rumors maybe you shouldn't dress it up to look like a click-machine.
- drakino, on 10/12/2007, -2/+4I'm still waiting for the color changing multi processor (not multicore) PowerBook G4 that "rumor" site predicted a while back.
That site is like the tabloid quality site of the rumor sites. And considering most of the rumor sites are so off base on any day, MOSR just looks even more horrible with their predictions. - canyonblue, on 10/12/2007, -1/+3only half agree with you. clearly mac users hated intel now say they love intel... but the reality is fans of any product often switch views on a dime to continue to "love" their product. the same is true for windows fans, people who said they didn't need certain features of mac os x are now jumping for joy that vista may have those features. it isn't a crime, just a funny quirk of human nature.
- rdas7, on 10/12/2007, -10/+12but there's no good software for windows :( no iPhoto, Garageband, iMovie, etc...
- LaughingMan11, on 10/12/2007, -2/+44 socket and 8 socket interfaces are going to have huge hurdles with existing Intel technology. Intel does not have nearly a robust interconnect technology as AMD does with Hypertransport. The first round of Intel Core Microarchitecture procs will still use traditional P4-style front side busses, and will inherent many of the problems with that bus in lack of scalability.
The article came close to the truth on one point: multiprocessor performance will not scale well past 16 cores... but this has more to do with Intel's choice in multiprocessor architecture than with Apple and their software... any optimization that is to be done in software is secondary to Intel's responsibility of improving their multiprocessor architecture.
Past 4 cores, the memory subsystem will become a huge bottleneck as Intel will have to resort to their shared bus and dumb memory controller because they do not have an interconnect technology like Hypertransport and an on-die memory controller.
Lets take your example of a 4x4 configuration... What you need to understand that on Intel's current roadmap, the first quad core processors will be MCMs, multi chip modules, where they tie together 2 dual core pieces of silicon on the same package with the front side bus. While Intel's new dual-core Core microarchitecture will have a shared L2 cache, where 2 cores can talk to each other without hitting the FSB, a 4 core config will still require communication over the FSB...
A 4x4 config will actually imply 4x2 connections to a shared FSB... That's a heck of a lot.
The bottom line is that yes, there are scalability problems, but I don't believe at all that they have as much to do with the operating system than it does with the architecture. This article is BS. - HackWithRamzi, on 10/12/2007, -0/+1"mean, it can't be Spotlight, because Windows has had the option of indexed searching long before the Apple, and even Google Desktop Search predates Spotlight."
Have you ever used Spotlight? Try using it, then use Google Desktop Search. Spotlight is way better. - zonk3r, on 10/12/2007, -0/+1i wonder if apple will continue to have osx running in the backroom on powerpc chips after they eventually kill off (probably a few years from now) the powerpc line like they did with the osx for intel. seems like it might be smart to do in case things steer in a direction where, say, the cell processor seems the future years from now...
- Angostura, on 10/12/2007, -1/+2Aside from the MS bashing, you have a point - one of the thing that was notable about the OS X releases up until (but not including) 10.4 was the way that each was faster than the previous one. Apple spent a lot of time optimising and tweaking for speed.
It seems reasonable to assume that once they got 10.4 Intel out of the door they would get serious about squeezing every last ounce of power out of the Intel architecture. - Swift2, on 10/12/2007, -1/+2Well, to those who say Apple is getting all the news, this is a user site. If you've got some hot news from the Windows world, feel free to post the news.
- moisie, on 10/12/2007, -3/+4The thing is that I believe that this could be true. I'm not knowledgeable enough to understand the technology of it but basically if any company is gonna take advantage of increases in hardware technology then it has to be apple. They've gone through the bad patch with their OS, bit the bullet and made a transition to a new system and are now reaping the benefits. Microsoft are a bloated mess who take forever to do anything. Irrespective of which OS you prefer (I'm ingoring Linux I know, but because it's not one company, not because it's not good) you have to agree that Microsoft are having problems, problems which Apple have largely moved past. Because of their size (for one) Microsoft are gonna find those problems far harder to overcome and as such I don't see them really taking advantage of new technologies anywhere as near as fast as Apple can.
- inactive, on 10/12/2007, -2/+3http://www.macosrumors.com.nyud.net:8080/20060402B.php
Here's the Coral .. site is Dugg - PhysicsTheory, on 10/12/2007, -2/+3Do you honestly think that by the time you can actually buy a 32 core chip that Intel's FSB bus will work in the same way it does now and the chips will only have 2MB L2? ...
- geminitojanus, on 10/12/2007, -1/+2I don't believe so. Apple has already been hard at work on GCC in a lot of different areas of GCC (go read a changelog ;), so I believe that will maintain the way they look at things. However, I do believe that working with Apple may actually bring more Intel developers into working on GCC, and I'm also sure that some of the algoritms developed will find their way into Intel's internal compilers, as long as licensing jeopardy isn't attached. It's a good thing for both Apple and Intel, and every open source developer in the world who relies on GCC as their compiler of choice.
- canyonblue, on 10/12/2007, -1/+2they seem to imply that it will only reach alpha/beta in jan 07 which seems odd to me. my understanding is jobs commited to late 06 / early 07 (in other words... early 07 hahaha) and is expected to demo it at the developers conference this june.
all i can say is i fully expect apple to demonstrate that vista while taking forever is being left in the dust before it even comes to market. - inactive, on 10/12/2007, -2/+3Or you'll see all the Disney movies get released by iTunes and all the rest of the movie studios get *****-hammered by stockholders because they aren't utilizing an avenue for profit... (In my view the only one that could stand against Apple+Disney is News Corp (Which owns 20th Cent Fox/Fox TV)
btw Disney is any movie/show by:
Disney, Disney Animation, Pixar, Miramax, Hollywood Pictures, Touchtone, Disney Channel, ESPN, ABC, Buena Vista TV, etc
http://en.wikipedia.org/wiki/The_Walt_Disney_Company
http://en.wikipedia.org/wiki/Walt_Disney_Pictures
http://en.wikipedia.org/wiki/Touchstone_Pictures
http://en.wikipedia.org/wiki/Hollywood_Pictures
http://en.wikipedia.org/wiki/Miramax_Films - DocDEB, on 10/12/2007, -0/+1Golly, lmlloyd, here's an idea... make a comment that is worth responding to.
- swisswuff, on 10/12/2007, -0/+0"Because GCC is FREE SOFTWARE, it's redistributable, everyone knows how to work with it, everyone knows how to optimize for it and around it, because it's not exceedingly quirky in the way it generates code, and unlike IBM's own internally developed compiler, it works on x86 as well as PPC, by changing a few lines around on the back-end rather than by having to shift all of your front end tools to a new compiler suite (See Adobe/CodeWarrior debacle). Furthermore, your whole argument about Macs not being powerful should be shoved up your own ass as seeing they've made supercomputers out of them that were cheaper than all other presented alternatives, and that GCC was used in writing the applications for (at least one) of those Supercomputers (one of the kids I just hired wrote part of it). Would it have been more powerful if they bought 1/3rd of the number of PCs and threw Itaniums in it? Yes, it would have been a hell of a lot more powerful. It would have also costed nearly three times as much, would have needed way more in the way of energy, and would have netted a slightly higher amount of airconditioning (which means more cost)."
First of all, we did compare speed results of PPC versus Intel using GCC, and Mac software that was not specifically tweaked ran more than significantly slower. It caused us to leave Apple solutions for time being. You really have to love Macs to death in order to overcome such significant performance problems.
Secondly, "super computers" of the kind that you mention - clustering boxes by "throwing together" some "units" - are typically built under some type of flag-ship contract. The company offering to build that "super computer" will expect a huge return by that particular "super computer" installation being "scientific advertising" of some sorts, and while they will likely offer the "super computer" setup as cheaply as they can, they'll have to pull the missing revenue in through adding price to the single user machine. So unless you are buying a flagship setup, you'll not see much of the assumed cheap price. Apple wouldn't be the only company to do it like that - other manufacturers work in the same way. So, saying that Apple has installed a cluster somewhere doesn't really prove that much other than their interest in trying scientific advertising.
So, installation of clusters always adds to a company's myth without saying something about the real performance. Now, if Apple based clusters were indeed a factor in terms of low cost and high performance, they'd abound. But they don't. So, very likely performance and cost end up not adding up after all. If you buy 1536 units of the XServe model for a unit price of 7048 US dollar as advertised on the Apple website, you will have to shell out more than 10 Million US Dollars just for the computers. Seeing as if such installations indeed warrant an investigation not only on performance itself, but - with so many units - reliability, unit surveillance, power consumption, software / hardware optimization and floor space, it does make sense to evaluate other options under pure price/performance considerations. Looks like out of the top 500 fastest setups, not many builders shared your view. Maybe BlueGene does offer a better, cheaper, more powerful integration (one unit apparently only consumes 17W; XServe is a whole dimension higher).
I like cheeky advertising for it's sometimes entertaining value, but I'm not sure I buy it each and every time. - mark1372, on 10/12/2007, -2/+2But you can get a Life Experience Degree for only $199! And it's BOUA accredited, everyone is approved, and it will arrive in five days – with free delivery!
The best part of that banner ad is that it's entitled WARNING! with a flashing red X. How appropriate. - OneAndOnlySnob, on 10/12/2007, -0/+0"the same is true for windows fans, people who said they didn't need certain features of mac os x are now jumping for joy that vista may have those features."
I'm not. I'm switching to Linux and/or continuing to use XP till the bitter end. Vista is repulsive to me and most Windows users I know. The only views I know on this topic are:
That's it, I'm switching to Linux (nobody but me forces me to use bloated software.)
That's it, I'm switching to OS X (I don't fall into this camp so I'm not sure what the rational is. (Not that I'm against these people taking OS X over Vista. (Though I hope Linux gains popularity too if Microsoft slides, because of what it represents.)))
I'm sticking to XP. (XP does everything I want an OS to do so I'm just going to leave it alone.)
That said, Vista's biggest competition is clearly XP. It's going to take a LOT to convince most Windows users that XP isn't good enough for them anymore. Lucky for Microsoft, Vista will come on new computers regardless. - LaughingMan11, on 10/12/2007, -1/+1If you take a look at Intel's roadmap going forward, http://en.wikipedia.org/wiki/Intel_Next_Generation_Microarchitecture, you'll know that the aggregate amount of L2 cache will be at sizes of 3MB, 4MB, and 6MB in the immediate future. The amount of L2 will likely scale with more cores, and also will increase with the die shrink to 45nm.
The shared nature of Intel's L2 cache shouldn't be looked down upon. The "sharing" happens only among the 2 cores on the same piece of silicon, and is actually a vastly more intelligent architecture than discrete L2 caches per core, because it allows for on die core to core communication without the FSB. - rino, on 10/12/2007, -1/+1BTW: WWDC was moved to August so no June love for developers, watchers and the like.
http://developer.apple.com/wwdc/ - balle13, on 10/12/2007, -0/+0It could actually be very true.
I believe it means that Apple has begun using OpenMP, and plan to add support for it to the gcc compiler.
For more information about OpenMP:
http://www.intel.com/cd/software/products/asmo-na/eng/compilers/cwin/219880.htm
http://www.openmp.org/drupal/ - lmlloyd, on 10/12/2007, -0/+0"And when people need that help, I get a phone call. Ask me how many times I've received a call from Microsoft in 11 years (read: once, and it was a wrong number). Ask me how many calls I've received from Sun in 11 years (none). Ask me how many times I've received calls from Apple, even in the more recent two years of my experience? I have a conversation from someone working for them every few days..."
Ok, I don't doubt what you say about your expertise, but am I reading this right? Are you in fact suggesting that Apple is a better company specifically because they call you? I'm sorry, but that is a bit on the egotistical side. Did it ever occur to you that there just might possibly be one or two other people out there in the entire world who know as much about your area of expertise as you do, and those couple people might work for Microsoft or Sun?
I suppose it is possible that you know more about this subject than anyone else in the world, and thus are the best person in all of the human race to talk to about these matters. If that is true, then it certainly shows a large amount of arrogance on the part of Sun and Microsoft, but it seems somewhat unlikely. The more likely scenario seems to me to be that you are quite knowledgeable, and have a more favorable view of Apple, because they help justify your paycheck with their constant calls, unlike Sun and Microsoft, who have their own just as competent in-house people to do what you do. - kolop1, on 10/12/2007, -2/+2 I remember a day when Mac people said Intel sucks. Now they all love Intel for some reason. Mac fans are nothing but a bunch of whiny hypocrites.
- OneAndOnlySnob, on 10/12/2007, -1/+1People are stupid for modding this comment down. It's 100% true. It is simply not feasible or practical to do this. Making a thoroughly multithreaded application is difficult and requires careful planning by developers. Not even Apple can magically make that problem go away.
Not to mention the fact that you're not going to get any performance benefit from multithreading most applications. Text Editor isn't gunna scream once it's running multiple threads. It will still spend 99.9% of its time waiting for input. Applications that would benefit from this are already going to be multithreaded or they'd be abysmal failures as applications in the first place.
You Apple fanatics are insane! - MattH, on 10/12/2007, -1/+1The First Studio will be NBCUniversal and then Fox acording to a industry source who is a friend of mine .
- lmlloyd, on 10/12/2007, -1/+0Wow, Intel working with a company to optimize code for their processors, how amazing! Why didn't Microsoft think of that? Wow, those guys at Apple sure are innovative and groundbreaking. I can't think of a single other time anyone has ever worked with Intel to optimize code for one of their chips.
Get serious! I know that macheads just found out Intel made chips when the new Macs were announced, but that doesn't mean that somehow now Intel doing what Intel has always done is some revolutionary new thing brought to you by Apple! - champs, on 10/12/2007, -2/+1so... what they're saying is that Apple's come full circle to take the BeOS approach?
- paulryan21, on 10/12/2007, -1/+0You can bet Jobs is going to move heaven and earth to ensure Leopard is out before Vista - wheter it is true or not - Apple continue to roll with better and better hardware and even better operating systems as the years roll by.
- zaduma, on 10/12/2007, -1/+0http://www.shortText.com/gnsp
Heres the shorttext (loads faster than coral) - SectorNation, on 10/12/2007, -3/+1The author of this article clearly has no fundemental understanding of how processes and threads are distributed to multiple CPUs on OS X and Darwin. They are implying that processors fill up like dixie cups, one at a time. To say that the first core is busy the second is moderate the rest are so so is complete and utter hogwash.
The bigger issue here (but hopefully not, Apple) with lots of cores in a future Mac Pro and XServe is will the bus architecture have enough bandwidth to provide sufficient enough I/O from memory, disk, whatever to keep the cores from starving (basically sitting around doing nothing).
MacOSRumors indeed.
- yvovandoorn, on 10/12/2007, -3/+1The problem with these types of articles is that it is all hype unfortunately. We are all going to have to wait until the first betas leak on the web and people start installing it on their Macs.
- Ramble, on 10/12/2007, -4/+2Do they honestly think that a 32 core processor will provide any sort of perfomance benefit over Intel's FSB core link (Not to mention the measly shared 2MB of L2 cache)?
- inactive, on 10/12/2007, -6/+4You called?
- OneAndOnlySnob, on 10/12/2007, -3/+1As a programmer and a computer scientist, I think this sounds both ridiculous and far-fetched. Plus it's macosrumors.
-
Show 51 - 76 of 76 discussions
Browsing Digg on your phone just got easier with our enhancements to the 
© Digg Inc. 2009
— Content created and posted by Digg users is