Slashdot Mirror

The link to Arstechnica (http://arstechnica.net/) is not really to the correct site, Arstechnica...

Instead, you'll get a parked domain rife with popups.

Your kidding right? Theres no way in hell osx can compete against xp/2003 in multitasking. OSx is pretty much single threaded, OSX(Pre tiger) used funnel locks, locking down pretty much everything. Aren't you glad your osx is based on a 30 year old os? XP uses fine grained locks and 'push locks' which have no contention problems. Luckily for you, Tiger got rid of of the funnel locks and replaced them with kernel section locks, it still can't compete against windows xp or Server 2003.

ArsTechnica

I use the hot rod spec whenever I am looking for a new mobo. The rest I just shop around for on a part by part basis, paying close attention to price breaks on video cards.

Which USB stick do you have? The Lexar JumpDrive Lightning reads at 22 MB/s. ArsTechnica review.

Meanwhile, Microsoft has announced that IBM will make 3.2 GHz triple-core G5 derivatives available to Microsoft for Xbox 360.

The PowerPC processor included the capability to emulate 68K instructions, allowing almost all 68K applications to run.

Support is eventually dropped for all older hardware in the current OS (for example, for PowerPC G3-based systems).

"SPEC is subject to all kinds of problems."

All benchmarks are subject to problems. If you are testing for the CPU, somehow you must control for variables of compiler, OS, system architecture, and the amount of time and expertise of the tester.

Uh, how exactly do you get "GCC, is said to generate code that less well optimised than Intel's" from "GCC, is said to generate code that less well optimised for x86"?

Sigh. read: "Dell's own figures were calculated using different compilers and host operating system: Windows XP Pro, Intel's own C++ and Fortran compilers, and the MicroQuill SmartHeap Library 6.01. Secondly, the compiler used by VeriTest, GCC, is said to generate code that less well optimised for x86." QED less well than intels compiler in the previous sentence

GCC for PowerPC is not as mature: "The gcc scheduler is not really designed ideally for a processor like the 970 and the Power4...that was one of the things that we're continuing to work on to try to get the best performance out of the processor."

GCC on intel is far more mature with a long history, read a little of the history: "...When Intel released the Pentium some of their team produced a version of gcc with enhancements which gave 30% speed improvements on some benchmarks..."

Look at these redhat GCC 3.3/4.0 benchmarks. Notice how the 2-way PPC970 is twice as fast as the 4-way P4 on many tests and at close to par on the others. Now this is not the end all, am I'm sure you could come up with a different test that shows the P4 beating the G5, but certianly the G5 is not a "peice of crap".

You arguement about standardizing compilers is equivalent...

Standardizing of compilers is scientific method. Ideally you'd do a bank of tests, and unroll the variables: Standard compilers, standard OS, standard CPUs. Or you could tune each system to the max and then compare, that was LinPack and you didn't like that one either.

Hmm, does this appear to be vector processing done by a compiler?

Exactly my point! Intels compiler does auto-vectorization. GCC doesn't. If you test C code, P4+intel against GCC+G5, you are crippling the G5 by leaving out the altivec unit, which is a more capacble vector unit than SSE2

"hand coding...becomes completely out of reach for humans"

Hand coding is still done frequently on high performance algorithms:

• "I've also recently started hand coding the low level math kernels... P4, this gives a ~30% boost to performance on this particular MILC code."
• "However, once that level of optimization becomes necessary it's generally just easiest to hand-code the instructions, rather than jumping through a bunch of hoops to try to trick the compiler into doing what you want."
• "In some cases, complete vectorization is not possible and you may want to include hand coded SIMD instructions for the best possible performance"
• "Altivec requires hand-coding to exploit"

Here are some insights offered by someone who has written support for the Tulip ethernet chipset in IOKit. It's fun to read if you're into this kind of thing.

Apple won't ship an Intel PowerMac until probably late 2006/early 2007. Since they update their PowerMac line about twice a year, that means at least two major updates to the existing G5 line. So if you're worried about better PowerMacs in a year, you should be worried about a dual 3+ GHz G5 PowerPC machine, not a Pentium.

The best estimated roadmap I've seen of Apple's plans is still this column.

What I really find funny about the 'not-finished' discussion is the blissful ignorance surrounding computer advances. Lots of jokes about Longhorn flying around, but nobody seemed to catch the demo of the OS's UI running on the video card. Etc.

That's not really an advance. It has been demoed long, long ago on other systems. And in fact, it is shipping in Mac OS X 10.4, although admittedly disabled by default.

You can download X versions for linux too which run the UI on the graphics card. So, I repeat, to demo this is nothing worthy of the label "advanced".

The only thing I consider advanced surrounding the whole longhorn spiel is Monad/MSH. I haven't yet seen demoes of piping objects around a shell instead of strings.

Higher resolution doesn't always mean bigger files. The textures are already quite big. Increasing the poly count on the models won't be that much of a hit.

And at least for the xbox 360, the keyword is Procedural synthesis.

http://arstechnica.com/articles/paedia/cpu/xbox360 -1.ars

Day 3: Apple annouces their spreadsheet and database app and complete iWork, their office replacement, and charge 1/2 of what Mac MS Office costs.
Day 4: Admins rejoice as they can now replace most of thier paperpusher's PCs with Macs and worry a hell of a lot less about security.

I have to question if you understand the word pipeline. A decoder and its associated execution unit are the same pipeline, not two separate ones. Athlons have three pipes.

What that means, is that an Athlon performs ~9 operations per cycle, or 9 * 2.8 Ghz = ~25.2 billion instructions per second, and the intel would do 6 * 3.8Ghz = ~22.8. Those are very, very rough estimates.

Again, completely wrong. AMD's architecture can at best issue 3 instructions per cycle (and this depends on there being 3 instructions present in a 16 byte fetch window, plus various other factors such as branching into the middle of the window and predecode bits being correct), and they can only retire 3 instructions per cycle. As a result the best possible IPC is 3. In reality, a good application will exceed an IPC of 1.

So basically, the Ghz of the processor doesn't mean a whole lot. There's a lot more to take into consideration.

This at least is correct. You might want to read this article for a start, although it really only covers the processor's front end:

Arstechnica processor overview

Specifics:

Why not just implement enabling technologies instead of trying to make second-rate implementations of good ideas and hard-wire them into the operating system? I'm thinking of the horrible Show Fonts gadget, the Show Colors gadget, and the Address book. Changing fonts in OS 9 took a click, drag, release. Now it requires a click, a drag, a release, a select, and then close the dialog. Why is this better again? The whole idea of building an Address book into the operating system, which they never improve and which nobody else can really make good use of. The Color gadget. WTF?

More: The random mixes of interface styles between applications. The non-spatial monstrosity they mockingly call "Finder", the lack of thought to number of mouse clicks and distance between events, and other GUI errors...

I now believe that they no longer have any real standards, just a set of guidelines that they feel free to ignore. Such details were understandably glossed over after OS X came out at first, but holy crap, Mail hasn't changed for years now, as far as I can tell. They update it, but it never improves. It's depressing.

Ah yes, Mail.app. The whole painful Mail.app interface with its weird sort-of-heirarchical menus and color labels that don't show up on messages unless they ARE NOT selected. Its underpowered rules filter. The weird implementation of Spotlight technology. The hit-and-miss interfacing with IMAP servers, misplacing messages. The ability to say "use this mailbox as the Trash," but afterwards, you can't set that back to the Trash! Endless nits like this, too many to mention, but each reflecting a lack of thought, and implying to me that Apple employees must use Eudora for their mail needs. Now it's version 2.0. Can somebody explain why this got a whole version bump? Was it just Spotlight? It's weird. Seemingly just a rewarmed Next gadget, Mail is.

Another thing is the organization of the Applications directory. Why is Grab a utility but Preview isn't? Some things go into folders, some don't. It's exactly as if each little project just picks a random spot and sticks their application there. Quick quiz: where is Stuffit Expander located? You can't move them because then they won't get updated properly. It's just crap. Crap I say!

And another thing: after software updates, sometimes the installed application is an updater such as iPod update. The installer asks me to restart, then the installed application cancels the logout to install something I don't even own. I can't choose NOT to download the update. It's pretty messed up.

Oh, I guess I'll lose karma for this post! :-) But it's the truth. OS X is a hard-core sweet technology with tons of power and is the best thing out there, but it could be a lot better if they would shine it up a bit. The spit and polish is gone, replaced mainly with spit.

Apple would still have control over the hardware. I'm sure Dell could produce MacTel boxes that would satisfy Cupertino's equipment requirements.

Expanding OS X's install base would be a GOOD thing for existing Mac users: more drivers, more software, more everything (including, perhaps, viruses).

Expanding OS X's install base would improve Apple's research and development ROI. Good for stockholders, and eventually for consumers.

Expanding OS X's install base would be a very good thing for Windows users, who would have more opportunities to purchase increasingly price-competitive MacTels.

Michael Dell's out-loud thinking is entirely consistent with his signal last week that Dell is ready to launch a premium line of computer hardware.

I think Apple ought to do it. They've traditionally been a computer hardware vendor, with software used to sell the pretty boxes. But these times, they are a-changin'.

I stand corrected. The error had two sources: First, a friend of mine who told me so, and second, a misreading of the bold from the review at Ars Technica:

If Steam is having a bad day and you can't download a necessary update, you may not be able to play the game that's already installed in your computer.

My bold, my bad. Thanks for the correction.

I've already written too much in this forum today, but I did want to add that Google, page ranking, and SEOs have had an affect on blogging as well. I don't mind teenage angst blogs. They don't come up in my searches and if they do I can ignore them easily. MLM and adblogs are easy to spot as well. But nothing is more irritating than lousy SEO spam blogs coming up at the top of a Google search when the blog offers nothing to the subject except keywords. When Google began ranking blogs higher (perhaps around the same time they bought blogger.com?), they made blogs attractive to SEOs as methods of increasing page ranking for their clients. Google adsense (which I use on my blog, I should disclose), also added into the mix creating a market for reams of digital crap whose sole purpose is to keyword web patrons to death with pages filled with text ads and no relevant content. Look into the fiasco the WordPress guys ran into last March (http://arstechnica.com/news.ars/post/20050331-475 9.html)

How is this all Google's fault? As I see it, they have a responsibility to guarantee the relevance of their service by filtering out abuses in a timely manner. They don't have to do this, of course, and nobody forces me to use their service, but Google cannot choose the consequences of their actions. What Google deems as important affects our daily web experience. Like it or not, they shape the Web. If Google allows sites to poison Google's search results then Google not only dilutes the value of their search and blog services, but inconveniences the web as a whole.

Rant all you want about the late comers to blogging ruining things for the "real" bloggers, but some girl's rant about her girlfriend's lipstick isn't going to spoil your internet experience or ruin blogs for you. I read blogs based on content. I'm sure you do as well. So I avoid those blogs like the plague. But when blogs are being used to promote Google's page ranking and Adsense revenues, thus making it harder for us to find legitimate content, then Google has a responsibility to separate the wheat from the chaffe and restore some much needed enlightenment to the blogging world.

You are entirely correct. See IBM v. United States or more recently SCC v Lexmark

Basically, using technical or licensing means to shut a competitor out of one market if they want to use your product in a different market is illegal, whether you phrase it as a licensing restriction (IBM) or even as a DCMA copyright case (Lexmark).

So, yeah. Should they go this route, Apple would have a substantial uphill legal battle.

By the time the OS is ready, Intel will have the Dual Core Desktop Pentium M available.

Jobs has already stated that the Pentium M is a better product than the G5, because of thermal concerns. Hence, most boxes will likely ship with the 32bit, 64bit for workstations, and low power for laptops.

Check this http://arstechnica.com/columns/mac/mac-20050608.ar s

I read a great article on Ars Technica here that shows how Apple is moving the rendering from the CPU to the GPU. Included are some nice graphs that show the relative available bandwidth between the components. To get to the point, it's not that the CPUs aren't fast enough for rendering, it's that the bandwidth to fill the GPU isn't there. Hence, they're moving the rendering to the GPU.

Intel just anounced a cheap low voltage itanium 2 for the 3Q of the year

http://arstechnica.com/news.ars/post/20030908-2787 .html

Do you have a URL for that? Because I had originally assumed that both of these chips were crippled versions of the 970 core and that accounted for the high clock speed... but when I suggested that in another discussion I wasn't able to find a reference.

Do you have a URL for that? Because I had originally assumed that both of these chips were crippled versions of the 970 core and that accounted for the high clock speed... but when I suggested that in another discussion I wasn't able to find a reference.

Do you have a URL for that? Because I had originally assumed that both of these chips were crippled versions of the 970 core and that accounted for the high clock speed... but when I suggested that in another discussion I wasn't able to find a reference.

"The day someone makes a knockoff of Slashdot that's a bit more computer-science oriented and isn't solely aimed at producing the same tired old trolling every day"

Have you seen Technocrat.net? Looks to be just starting, but I'm already impressed: slashdot ran an article on a nanotech textiles protest - technocrat ran one on a group of scientists demonstrating a refined iteration of a carbon nanotube CPU. Comments are on-topic too, touch wood.

(Or there's always ars for CS stuff, but they're hardly a /. knockoff.)

IBM is showing you a 3-core 3.2 GHz "G5" and a "G5" with 8 integrated DSPs, either of which could have been used in a Powermac if Apple was actually interested in them.

Freescale is showing you a G4 that'll run as fast as a 3 GHz Pentium 4 and cooler than a Pentium M and its bridge chips... because it's an integrated CPU with multiple independent memory and I/O ports.

...from John Siracusa of Ars Technica

Q: Will x86 Macs be cheaper than today's Macs?

A: A better question would be, "Will x86 Macs be cheaper than 'equivalent' PowerPC-based Macs would have been had the IBM relationship not gone south?" My answer is "no." Expect Macs to remain more expensive than PCs.

Q: Will I be able to run Mac OS X on a non-Apple PC?

A: No.

Q: Try and stop me!

A: Apple most assuredly will--try, that is. And they'll fail, just like Microsoft failed to stop people from installing Linux and MAME on the Xbox. But like MS, all Apple has to do is make sure that only Slashdot-reading, VoIP-using, PC-assembling, DMCA-breaking geeks hack their way to an "unapproved" configuration of hardware and software. If it's illegal (thanks to the Mac OS X EULA or the DMCA) or at least "technically complex and/or annoying" to run Mac OS X on non-Apple x86 hardware, Apple will be able to absorb any loss in hardware sales attributable to geeks and hardware hackers.

Q: Will future Macs use Pentium 4 CPUs like Apple's x86 developer kit announced today?

A: Probably not. I expect Apple to start with Intel's next generation of multi-core CPUs. Hannibal has more to say about this issue.

Q: Will I be able to run Windows applications on an x86 Mac?

A: Not unless you also run Windows on it.

Q: Okay, will I be able to boot an x86 Mac into Windows?

A: No.

Q: Try and sto--

A: See earlier answer about running Mac OS X on a non-Apple PC. Update: I missed this quote from Phil Schiller. "That doesn't preclude someone from running [Windows] on a Mac. They probably will. We won't do anything to preclude that." My reaction to this new information can be found in the article discussion thread.

Q: Will I be able to run Windows on an x86 Mac?

A: With something like Virtual PC, yes. (Well, VMware, really.) Only it'll actually be fast now, close to native speed if all goes well.

Q: Will Apple provide a VMware-like environment to run Windows applications at near-native speeds on x86 Macs running Mac OS X?

A: No.

Q: Okay, then will someone other than Apple provide one?

A: Yes.

Q: Will Apple continue to design its own motherboards, or will it use commodity PC parts?

A: I think Apple will continue to produce custom designs, or will "bless" a particular PC motherboard/chipset maker (like Intel, for instance...) and contract them to build boards/chipsets that suit Apple's needs.

Q: Will Apple's planned emulation of the PowerPC ISA on an x86 chip really work?

A: It'll be "good enough," but not nearly as good as 68K emulation was on the PowerPC.

Q: Will developers get onboard with such a big change, or will they revolt and abandon ship?

A: If history is any indication, enough developers will ride out the storm to maintain the life of the platform.

Q: Will porting Mac OS X applications to x86 really be easier than porting classic Mac OS applications to Mac OS X was?

A: Yes.

Q: Will Apple maintain an internal PowerPC build of Mac OS X even after moving its entire product line to x86 processors "just in case" they ever need to switch back?

A: I hope so, if only to continue to enforce the discipline of portability.

Q: Is Microsoft worried that every Windows user is suddenly a potential Mac OS X user if Apple ever decides to give up or de-emphasize its hardware business?

A: You bet your ass they are. Don't believe the hype. Microsoft worries about everything, and this is more than a little blip on their radar.

Q: Would Apple ever do that? You know, sell Mac OS X to current Windows users to install on their existing PCs?

A: Someday, maybe, but not soon, and probably only after Apple is convinced that such a market exists and is big enough to be worth sacrificing their own hardware business. How will Apple be convinced of this?

Well given Bob Cringely's well thought position on the strategy of Apple, it makes great sense that Microsoft would position a product in the face of Adobe Photoshop. Acrylic is certainly not the first image editing software Mcrosoft has given away, included with the OS, or with Office. What will be interesting, is the direction MS will take their OS, for it is a rational guess based on their recent move with the XBOX 360 that they will pursue the IBM lineage of microprocessors. This is kinda like Melrose Place, where by the end of it's run, everyone had slept with everyone else before settling down.

Macrumors states that Apple looked into the Cell, but it's not intended for PCs.

The Cell is better suited for a PCI-X support solution, kind of like the ARTVPS render cards, not as a main processor. Although this chip is a beast, it's weaker in some areas then the 970FX Apple is currently using. As an example, its watered down vector unit, doesn't hold a candle to Altivec, which is used by the G5.

Here's a tid-bit from Asrtchnica:
http://arstechnica.com/articles/paedia/cpu/cell-2. ars

"Finally, before signing off, I should clarify my earlier remarks to the effect that I don't think that Apple will use this CPU. I originally based this assessment on the fact that I knew that the SPUs would not use VMX/Altivec. However, the PPC core does have a VMX unit. Nonetheless, I expect this VMX to be very simple, and roughly comparable to the Altivec unit o the first G4. Everything on this processor is stripped down to the bare minimum, so don't expect a ton of VMX performance out of it, and definitely not anything comparable to the G5. Furthermore, any Altivec code written for the new G4 or G5 would have to be completely reoptimized due to inorder nature of the PPC core's issue.

So the short answer is, Apple's use of this chip is within the realm of concievability, but it's extremely unlikely in the short- and medium-term. Apple is just too heavily invested in Altivec, and this processor is going to be a relative weakling in that department. Sure, it'll pack a major SIMD punch, but that will not be a double-precision Alitvec-type punch"

Astroturf much? Do you even understand the Cell's design enough to realize what a PITA it's going to be to develop for?

It looks like fun to me.

(I presume that you astroturf much. How's the weather in Redmond?)

There has already been comments on Cell-Core vs Cell-SPE so I'll be brief.

The problem (and this is a very real problem for all processors, not just Cell or consoles) is getting data into and out of the processor in a timely fasion. If a processor (core/SPE/whatever) doesn't have input data or if it's unable to send data out of the chip then it stalls. If it stalls then that processor part can't get any work done until this block is resolved. This is why it's important to manage input and output of the CPU in order to keep the processor working as much as possible.

Besides this little misunderstanding you can't have "2 running the data", a processor (of any kind) doesn't work that way. The processor itself loads data from memory as part of the operations it executes. So each of them "runs their data" on their own just fine.

And the Cell doesn't have 5 of anything, it has one core and 8 SPE (vector units). Finally if you have 3 actions (data, sound, graphics) and 5 cores to execute them on that gives you 243 different combinations. (And this is counting many duplicates as sound-sound-gfx is calculated separately from sound-gfx-sound.) In any case it's pretty far from an infinate amount of combinations.

"Hannibal" over at ArsTechnica has some great no-nonsense articles (1 and 2) on both the Cell and the X360. I recommend that you read them as he tends to dispell the hype and misunderstandings quite well.

There has already been comments on Cell-Core vs Cell-SPE so I'll be brief.

The problem (and this is a very real problem for all processors, not just Cell or consoles) is getting data into and out of the processor in a timely fasion. If a processor (core/SPE/whatever) doesn't have input data or if it's unable to send data out of the chip then it stalls. If it stalls then that processor part can't get any work done until this block is resolved. This is why it's important to manage input and output of the CPU in order to keep the processor working as much as possible.

Besides this little misunderstanding you can't have "2 running the data", a processor (of any kind) doesn't work that way. The processor itself loads data from memory as part of the operations it executes. So each of them "runs their data" on their own just fine.

And the Cell doesn't have 5 of anything, it has one core and 8 SPE (vector units). Finally if you have 3 actions (data, sound, graphics) and 5 cores to execute them on that gives you 243 different combinations. (And this is counting many duplicates as sound-sound-gfx is calculated separately from sound-gfx-sound.) In any case it's pretty far from an infinate amount of combinations.

"Hannibal" over at ArsTechnica has some great no-nonsense articles (1 and 2) on both the Cell and the X360. I recommend that you read them as he tends to dispell the hype and misunderstandings quite well.

I agree that it will probably never happen, but an Cringley's article reminded me of a bit of news last week:

http://arstechnica.com/news.ars/post/20050603-4968 .htm

It sure would be funny if Dell's "unnamed premium brand" turned out to be Apple on x86...

...how many things REALLY take such advantage of Altivec that its worth keeping it around?

...how many things REALLY take such advantage of Altivec that its worth keeping it around?

Apple loved to pull Phil Schiller onstage to do side-by-side speed tests showing how much faster in real life the G4s and G5s were than their Pentium equivalents. Was that so much BS?

So is Intel going to do a cheaper Itanium for Apple or is Apple going to pretend that 64-bit never existed? Yes to both is my guess, which explains why the word "Pentium" was hardly used in the Jobs presentation. Certainly, he never said WHICH Intel chip they'd be using

The vaunted Intel roadmap is nice, but no nicer than the AMD roadmap, and nothing that IBM couldn't have matched. If Apple was willing to consider a processor switch, moving to the Cell Processor would have made much more sense than going to Intel or AMD, so I simply have to conclude that technology has nothing at all to do with this decision.

In other words, it was an PA-8000 emulator, running on PA-8000. And it very often ran faster!!!! (Between 5 and 40%, occasionally slower, but then it switched itself off and ran natively.)

Obviously there was a trick; and it was that it was able to do stuff like straighten out code, which improved cache usage, and measure how the code actually ran, rather than how the compiler thought it might run, and generally do great run-time decisions.

Apple loved to pull Phil Schiller onstage to do side-by-side speed tests showing how much faster in real life the G4s and G5s were than their Pentium equivalents. Was that so much BS?

So is Intel going to do a cheaper Itanium for Apple or is Apple going to pretend that 64-bit never existed? Yes to both is my guess, which explains why the word "Pentium" was hardly used in the Jobs presentation. Certainly, he never said WHICH Intel chip they'd be using

just mentioning an unnamed 3.6-Ghz development system -- a system which apparently doesn't benchmark very well, either (it's in the links)

If Apple is willing to embrace the Intel architecture because of its performance and low power consumption, then why not go with AMD, which equals Intel's power specs, EXCEEDS Intel's performance specs AND does so at a lower price point across the board? Apple and AMD makes far more sense than Apple and Intel any day.

So why would Steve Jobs --snip-- pre-announce this chip change that undercuts not only his present product line but most of the machines he'll be introducing in the next 12 to 18 months?

The vaunted Intel roadmap is nice, but no nicer than the AMD roadmap, and nothing that IBM couldn't have matched. If Apple was willing to consider a processor switch, moving to the Cell Processor would have made much more sense than going to Intel or AMD, so I simply have to conclude that technology has nothing at all to do with this decision.

The boolean query problem is too bad, since Spotlight itself actually supports it, it just isn't in the UI for any Apple programs.

We need something to help that is clear from the number of digital objects we have lying round on our computers these days. Some method of collecting these objects into conceptual sets or classifications (apart from file extensions which is not always the most useful) could be really useful - I have read some interesting stuff by people who are Metadata crazy (seem to have lost the links though - the tiger review of metadata writer was really interesting...) Maybe the answers are somewhere there.

But for most people, some method of grouping data, adding categorical schemes, visually and texturally organising and generally making files/objects more plastic in the way that we store them would be a great step forward.

But in any case, nested folders *do* still have uses. And I think we need --in addition to-- rather than --instead of--.

---- Posted anonymous as bloody slashdot is banning IP

Sure it's a nice intro to pipelining for those that haven't taken an introduction to computer engineering; but it doesn't have that much to do with how a modern CPU works.

It is true that modern CPUs use pipelining, but the basic model described here is only used if you want to try and implement your own CPU for fun. Modern CPUs are super-scalar monsters with out-of-order execution and custom internal micro code. But I guess you'll have to start somewhere.

If you want look at what a real modern CPU looks like I recommend Hannibal's excellent CPU articles over at ArsTechnica. He has gone over quite a few different architectures as well as the CPUs found in the X360 and PS3. Interesting reading for all! There are also some articles on stuff like cache and memory architecture.

A curious fact about Apple's choice of Intel over AMD, as I learned over on the Ars Technica forums -- AMD's CEO, Hector de Ruiz, was formerly the director of semiconductor products at Motorola.

I think this is one big reason why Steve Jobs and Apple could not / did not consider AMD -- they notoriously burned their bridges with Motorola/FreeScale over the G4's lackluster performance and slow development. Thus, Jobs and de Ruiz probably don't have a particularly good relationship.

I remember Steve Jobs Macworld keynote in January 2003 where he claimed that 03 would be "the year of the laptop".
http://arstechnica.com/columns/mac/mac-12302003.ar s/2

Maybe he was just a couple of years early?

"Have a look at how Core Image works (for example). It's not heavily tied to the PPC design - it's completely and irrevocably bound to the PPC design. It must have AltiVec to work. Sure, you could simulate AltiVec on x86-64 (but not x86) but it'd be crap. Apple would drop Core Image in an x86 transition."

I'm not sure where you got this impression, but it's incorrect.

Currently, Core Image filter graphs are just-in-time compiled to work with two types of SIMD hardware: a GPU or AltiVec enabled G4 or greater processor. This decision is make at runtime based on the hardware present and other performance issues.

Apple could easily include additional SIMD targets for Intel (or even an IBM Cell Processor) to Core Image and any existing code will automatically utilize it if present.

http://developer.apple.com/macosx/coreimage.html

http://arstechnica.com/reviews/os/macosx-10.4.ars/ 15

Apple's mantra has always been use our libraries and you'll get new hardware support and optimizations for free. This could be one of those cases where it really pays off.

... surprisingly, his estimate isn't too far off. Out of curiousity, I dug around to see what Ars' current recommended systems are, and hardware parity with the Mac you describe is somewhere between these two systems, except for the monitor and harddrive:

http://arstechnica.com/guides/buyer/system-guide-2 00505.ars
http://arstechnica.com/guides/buyer/system-guide-2 00504.ars/2

The drives listed are too small, but even huge drives are cheap, and depending on your preference you may or may not want to add another $200 for the monitor (that's the big difference).

So, his range is wrong... $300 is too low for parity. But, parity for $1000 should be easy, and $800 might actually be close to the mark,

... surprisingly, his estimate isn't too far off. Out of curiousity, I dug around to see what Ars' current recommended systems are, and hardware parity with the Mac you describe is somewhere between these two systems, except for the monitor and harddrive:

http://arstechnica.com/guides/buyer/system-guide-2 00505.ars
http://arstechnica.com/guides/buyer/system-guide-2 00504.ars/2

The drives listed are too small, but even huge drives are cheap, and depending on your preference you may or may not want to add another $200 for the monitor (that's the big difference).

So, his range is wrong... $300 is too low for parity. But, parity for $1000 should be easy, and $800 might actually be close to the mark,

Can you tell me how OS X kernel extensions are somehow inferior to having to recompile a linux kernel for driver updates?

Wanna have a throwdown and you try to tell me that /etc/rc shell scripts are better than XML-based launchd (like SMF in Solaris 10)?

Does linux have fine-grained locking in its kernel? Does linux have or support Access Control Lists? How does Linux file-type; MIME or filename extensions, with their weaknesses? OS X Tiger supports both, along with Classic type and creator codes, and the new Uniform Type Identifiers.

Go read John Siracusa's review and analysis of Tiger.

As a man in an orange Bandicoot suit once said, "Booya, Grandma! Booya!"

"...Dynamo is an odd beast. It is, in essence, an interpreter for HP's PA-8000 instruction set that itself runs on a PA-8000 processor. That's right -- it interprets programs that could just as easily be executed natively on the same hardware. For a research prototype, this isn't as strange as it seems. The Dynamo project was started to investigate issues in what was seen as an increasingly important area -- dynamic translation of non-native binaries to native code. For that purpose it doesn't really matter if the original binaries are non-native or not, only that, whatever they are, they're read into some internal form, munged, and spit back out for native execution. The question is only, "How can this translation be efficient, both in time and space?" What's surprising is that Dynamo "inadvertently" became practical. Programs "interpreted" by Dynamo are often faster than if they were run natively. Sometimes by 20% or more. "

It has been a while since I've followed the redhat camp. I was wondering if a reasonable solution exists yet for the dependancy problem that many complained about during package upgrades. As recently as a few months ago, my brother in law switched to Gentoo in frustration after he ran into too much hastle trying to get a SQL package installed on his redhat laptop. The last thing I heard with a bearing on this topic was work being done by Ian Murdock attempting to bring Redhat and Debian closer together. Does anyone know how much that effort has progressed?

I think title of TFA is (aptly, it would seem) entitled "Basics of Intel CPUs. What does the IBM Power series have to do with Intel CPUs?
If you broaden the scope of an article like this to include wildly different architectures like the Power series, you'd have a book instead of an article. Seriously. WTF. If you care so much about the PowerPC, check out Ars Technica's nice collection of related writeups. They're much better than the BigBruin articles...

I think title of TFA is (aptly, it would seem) entitled "Basics of Intel CPUs. What does the IBM Power series have to do with Intel CPUs?
If you broaden the scope of an article like this to include wildly different architectures like the Power series, you'd have a book instead of an article. Seriously. WTF. If you care so much about the PowerPC, check out Ars Technica's nice collection of related writeups. They're much better than the BigBruin articles...