Yellow Dog Linux Finds New PPC Hardware Vendor

inditek writes "C|Net's News.com reports that Terrasoft Solutions, the vendor that sells and contributes to the development of Yellow Dog Linux has found, and continues to look for, some hardware alternatives based around the PowerPC now that Apple is moving to Intel chips. They say Apple's move makes for a good opportunity and more open space for a chip they think has a lot of life left in it." team99parody also writes "This is great news for customers like the US Navy who rely on Linux-on-PowerPC for important tasks like sonar imaging systems."

Comments

pegasos

[larry bagina]

Pegasos sells non macintosh, linux-based PPC machines. At least, they would if they weren't currently out of stock.

Re:pegasos

[erikharrison]

Out of stock is a pretty serious issue.

Besides, legend (okay, Theo De Radt) has it that the firmware is really awful

Terra Soft would probably be best to team up with another minority vendor, like those produce the Amiga One

Re:pegasos

[larry bagina]

i mispoke, the G3 motherboard are out of stock, it looks like they are selling G4 desktop units and motherboards.

Re:pegasos

[Anonymous Coward]

The Pegasos G4 is not out of stock. Genesi sold out the G3
a while ago for the incredible price of 299 USD. With
their great partnership with Freescale and the advanced
AltiVec engine, they focus on the G4 with 7448 upgrades
as the next thing.

Hopefully before end of year they will release a very tiny
(less than Nano-ITX) 2 chip motherboard called EFIKA 5k2.
It is based on the Freescale 5200 CPU at 400 MHz. No heatsink
nor fan required. Very low power with ethernet, USB etc
builtin the CPU. The rest on the support chip. This will
carry 1 2.5" IDE port and 1 PCI. They are currently
looking for a good graphics chip to also put onboard.

Pictures /ironfist

Re:pegasos

[Anonymous Coward]

Besides, legend (okay, Theo De Radt)

Theo might be abrassive and exhibit some pretty anti-social behaviour at times, going overboard with exagerated responses. But he gets some very cool shit done. Love the guy or hate him, the good stuff he does is really good. They have just found a bug in X source which is 10 years old, because of their proactive stance and active mechanisms.

They are embarking on a new memory model change which will spotlight even more bugs in their own software and the software of other open source code. Stability and even security may be harmed in the short term, so that these can be dramatically improved for the long term. It is yet more exciting stuff from them, which ends up helping code quality even outside of their own.

For the trolls, please don't bother complaining about OpenBSD performance BTW. OpenBSD switches everything they can to "secure" even if it means switching away from performance.

Re:pegasos

[anothy]

there are also some pretty serious questions around Genesi as a company. business practice sorts of questions. i know several people who've tried to order hardware from them and come away thinking the whole thing is basically a scam. at the very least, they bill themselves for much more than they are (although that's true of many companies).

Re:pegasos

[oudzeeman]

Genesi was supposed to ship me a demo box over a year ago... I have yet to recieve anything from them.

Re:pegasos

[erikharrison]

For what it's worth, I like what Theo is doing, and has done, which is why I mentioned his comment about Pegasos firmware as being likely legitimite

Re:pegasos

[anothy]

um, no?

but, uh, thanks for asking, i guess.

Continuing PPC Support

[Sv-Manowar]

It should be interesting to see the effect on Yellow Dog post-x86 macintosh, to see how the PPC Linux platform can compete on its own merits. Of course the comparison will be affected by the existing base of PPC hardware and the potential of stalled development given reduced demand for the platform. IBM have been using PPC in their own products, and its possible that their own demand will continue to drive desktop PPC chip development at the same pace as current.

Re:Continuing PPC Support

[PDXNerd]

It should be relatively easy to predict - this is a niche market. Always has been, even when Macintosh was PPC. Now that the major marketer for these chips is gone, the remaining market will still be what it was - a niche segment. Add into that the relatively low market share of Linux and I think we are all literate enough to read the writing on the wall.

AMD/Intel are too cheap, too powerful, and too prevelant.

Re:Continuing PPC Support

[Krach42]

The thing was actually, that Apple was even a niche market within that niche market.

This is why Apple couldn't pull any leverage against IBM or Motorola or FreeScale to actually make the chips that they wanted.

Even Apple wasn't that big a market share in the PowerPC world. In fact, there are more PowerPCs out there than there are x86 chips. "Where are they?" you ask? They're in things like your car, and other embedded devices.

It's like ARM. You just don't realize how pervasive they already are, because the only CPUs you usually ever hear about are desktops.

Re:Continuing PPC Support

[PDXNerd]

I'm sorry, but I seriously doubt your claim that there are more PowerPC's than x86 chips. x86 has been around since the 8086 days.

Do you have any proof to back up this claim?

Re:Continuing PPC Support

[Amiga Trombone]

It should be relatively easy to predict - this is a niche market. Always has been, even when Macintosh was PPC. Now that the major marketer for these chips is gone, the remaining market will still be what it was - a niche segment. Add into that the relatively low market share of Linux and I think we are all literate enough to read the writing on the wall.

You probably have a point there. While Power and PowerPC derivatives will certainly continue to be used in servers and embedded applications, you have to wonder how much R&D IBM will be willing to put into implementations suitable for desktops and notebooks. Even Apple didn't command enough of a market share to make the expense cost-effective for IBM. Somehow I doubt we're going to be seeing any exciting new PPC chips targeted at the consumer PC market.

Which is a shame, really. It's a great chip.

Re:Continuing PPC Support

[Krach42]

It's like the fact that more Ikea catalogs get printed every year than Bibles.

It's damned strange to hear, and you wouldn't believe it, but if you actually bother to do the math, and look at where the things are going, you find out that it's got some darn good proof.

As for real proof. No, I don't have any concrete proof, but I do know that IBM is the largest chip manufacturer in the world, and a large part of their production is PowerPC. Just like I said, they go into cars and other vehicles, not desktop computers.

Still, even were there less PowerPCs than x86 chips, the point still stands that Apple was not the major consumer of PowerPC chips... just the most notable to date.

Re:Continuing PPC Support

[log0n]

I've got no proof to backup the previous poster (taken at face value I'd disagree with it), but if I could modify the argument to say that there are far more risc based (which PowerPC is - a lot of people think the two are interchangable) chips out there than traditional cisc chips (which x86 is - well, vanilla x86 is) then I'd say w/o a doubt that is true.

There are a metric ass-load of risc chips out there - in cars, buses, airplanes, fancy refrigerators, satellites, CNC routers/mills, PDAs, etc etc etc.

Re:Continuing PPC Support

[Lars T.]

It doesn't help much if you want to build a high-end desktop/server/notebook computer. Are you going to use automotive chips in it? Apple's move at least means the 970 and 74xx will get more expensive. The XBox CPU seems to be an exclusive deal, and nobody knows yet how Cell will actually perform for general computing tasks, nor if it will be SMP-able.

Re:Continuing PPC Support

[Guspaz]

Let's not forget consoles. At the very least the GameCube can account for a few million PowerPC chips, and since both the XBOX360 and the Playstation 3 feature PowerPC-based chips, that will put a few hundred million more PowerPC chips into the world.

It depends how you define PowerPC chip though. Both the 360's processor and the PS3's Cell processor use the PowerPC instruction set, but probably don't have "PowerPC" in the model number. I'd say they count, though, just like you would count x86 processors based on chips using the instruction set.

Re:Continuing PPC Support

[Anonymous Coward]

I couldn't find solid numbers.

But I went hunting for the types of systems that Power chips are being used in.

From http://www.absoluteastronomy.com/encyclopedia/p/po /powerpc2.htm

"Design win summaryPowerPC processors are used in many products, among which are the following: Apple Macintosh post-68k models (called PowerMacs), IBM RS/6000 UNIX workstations, Cisco routers, Pegasos (a Commodore Amiga spin off), Amiga acceleration boards, the Nintendo GameCube video game console, and many embedded systems such as the TiVo personal video recorder. Sonnet Technologies and Daystar manufacture PowerPC-based CPU upgrades for use in Macintosh systems. Microsoft's Xbox 360 game console, to be available from the 2005 holiday season, includes a 3.2 GHz custom IBM PowerPC chip with three symmetrical cores."

There's some more on http://en.wikipedia.org/wiki/PowerPC#Embedded_Powe rPC_microcontrollers

Considering that the embedded market is many times larger than the desktop market, I'd be surprised if there weren't more Power ships than x86 out there.

Re:Continuing PPC Support

[DenDave]

Apple is still a PPC vendor and will be for some time.. Don't expect IntelMac to roll out on jan 06 and wipe out the G5...

I actually think that linux on PPC G5/IBMPower5 will become a more important player in the high end market. Customers will appreciate the amazing technology and thanks to IBM's continued support to OpenSource they will actually be able to benefit rapidly.

Whether YDL will be at the forefront of this is of course a different matter altogether. In the high end PPC segment I expect RedHat to move quickly and provide support. Currently AMD64 is the mainstream Linux 64 bit platform but I cannot imagine RedHat not taking advantage of the IBM Power5. This being said, anything RedHat does will be availabel for YDL to implement and hence it will be a marketing question.

I expect to ride Apple PPC until they are obsolete at which point I will asses the performance of IntelMac's versus IBMPower5 systems bearing the cost factor in mind, but I am a private user and so my needs are not nearly as high performance oriented as the professional market.

Re:Continuing PPC Support

[TheRaven64]

So use some of the ones targeted at the embedded market. The G3-series was a very nice chip, especially in terms of performance per watt - and per dollar. For most uses a 1GHz G3 is more than fast enough.

Yesterday, Intel announced XScales at 1GHz due for introduction in Q4 and demoed one running at 1.25GHz. I think the line between embedded and consumer in terms of power is blurring. There are a few things I do that I couldn't on a 1.25GHz XScale (or 1GHz G3) system with 4GB of Flash instead of a hard drive - but not many.

Re:Continuing PPC Support

[constantnormal]

"AMD/Intel are too cheap, too powerful, and too prevelant."

You have data to support this? Everything I have heard was that IBM was considerably cheaper than Intel (don't know about AMD) due to the smaller die size for the PPC chips.

Apple is not your typical razor-thin margins PC manufacturer, and could easily dial up the pricing in concert with a dose of hype to handle a hundred-dollar-per-Mac (or something like that) bump in cpu costs. Please take note: We Macheads prefer the term "boutique market" to "niche market".

It's possible that Apple will get humongous discounts due to lumping Macs in with iPods in their agreement with Intel, and wind up paying less for Intel Mac cpus than for IBM PPCs -- or not. They could easily see their way toward paying more for Intel cpus of lower compute power (compared to then-available G5 chips) and lower thermal power (compared to then-available G5 chips) so long as they get great pricing for iPod cpus of much greater power (throughput and watts) that they don't have today.

My bet is that AMD will get their shot at Apple when The Steve tries to make Intel do the Apple crazy customer dance for less than 5% of its manufacturing volume.

Re:Continuing PPC Support

[encia]

Around Q1 2005, PPC (includes embedded class processors) unit sales per year is about 60 million.

Intel reaches ~200 million units(laptop/desktop/server class processors) per year.

Refer to www.tundra.com/NewsRoom/PressReleases/2005/pr_03_0 1_05.cfm

Re:Continuing PPC Support

[Anonymous Coward]

Try a bit harder... Refer to http://www.tundra.com/NewsRoom/PressReleases/2005/ pr_03_01_05.cfm

Re:Continuing PPC Support

[BJZQ8]

Do you work for Microsoft? I'll bet they have a job at their Linux Lab for you. They have some comparison metrics they want you to help with...

Re:Continuing PPC Support

[Doug Dante]

Bibles last longer than Ikea catalogs.

Re:Continuing PPC Support

[Fishstick]

yeah, plus it's not like they come out with a new version every few months

"ooh! did you check out the spring 2006 Bible!? No? You have *got* to go and get one!"

Re:Continuing PPC Support

[zapp]

I am a developer/support engineer type person at Terra Soft, and the general opinion here (of both higher ups and engineers) is that Apple leaving the arena will be nothing but good for us.

Literally 50% of our development time is spent re-supporting machines after Apple changes them. Things like track pads on power books, northbridge chipsets on towers and XServes, video, sound, thermal control.... they change all these things regularly without a rev bump in model number, or a warning, or providing specs.

With the hardware support side of things going smoother (with our new hardware suppliers), we will have much more time to truly innovate and make a stronger product.

Re:Continuing PPC Support

[confused one]

for every one desktop PC made, there are over 10 embedded processors sold. PowerPC (whether it be IBM or Freescale) is a major player in embedded hardware. x86 just doesn't see that much use in embedded applications.

The company I work for builds instrumentation -- we use PIC or ARM for the low end; and, PowerPC for the high end. It's anecdotal, but representative...

It's Surprising

[under_score]

It's surprising where "odd" hardware/software combo's show up. I would never have suspected Linux/PPC in the Navy. How did it get there? Who knew about Linux, and PPC and had the influence to get it used there? Was it a really good sales job (and the connections that make it possible)? Or was it an insider who went looking for a platform from a clean slate?

The answers to these questions are extremely important to the further expansion of the use of Linux (or any other product/platform/system).

Re:It's Surprising

[ArbitraryConstant]

PowerPC was probably the best choice for sonar analysis when they designed the system, and now that it's not the fastest they've still got to support the systems they have in place because it's fast enough.

Re:It's Surprising

[maetenloch]

Also depending on where these systems are deployed and what kind of power/cooling/space constraints they had, the Navy may have been more focused on flops/watt rather than absolute performance.

Re:It's Surprising

[wargolem]

Maybe the Navy realized the security risks involved with a homogeneous network. That's at least one great reason to seek out a non-standard (yet reliable and efficient) platform, such as Linux on PPC.

Re:It's Surprising

[pkb]

AltiVec is big in military applications. Sonar, radar and such are imaging problems at heart.

Re:It's Surprising

[cmacb]

It probably doesn't matter that they are PowerPC or Intel processors. If the OSs used are Linux, and the applications are written in any language except the Microsoft lock-in variety, you just change processors, recompile, and you are back in business. Avoiding lock-ins gives you tremendous flexibility, bargaining power, and fallback options. Being locked into an OS, and thus destined to use whatever hardware your OS provider commands you to use is a sad sad thing, and hopefully a thing that won't last too much longer. Unless tech decision makers are even dumber than I think they are.

Re:It's Surprising

[dj245]

Bingo. A professor of mine (I'm studying marine enginering) once told a tale of designing a navy destroyer. Among other things, the anchor windlass turned out to be bigger than the manufacturer had initally said (due to a communications mishap) so they had to bubble the deck, an ugly and undesirable feature of a multi-million (billion?) dollar vessel.

In addition, the radar set they ended up using (because of the required output) put out so much waste heat into the radio room that a bigger AC unit had to be installed in that space than initially designed, but there wasn't anywhere to put it so they stole ceiling space and made the room 5 feet tall.

It could have easilly gone the other way and they could have searched out a better radio from an obscure manufacturer (Transmeta, if they made radios...)

Re:It's Surprising

[antifoidulus]

Um, that might work for your average PC program, but I wouldn't try doing that on a complex, incredibly expensive real time embedded system where failure can equal lives lost....
Something tells me the transition isn't nearly as simple as a /. post.

Re:It's Surprising

[Wyatt Earp]

Also PPC chips in the F/A-18E/F for a variety of systems, I think they are 200-400 MHz models if I remember correctly, my manual on it is...in a pile of books somewhere.

Re:It's Surprising

[larry bagina]

If the OSs used are Linux, and the applications are written in any language except the Microsoft lock-in variety, you just change processors, recompile, and you are back in business.

Or unless it uses assembly language.
Or unless it uses processor specific compiler glue (like gcc's SSE/MMX/Altivec support).
Or unless it makes byte-order assumptions.
Or unless they were using a PPC only compiler (like IBM's xlc or metrowerk's mwcc)
Or unless they were using 3rd party libraries
Or unless a lot of things.

Re:It's Surprising

[clymere]

You're just describing security through obscurity...which we all know is bad. Its also not entirely true in this case, its still Linux, and depending on their configuration, a lot of exploits would still be valid.

Furthermore, PPC Linux has a smaller userbase than x86. There are fewer people working on the code, fewer people testing it, and therefore fewer bugs getting worked out. In other words, one could make a pretty good case for why there is reason to believe it is a LESS secure platform.

I think that other posters are right on the mark with other good reasons why they would choose Linux PPC, power consumption and heat being on of them. I doubt they are running G5's.

Re:It's Surprising

[rindeee]

I can tell you based on my experiences with similar projects within DoD that they are pretty good about evaluating "conventional wisdom" and then evaluating any "unconventional" options. In a case like this, there would have been a set of requirements to be met. A team of folks would have been assigned to go out and find as many solutions as possible. After the solution set was identified, they would evaluate each one on it's merits and arrived at a final selection group which would then be decided on based on either price, longevity or some other more utilitarian criteria. Chances are very good that a "clean slate" approach is what led to Linux on PPC.

Re:It's Surprising

[Chandon Seldon]

It really depends on the specifics of the program. If the inital design wasn't heavily performance bound (and therefore was written for correctness), a unix program should be portable to pretty much any archetecture / unix OS.

It's true that a major change would require a new testing cycle, but it's definately possible shouldn't be *too* hard. If a program is so embedded that this isn't true, the fact that it's a unix app becomes irrelevent.

Re:It's Surprising

[Jeff DeMaagd]

A lot of flight control systems were PPC for this reason, but last I heard, there was a strong resistance to Linux, I think in part due to perception, and the rest due to the cost of validating / certifying it as a flight control operating system.

I don't know if the perception is warranted or not, I know it's a pretty tough set of shoes to sell. The Aerospace industry is very conservative, not wanting their products to crash and burn (literally of course), so it takes a decade or so to make changes.

Re:It's Surprising

[wax66]

I can attempt to answer that question to a small degree.

During a WWDC for Apple a few years ago, I sat down to lunch next to a Secret Service agent. I was curious to see what he was doing there, and he told me he was evaluating the security risks of Mac OS X (this was around public beta time), as well as other interesting information. He said he's the reason the Army switched to Macs running Mac OS 9 (that and having been hacked not too far back) running WebStar (I think that's the name). He also told me about trying to convince the Whitehouse to switch, and about how they were running Sparc/Solaris and their admins weren't that great. He was also trying to convince other branches of the service of what would be the most secure for their needs.

So it could very well be the Secret Service, or even that specific agent that did that particular push. Or something completely different...

Re:It's Surprising

[saider]

This system may be using a PPC and running Linux, but that does not mean that it is a simple PC. More likely than not, this is a very complex real-time system with many processors. The Linux is probably only used for the user interface area to give the user a somewhat familiar interface to the system. The software that the user commands is probably highly specialized to run on the particular hardware it is built on. Redesigning systems like this is not as trivial as being able to "just change processors, recompile, and you are back in business". Especially with a government contract, you will probably find it easier to pay a premium on buying end-of-life parts rather than designing a new system.

Re:It's Surprising

[vasqzr]

If they sell 30 million game systems, it won't be enough.

Last time I checked, they have been selling 150 million PC's a year for the last few years.

Re:It's Surprising

[wargolem]

Running a non-homogeneous network is not quite security through obscurity. I was implying that the Navy's network is most likely comprised of many different platforms, so if a bug gets loose, it will be isolated to a smaller portion of the network. I more think of security through obscurity putting plaintext passwords in a hidden file, but a non-homogeneous network provides damage control in case the shit ever hits the fan.

Also, just because PPC Linux has a smaller userbase than x86 Linux doesn't mean than PPC Linux is less secure. By your logic, MS Windows should be the most secure desktop platform because it has the highest desktop userbase. Plenty of brilliant people are working on PPC Linux, and you may have noticed a while back that even Torvalds is running Linux on PPC. So it's not the size of the userbase that matters; it's the caliber of the people writing the code.

And don't be so quick to rule out G5s for sonar work on a boat or submarine. I'm sure their power requirements are fairly strict, but while a G4 is consumes less power than a G5, the G5 doesn't consume that much more power. The G4 is rated at 10 watts, and the G5 consumes between 20 and 40 watts. Even a G5 is worlds better than a Pentium 4, which consumes over 100 watts. I agree that the G4 consumes less power, but both the G4 and G5 are energy efficient, and unless you know the Navy's requirements, you can only speculate as to which CPU they're using.

Re:It's Surprising

[clymere]

Good points. In that context, the hetrogenuous network thing does make some sense.

As far as windows...sure more people use it, but how many people are auditing the code? Its certainly not going through the same process. So i guess i should have been more explicit and said "code auditors" or "bug fixers" instead of users.

Its nice that Linus has run PPC Linux, but that does not inherently make it secure. He may have reasons for running it that have nothing whatsoever to do with security. It may even just be that a friend gave him a nice shiny new dual-G5 :) Just because hes the head kernel dev does not mean that security is his biggest priority...and in fact, there are many people hacking on the kernel who specialize in it, and likely know it a lot better than he does. In other words, its neat to know what the guy is running, but it probably has little bearing, other than knowing that PPC Linux isn't likely dying out anytime soon.

I'm also no expert on PPC linux. It may in fact be as or more secure than x86. I was just throwing out some observations.

Isn't it odd?

[Jukashi]

Isn't is odd that the only time something is described as having "a lot of life in it" is when its pretty apparent the thing is dying? Why not be a little more honest say something is "not as dead as it appears"?

Re:Isn't it odd?

[rockinrobotix]

Just because the chip has been left by apple doesn't mean it doesn't still have a huge amount of development being done by IBM for their servers, Motorola and the embedded market, and then the fact that two of the next major consoles will be using it. Even if it isn't exactly desktop, any development is good development.

Re:Isn't it odd?

[strider44]

From what I gather the point is that Apple's market share in that architecture is just a small (even insignificant, when looking at the extremely small profit margins) part. It wasn't really a big loss at all to IBM with Apple jumping ship. The PowerPC architecture is anything but dying.

Re:Isn't it odd?

[OO7david]

I'm sure if the PPC could speak for itself it'd go something like this:

Apple: Here's one-
Public: Ninepence.
PPC: (feebly) I'm not dead!
Public: (suprised) What?
Apple: Nothing! Here's your ninepence....
PPC: I'm not dead!
Public: 'Ere! 'E says 'e's not dead!
Apple: Yes he is.
PPC: I'm not!
Public: 'E isn't?
Apple: Well... he will be soon-- he's very ill...
PPC: I'm getting better!
Apple: No you're not, you'll be stone dead in a moment.
Public: I can't take 'im like that! It's against regulations!
PPC: I don't want to go on the cart....
Apple: Oh, don't be such a baby.
Public: I can't take 'im....
Apple: I feel fine!
PPC: Well, do us a favor...
Public: I can't!
Apple: Can you hang around a couple of minutes? He won't be long...
Public: No, gotta get to Robinson's by nine today....
Apple: Well, when's your next round?
Public: Thursday.
PPC: I think I'll go for a walk....
Apple: You're not fooling anyone, you know--(to Cart-master) Look, isn't there something you can do...?

(they both look around)

PPC: I feel happy! I feel happy!

(the Cart-master deals the old man a swift blow to the head with his wooden spoon. The old man goes limp.)

Apple: (throwing the old man onto the cart) Ah. thanks very much.
Public: Not at all. See you on Thursday!
Apple: Right! All right....

Re:Isn't it odd?

[Bastian]

Just like the z80 was all but dead when everyone quit making desktop computers that ran CP/M? I guess all those countless TI calculators, Game Boys, cell phones, and the like don't count. Come to think of it, I think I have 4 devices that use a z80 sitting on my desk right now. I bought two of them in the past year.

There are craploads of things out there that use PPC chips that are not Apple computers. It most certainly does have a lot of life left in it.

Re:Isn't it odd?

[Arker]

Huh? PPC dying? What planet are you living on? There are more PPCs shipping than ever before.

And btw what kind of crack are the moderators smoking? Parent is about as "insightful" as the guy that says "nice weather, huh?" in the middle of a hurricane.

Re:Isn't it odd?

[DrXym]

And how many would be capable of running Yellow Dog Linux, what with it being a desktop OS and all? In the context of desktop computers, the future of the PPC is all but cast in stone.

Re:Isn't it odd?

[niteice]

Isn't is odd that the only time something is described as having "a lot of life in it" is when its pretty apparent the thing is dying? Why not be a little more honest say something is "not as dead as it appears"?

BSD? :P

Dead parrot sketch? :P

Oh Dear

[eno2001]

I can just see the new troll: "PPC is dying. Anandtech confirms it".

Heh. I remember getting all excited about PPC back in 1994 when Apple first announced the move. It seemed like it was a natural and logical extension from the 680x0 family (one of the best CPUs ever for desktop systems). It's kind of sad how it didn't wind up being as much of a player as it should have. Even the guy who wrote Minix quipped back then that the future would be everyone running some kind of *nix OS on their PPC desktops. Now that dream is gone because even Apple went with Intel. I sure hope Intel can get it together and make a decent CPU/Mobo combo that dumps all backwards compatibility, BIOS and segmented memory.

Re:Oh Dear

[eexlebots]

They did and it is called the Itanium.

oh.

Rats.

Re:Oh Dear

[Mad Merlin]

Here's the quote you're looking for (circa January 1992 by Andy Tanebaum, "LINUX is obsolete" in comp.os.minix):

Of course 5 years from now that will be different, but 5 years from now everyone will be running free GNU on their 200 MIPS, 64M SPARCstation-5.

Part of a huge flamewar between Linus, AST and a number of other people on microkernel vs monolithic kernel design. Here's the entire original thread if you're looking for some good Sunday reading.

Re:Oh Dear

[LoadWB]

I'm curious about the timeline of announcement, official announcement, and final release of the PowerPC chip. I remember back in 7th grade (1986-87) I had a teacher who was a big Mac artist. It was around that time I was getting interested in digging into microprocessor architecture and was learning about the TMS-9900 (in my TI-99/4A) and the 6502 (in the school's Apple and Atari computers.) I constantly envisioned and drew crude plans for a system which would integrate the 99/4's PEB and an Apple II, including a design for a combination computer (keeping in mind that I only basically understood the concept of interfacing different chip architectures... so there were of course no real engineering plans or OS design.)

I digress... Mr. McPhail introduced me to 68000 CPU and steered me to some magazines in the school library which talked about the upcoming "Intel-killer" called the PowerPC. I was told it would be powerful enough to emulate the Intel CPU and would revolutionize the computer world.

Given that this was around the time that I learned of the Amiga and how it could emulate a Mac, I easily swallowed this concept.

So, does my memory deceive me, or is the time-line I remember accurate??

Re:Oh Dear

[KillShill]

even the macos9/10 still have/had segmented memory.

the "64bit" osx also is segmented but you won't run into that limit for quite a while. windows 32bit and 64bit face the same limitations.

there is no such thing as flat addressing, because costs keep it out of the picture. current end-user 64bit cpus use 48bit virtual memory addressing and 40bit physical, including x86-64.

in a way manufacturers have a point... most cpus won't last more than 5-10 years... i just wonder though how much die space and costs are saved through this kind of corner cutting.

Re:Oh Dear

[TheRaven64]

Itanium was the second attempt. The first was the i860.

Re:Oh Dear

[zxking]

The reasons the PPC did not become the x86 killer it was predicted to be are largely business-based rather than grounded purely on technical merit. The PPC is superior in a number of areas like SIMD however, IBM did not market it to Microsoft in the same way that Intel pushed the x86 for Windows. IBM placed its bets instead on the lesser(but very capable) players, Apple and Linux, who did not achieve enough market share.

However, the PPC is long from dead. IBM has been strongly cementing the PPC as the ideal platform for embedded computing in an effort to displace other commonly used processors such as MIPS and ARM. The biggest example of this is that both the PS3 and Xbox 360 will feature what are essentially PPC upgrades. Another good example is the embedding of PPCs in Xilinx FPGAs to provide a complete SoC solution.

Re:Oh Dear

[LoadWB]

Couldn't resist, eh? Well, some guys have hobbies like young boys and dirty magazines, and others like young technologies and "dirty" ROMs. What can you do?

Re:Oh Dear

[eno2001]

Hehehehe. You sound like me. I still remember all the hype about Pink Taligent being the "convergence box" of the future.

See here for a good synopsis of the history of the PPC chip. As usual Wikipedia comes through for me again.

As a side note, I remember I had this one math teacher in high school (1987) who got me into computers. He did this really cool "paper computer" in his class where we had to act as CPUs and use our own registers on paper to work out a simple math problem. You didn't have to know the answer to the problem necessarily since you were actually moving bits on paper and should get the right answer anyway. This guy was also into the Timex Sinclair ZX81/1000 (Zilog Z80 CPU) computer. I swear he mentioned something about having connected several of them together at home and used them for various home automation tasks. Pretty cool guy. Math teachers CAN rule.

Re:navy

[Krach42]

That their existing software is all written for it.

True, they could recompile for a different architecture, but that costs money, and test time.

So, they're better off continuing with PowerPC hardware.

So, pull your head out of your ass that everyone can just jump ship from a chip design when it isn't going well for them, and shut up.

Interestingly enough, the newest Apache modifications (that I had heard about, this was Spring 2001) put a Voodoo 4 in for the HUD displays. So, again, the question, "what shit can they accomplish with a Voodoo 4, that they can't with something else?"

Nothing, but their contract says they're doing the project at a certain price and they've already made their choice, and bought the chips.

Re:navy

[zulux]

What is so damned special about PPC for sonar imaging.

It's working.

Seriously, when something is working, don't screw with it. Just leave it alone, and use your talents to solve a new problem.

Continuing Cell Support

[Anonymous Coward]

"It should be interesting to see the effect on Yellow Dog post-x86 macintosh, to see how the PPC Linux platform can compete on its own merits. Of course the comparison will be affected by the existing base of PPC hardware and the potential of stalled development given reduced demand for the platform. IBM have been using PPC in their own products, and its possible that their own demand will continue to drive desktop PPC chip development at the same pace as current."

The Sony/IBM/Nintendo Cell comes to mind. Plus all the embedded systems out there.

--
The "are you a script" word for today is untidy...much like Taco's code.

Re:navy

[Osty]

Interestingly enough, the newest Apache modifications (that I had heard about, this was Spring 2001) put a Voodoo 4 in for the HUD displays. So, again, the question, "what shit can they accomplish with a Voodoo 4, that they can't with something else?"

They obviously couldn't have used a Voodoo3, since 22bpp isn't enough to allow for nice alpha on a HUD. Duh.

The battle rages on.

[keilinw]

This is such an interesting discussion. Which CPU is better? Better is obviously in the eye of the behold as price, power, Mhz, and apparently performance per watt matter. I was at WWDC and played with the new MacTel boxes. Interestingly the single 3.6 Ghz pentiums appear to run faster than the dual 2.7 Ghz G5's.

OK... so I'm not going to go there... but Intel is apparently coming out with some interesting new hardware. I don't know everything about it.. but it appears that they will be chainge the x86 architecture altogether. So was Apples move speculative or desperate?

Another interesting thing as brough up by the author of the post to which I am replying.. liies in the fact that certain companies are inexorably tied to their hardware. Some institutions, for example, running Pro Tools may not be able to upgrade to the new hardware as their software will not be availible. This is speculative... but it is possible.

So I'd be interested in an arcitcle that clearly lays out the differences between the PowerPC and Intel architectures and maybe even one that examines Inte's new architecture as well.

The PowerPC is undoubtably an excellent platform....but there are other factors to be considered.

Re:The battle rages on.

[Bastian]

So was Apples move speculative or desperate?

Probably more on the desperate side. Laptops are now slightly more than 50% of the market. I don't have any numbers, but I wouldn't be surprised if it's even more for Macs, where you don't get people buying big gaming desktops and the cheapest desktop isn't less than half the price of the cheapest laptop.

The G5 is power-hungry, hot, and decidedly not suitable for mobile and low-power applications. It probably never will be, given how little pull Apple has with CPU manufacturers. And the G4 is more than ready for retirement.

Academic arguments on the relative advantages of PPC and x86 just don't play into the issue. If Apple wants to continue to sell computers, they really have no choice but to jump ship on PPC.

Re:The battle rages on.

[VMEbus]

Here is a pretty good article on Anandtech that compares the architecture of PPC vs x86 and also has some performance tests. http://www.anandtech.com/mac/showdoc.aspx?i=2436

Re:The battle rages on.

[Anonymous Coward]

You can get the numbers from the Apple 10-Q fillings (page 26). The sales ($$) and units shipped numbers are close with laptops trailing slightly in 2004. In June of 2005 the laptop numbers have slipped. This is probably due to the age of the Mac Laptops. (No processor upgrades.) In the same time period sales of Mac desktops and servers have grown 50% while laptops sales have grown by less then 10% (comparing 3 months in 2005 and 2004). Meanwhile other manufacturers have enjoyed tremendous growth in laptop sales.

Apple switched to Intel because Apple needed a strong laptop processor. It is a desperate move after waiting too long for a competative laptop PPC chip.

Re:The battle rages on.

[Bastian]

Heh, those numbers aren't what I expected at all, but I guess I shouldn't be surprised.

The next numbers I'd want to see (and which I am again too lazy to look for) are how much of that 50% increase in desktop sales are due to the Mac Mini, which I am sure is not a candidate for a G5 CPU, either.

That said, I don't think Apple just waited too long. I have a feeling they've been working on the switch to x86 (at least as part of a contingency plan) since not long after the G5 came out at the latest. I'm sure craploads of non-Darwin OS X code needed to be modified, what with the switch from OpenFirmware to a BIOS and the endianness and all. The writing has probably been on the wall for anyone who is in the know on Apple's dealings with IBM and Motorola for quite some time.

Re:The battle rages on.

[cfuse]

Another interesting thing as brough up by the author of the post to which I am replying.. liies in the fact that certain companies are inexorably tied to their hardware. Some institutions, for example, running Pro Tools may not be able to upgrade to the new hardware as their software will not be availible. This is speculative... but it is possible.

Why doesn't apple (or some third party) simply release a hardware compatibility card like they did in the good old days? It can't be that difficult to wack a G5 onto a pci card and get it to handle the G5 specific code.

Mod parent up!

[Huh?]

Don't fix it if it aint broken. Just because you can doesn't mean you should. Etcetera...

Re:heh, linux on a mac - what a waste

[larry bagina]

not entirely. Linux is mainstream now. If you want to be a 133t p0ser, you gotta run linux on something obscure like an alpha, sparc, or ppc.

or Fedora Core

[Huh?]

http://fedora.redhat.com/

Re:or Fedora Core

[b17bmbr]

i couldn't get my G3 ibook to work with fedora 4 PPC. Don't know why. It'd install fine, reboot, go through boot process, then hang somewhere. Never could figure it out. Even booted with resuce disk, changed /etc/inittab to run level 3, nothing. No idea. Now the g3 is in my classroom and my g4 ibook is home and I can't play around with the g4. i owul dlove to run fedora on the mac. i had yellow dog 3.01 some time ago on the g3 and it ran much faster than jaguar. now, panther runs really well, plus i use keynote every day. still, i'd like to run gcj4.0 and native compiled eclipse. oh well.

Re:or Fedora Core

[superpulpsicle]

I used to have an old mac that I tried to run PowerPC linux with. I have never had a hard time with any distro. This one had me stumped for good.

I couldn't believe my eyes, but I gave up. That was also the last time I ever tried linux on mac hardware.

Re:or Fedora Core

[b17bmbr]

my old bondi blue 233mhz/160mb ram imac ran really nicely with yellowdog 3.01. I hosed it and installed panther for games for the kids. it is kinda sad that ppclinux just doesn't do as well as x86 linux with hardware, etc. but ppc is an awesome linux platform if it does work. in fact, i think it can be better than x86. i know it's supposed to be platform independent, but linux was built ground up to be unix on x86, and probably will be best there, unless a company puts resources into it, like sun or ibm. apple i imagine is in no great hurry to push linux, not they care the same level microsoft does.

Re:navy

[mclaincausey]

Perhaps the SIMD unit affords some sort of advantage for the applucation.

In the best of all worlds

[Joseph_Daniel_Zukige]

Apple, when the iNTEL road map lead turns out to be mostly a paper lead yet again, would keep both iNTEL and PowerPC lines. And then start porting Darwin to ARM, CELL, and other available CPUs.

(Yeah, CELL would require a port. That's probably the point that Steve got sidetracked on. My guess is the discussions of re-writing for CELL produced a lot of complaints, and a lot of, "if we're going to have to do that, why not re-write for iNTEL?" Silly middle management.)

And in the best of all possible worlds, Linux on non-Apple PPC would push commodity mobo vendors to start focusing on alternative CPUs and alternative hardware. Then we could get out of the blasted rut we are in, where, in so many shops, you can't buy a project unless it includes a whole bunch of false standards.

Re:In the best of all worlds

[KillShill]

cell = 8 simd vector units (memory starved--256KB?!)
with an in-order powerpc core.

tell me why cell is anything remotely like a new instruction set?

same old tech with bolted on vector processing units... which btw, the xbox360's ppc's also have 3 vector processors (altivec).

rehashed technology but somehow the fanaticboys cannot piece it together.

Re:In the best of all worlds

[wootest]

(Yeah, CELL would require a port. That's probably the point that Steve got sidetracked on. My guess is the discussions of re-writing for CELL produced a lot of complaints, and a lot of, "if we're going to have to do that, why not re-write for iNTEL?" Silly middle management.)

Your post is a sea of red herrings for all I know. First of all, Apple *never* ported - as in 'rewriting' - to x86. OpenStep was working perfectly fine on x86 when they got it, and they simply continued to maintain that port but dropping the public releases of it after Mac OS X Developer Release 2 - right after the Aqua interface was introduced. When Steve Jobs proclaimed that "Mac OS X has been living a secret double life for the past five years" at this year's WWDC, he meant it, and he also noted that *all golden masters of 10.0 thru 10.4 have been built for PPC and x86*. If you still think this is a "rewrite" port, I don't know what to tell you.

Second of all, from all I've heard, the Cell is very powerful for some tasks but more or less sucks at some other tasks, and it's very much a niche CPU. If a primary CPU needs to be anything, it's well-rounded. I doubt that Apple will release a Mac with the Cell as a primary CPU - however it's not impossible that they license it to Sony, whose PS3 page says that it can run "Apple's Tiger", or that they simply stick a Cell or two in as helper processors in late PowerPC models.

I also don't doubt for a minute that Mac OS X runs on at least one other architecture beyond x86, the new Intels and PowerPC. Intel was their contingency plan when they went with PowerPC - it's only logical that they'd want a new contingency plan now, even if the need may not be as big if they've got secret deals with AMD too (which it's hard to imagine they haven't).

Re:In the best of all worlds

[el_womble]

This whole double life business has interested me since the iNTEL announcement. At an application layer, you don't need to know or care whether your writing for x86, PPC or ARM, you write to the framework and APIs provided by the OS / runtime of your choice. If that OS /runtime works on other architectures you simply cross your fingers, say a little prayer and recompile.

My, admittedly limited, understand of OS writing is that the exact opposite is true. You are dealing with the processor directly and performance is crucial so ASM is used a fair bit. Now at a hardware level x86 and PPC are very different. They have different strengths and weaknesses which need to be addressed, and yet Apple said that with just a single, secret, dept. they managed to get their microkernal up and flying on two vastly different system architectures. (I think microkernal is probably the key word here)

My thought has always been, that if MACH is so well designed that it can be this easily ported to x86 and PPC simulateously, there is nothing stopping Apple from supporting Cell, ARM or SPARC within a few months of an announcement. Especially as they have no convinced a large portion of its loyal developers to switch to XCode.

One area I can see the Cell exCelling (sic) is set-top box TiVo and H.264. Its like it was purpose built for the job ;) OK, it might not be the perfect general pupose CPU, but for the iLife suite and Quicktime the tech specs look like it could really make them fly.

I could see Apple shipping the boxtop mini, for 200 bucks that plugs straight into you Hi-Def TV and provides web access, PVR and H.264 movie downloading out of the box via a remote control and if you buy a Apple Bluetooth keyboard and mouse for $50 extra you can do your homework from your armchair. OK so Mathematica will perform like a dog, and Pages won't be as snappy as it is on a $500 mini, but you knew that when you bought it.

Re:In the best of all worlds

[wootest]

The "boxtop mini" that you're describing sounds like it's part of a bigger strategy with the long rumored Video iPod, and might also go with a similar AirPort Express AV product. What Apple's great at is leveraging technology and building on top of existing products. iTunes is based on QuickTime, iPod is a simplified parallel of iTunes and AirPort Express builds from iTunes in a smart way.

So I think that if they *do* go all out and make a box set thing, they'll probably do the Video iPod first. And you're right that the Cell would be a perfect choice for such a device (the box set) - one of the first big Cell demos was about decoding MPEG4 streams in parallel, even. However, I wouldn't think that they'd allow you to turn it into a computer because I think that they know damn well that operating a computer on your TV has always sucked and been clunky. HDTV may mean a half-decent resolution but that's only part of their problem. I think they'd just want people to get a Mac mini instead.

Re:In the best of all worlds

[demon]

At an application layer, you don't need to know or care whether your writing for x86, PPC or ARM, you write to the framework and APIs provided by the OS / runtime of your choice. If that OS /runtime works on other architectures you simply cross your fingers, say a little prayer and recompile.

Not exactly. Especially with anything requiring high performance, like using vector features of the host CPU, often requires some customization for the architecture. In Apple's case, they provide APIs for a lot of that sort of thing, but they can't cover every need.

My, admittedly limited, understand of OS writing is that the exact opposite is true. You are dealing with the processor directly and performance is crucial so ASM is used a fair bit.

Er... not _that_ much. Do you use Linux? Have you looked at the kernel? You should. Particularly you should observe the relatively small section of the kernel tree that's architecture specific, let alone the amount that's actually assembly. It's not that much really. That's because the kernel, once it knows the basic operations on the CPU, will be dealing with a lot of the same (or very similar) hardware regardless of the CPU - PCI bus controllers, display controllers, input devices, storage controllers, network controllers, IPC code, etc. The wide majority of things, once you get past the basics, are going to be pretty universal.

and yet Apple said that with just a single, secret, dept. they managed to get their microkernal up and flying on two vastly different system architectures. (I think microkernal is probably the key word here)

If they couldn't keep the core of the OS working on more than one architecture (and considering the pedigree of some of their developers, this should be a cakewalk), then I would have to call them "stupid". As was pointed out, OpenStep and Mach were already platform agnostic before Apple took an interest. That's what OS X is built on. And no, the microkernel has little, if anything to do with it, as OS X is not a "true microkernel" OS. It runs a mostly unmodified *BSD kernel on top of Mach, just making use of some handy Mach features (like its IPC, which is (ab)used widely as part of Aqua/Quartz).

My thought has always been, that if MACH is so well designed that it can be this easily ported to x86 and PPC simulateously, there is nothing stopping Apple from supporting Cell, ARM or SPARC within a few months of an announcement.

Er... Cell == 64-bit PPC core with lots of vector units strapped on. It wouldn't be much of a "port", and given they've abandoned the Power family, I doubt they'd turn around now. Apple doesn't care about SPARC - that's Sun's baby, and they're rather on the expensive side. (Oh yeah, and there's only one company making SPARC-based laptops - Tadpole - and they're not going to be setting any records for battery life.) ARM has lost its former glory now that it's owned by Intel (see xScale), these days it's almost exclusively used for embedded purposes. In short, you're drawing at straws.

I could see Apple shipping the boxtop mini, for 200 bucks that plugs straight into you Hi-Def TV and [...]

Er... no. Apple already has a business. They tried going in on the Pippin years ago, and that died a pretty quiet death. I doubt Steve Jobs cares a damn bit about the set-top box market.

Re:In the best of all worlds

[squiggleslash]

My, admittedly limited, understand of OS writing is that the exact opposite is true. You are dealing with the processor directly and performance is crucial so ASM is used a fair bit.

Believe it or not, a large proportion of OS X's kernel is written in C++. You can download the source to Darwin, which contains the kernel, from various websites. I don't think Apple cares that much about performance in terms of the types of things achievable with assembler tweaks.

There are various low level operations you generally have to resort to assembler for, but for the most part, modern programmers tend to avoid such things, especially as the days of the 68000 and 6809, where you could program something in assembler, look back, and say "That's a beautiful program", are no longer with us. Putting too much in assembler results in maintenance problems. And also means you can't get around a future processor incompatability simply by tweaking the compiler and recompiling all your code.

Plug plug plug

[NekoXP]

We finally get on Slashdot and nobody mentions the bloody company name!

ARGH! :)

http://www.genesi.lu/

Neko

Re:Plug plug plug

[Anonymous Coward]

Why not also mention what a con-artist company it is?
http://www.morphos.net/

or what OpenBSD thinks of Genesi?
http://www.openbsd.org/pegasos.html

Re:Plug plug plug

[TheRaven64]

Unless you R'd a different FA to me, the company name was Mercury Computer Systems. Genesi was mentioned only in passing.

Re:Plug plug plug

[NekoXP]

We have a good relationship with everyone who is not a dickhead.

Dale Rahn is a nice guy, who had some tangible problems with the way our
engineering team operates. It's a shame about his boss, though.

IBM?

[mr_zorg]

Uh, how about IBM? They continue, and do doubt will continue, to use PPC in a lot of *their* hardware. And they're big supporters of Linux... No doubt you could get a Linux on PPC setup from them!

Re:IBM?

[killtherat]

I've worked with IBM, bought a bunch of JS20s (2.2GHz 970 PPC powers those things). But typically IBM pushes Suse Linux Enterprise Server 9 for Power. Yes, that will run on a PowerPC chip. This is mostly because it is compatible (and certified to run) with IBMs cluster management software CSM and their Global file system GPFS, it's the natural choice when installing a set of these machines.

Yellow dog just isn't big enough to offer the kind of product certification and support that Novel can offer.

Re:Yellow Dog versus Debian?

[commodoresloat]

yellow dog is rpm-based. debian is debian. yellow dog is probably running several bleeding edge components and comes configured with a cool interface. debian is debian. you're going to hear all kinds of arguments on both sides of this question; issues about RPM vs. apt-get and whether stability is better than currency and so forth; but the bottom line is, use debian. I mean, OS X.

a lot of life?

[ChipMonk]

The chip may perform well, but when the memory management requirements result in code that "can be used to scare small children" (L. Torvalds), I have to wonder: What good is an excellent chip if using it is so difficult?

Re:a lot of life?

[Neil Blender]

What good is an excellent chip if using it is so difficult? by ChipMonk (711367)

You're not really a Chip Monk, are you? Or is this a "one hand clapping" type of question that Chip Monks ask their students?

Re:a lot of life?

[Gorbag]

There's difficult, and there's difficult. While operating system programmers may have to sweat out these details, the vast majority of us write applications, and such details should be completely hidden at that level of abstraction.

Re:a lot of life?

[OverCode@work]

I don't think it would be wise to show kernel memory management code to a child in the first place. :) It does tend to be pretty gnarly.

PPC code actually doesn't look that bad in the end. Be sure to turn on your sarcasm filter when reading Linus.

-John

Re:navy

[NewbieProgrammerMan]

"It works don't touch it" might be fine for your little project that doesn't really matter, but in the real world you always need to be planning for the future.

While you might hope that big organizations would "plan for the future," I haven't seen it happen very often. Often it takes a lot of money to develop those big "real world" projects, and a lot of project managers aren't willing to pay to upgrade to a new platform solely because it might be harder to upgrade in the future. Sometimes it's just more cost-effective to load MS-DOS 6.2 on that shiny new 3GHz dual Xeon instead of paying somebody to update some application to run on WinXP.

Re:navy

[kinema]

It's working.

I would say that it has more to do with the fact that the system designers looked at the availiable COTS CPUs and decided that the PowerPC was better suited to the task. Most likely the PowerPC's SIMD/vector unit (AltiVec) was superior to the offerings of other similar processors namely Intel's SSEn on the Pentium IV.

After the choice was made to go with the PowerPC/AltiVec processor piles and piles of hand optimized ASM code was created by some very well funded geeks to perform what I'm sure is an ultra high bandwidth and sample rate siganl processing system.

Re:heh, linux on a mac - what a waste

[KillShill]

and you didn't name a single obscure platform.

if more than 2 people have heard of it, it isn't obscure.

Re:Yellow Dog versus Debian?

[Arker]

I believe that Debian is ported to the PPC. How does Yellow Dog compare to Debian as a distribution? If I could use Yellow Dog on the x86 would I have a reason to use it instead of Debian?

No.

Yellow Dog is based on Redhat. Debian is... Debian. Score one for Debian.

Yellow Dog comes from a single company that will sell you a support contract. Debian is an open standard, if you need a support contract you can choose from several competitors, and if the one you choose initially gives you any problems, you can dump them and move to another without having to change your software. Score two for Debian.

Debian supports nearly as many platforms as NetBSD, meaning that you can run a very heterogenous environment, PPC here, X86 there, ARM over in that corner, SPARC behind that wall there... and have the same tools, use the same methods to administer each one, regardless of platform. Yellow Dog runs on PPC, so if you have anything else in your environment, you'll have to learn to admin Yellow Dog, plus something else. Score three for Debian.

You can get a PPC Linux system from IBM right now*

[Anonymous Coward]

*if you understand the fact that the difference between POWER and PowerPC is pretty small and IBM uses the term 'Power Platform'. They are API compatable in linux-land. (meaning the system that I use on my PowerPC laptop will work just fine on a Power 5 server without recompiling...)

Go check out OpenPower for starters.

These are server stuff specificly to use the Power 5 proccessors with a Linux-specific machine.

For example the low end of it is a Power 710 'express'. A rack mount system with a 1.65 ghz Power 5 proccessor with 36 meg cache(!), 2 gigs of RAM, and 2 73gig 10k SCSI drives.

Very fast, huge cache, ok amount of memory, ok fast harddrives. 4,500 USD

You can get dual proccessors for under 10k, which realy realy realy kicks the ass of anything you can get from Sun for that price range. The Power 5 systems with their 36meg cache and HUGE transistor counts blow the AMD opterons out of the water.

IBM does not have a OpenPower workstation, and does not have a OpenPower desktop though. These are server/database systems and it shows. They AIX workstations you can order, but have Linux installed on them instead if you wanted to, and those aren't much more expensive.

IBM's stuff has always been expensive though. I'd rather have a army of 3rd party manufacturers make PowerPC machines.

However I don't see much of a point, other then platform snobishness.

Personally I like my PowerPC lappy; a Apple Ibook, but it's the last one I'd buy because it's video card is the ATI 9200 and is the last supported by Open Source drivers.

There is the R300 project for newer cards, but I don't think that it's paticularly usefull at this point (although I am gratefull for it, don't get me wrong.)

Having a PowerPC machine realy drive home the values of having free software.

Free software is stable, it's cross-platform, and it 'just works'. All propriatory software runs like ass on my system, if at all. It's a night and day difference.

Trouble is, what is the advantage to PowerPC desktops over x86?

NONE that I see. The newer intel setups are faster, use less power, and are supported well by open source drivers.. much better then the overpriced apple hardware. The ibook when I bought it was vastly superior to all small Pentium 4-m systems aviable and was cheaper.. for the 12 inch long-lasting-battery form factor.

Since then Intel has surpassed it wholy with it's Centrino/Sonoma stuff.

(and beleive me, the x86 Apple stuff will be overpriced, too. I'd probably avoid it personally)

Truth is they are both proccessors, they do both the same thing. Other then price and speed, the differences are purely academic at this point when considuring their use with Linux. Both work fine, x86 allows propriatory applications easily, PPC doesn't.

To me they are on equal ground. If third parties start suppling powerpc laptops that are well supported by free software, I'd strongly considure it.. but otherwise I realy don't care to much.

IBM needs to realy get in gear about their Power systems otherwise they will simply lock themselves into a small high-end market with slowly, yet consistantly, shrinking share.

If the average geek AND the average developer can't have easy access to PPC machines then Linux will stop being cross platform in a few years. It's ineveitable, and there is nothing nobody can do about it. Most people don't have 5000 dollars to burn just to have a extra server in their basement, or feel like spending 1000 dollars for a slow ass machine of lesser quality then what they can by at walmart for 300 dollars.

The biggest hope for future PPC machine in my future will be the Sony PS3. If they release a Linux distro for it, I'll buy it in a second. At 3ghz with a limited core it will be somewhat faster then my aging AMD desktop and my 1.2ghz Ibook. The SPE's offer interesting possiblities and will be fun to mess around with, especially when it comes to things like ray tracing and whatnot. If Sony gets Nvidia to release drivers for it it can actually have the possibility of being a rather kick-ass Linux box, otherwise it will just end up being a nice toy.

Re:info on mac linux support

[Arker]

Look here

Re:Yellow Dog versus Debian?

[jericho4.0]

Why would you have a reason to use debian in the first place? Assuming you need a workstation, rather than a server, ubuntu, mepis or knoppix are much better choices for the desktop.

Secret life

[otter42]

Isn't this the moment when Yellow Dog Linux should declare that Linux has been leading a second, hidden life all this time, where everything has been compiled on both PPC and x86?

Yeah! And then we can all debate the wiseness of changing Linux from a PPC platform to x86!

Now, all we need is someone to crack Yellow Dog Linux so it'll run on an x86... I've got a developer's P4 with Linux already installed. I'll put the installation iso on bittorrent as soon as I make sure that there are no "unique identifiers".

Power for military use

[slashflood]

This is great news for customers like the US Navy who rely on Linux-on-PowerPC for important tasks like sonar imaging systems.

IBM just teamed up with a company called Mercury to build Cell-based computers for (military) applications:

As a result, demanding applications such as radar, sonar, MRI, digital X-Ray, and many others can be taken to new levels of sophistication and performance.

And as we all know, the Cell is basically a Power processor.

Re:Power for military use

[slashflood]

the general purpose core sure is yes, a slow one though. The SPEs might be good for the applications you mentioned but as far as a processor for general purpose computing its not even close to state of the art.

We discussed this already.

You can find all the available information here.

But the Power-compatible Cell processor does a good job in imaging applications.

Re:Power for military use

[encia]

Note that Cell's PPE (control CPU) is dual issue in-order processor. It's nothing like PowerPC 970.

Re:heh, linux on a mac - what a waste

[Anubis350]

hmmm... what about a toaster

Sure it's an every day appliance, but as a computing platfrorm it's both obscure and suprisingly functional (now you can have your OS and eat it too)

First bsd, next step linux (perhaps a beowolf cluster).... ...and my karma dies a painful death

Re:Yellow Dog versus Debian?

[vhogemann]

Well,

I had both Debian and YellowDog running on a PowerMac G3 system for a while. Debian at that time (Sarge/Testing) was superior, both in terms of number of avaliable packages and performance.

YellowDog 3 CDs just can't compete with the Debian repositories. There are 4 DVDs worth of software on Debian!

And now, with Ubuntu... I don't really see a reason to go with YellowDog as a Desktop distro.

Of course they offer support, and many companies depend on this "feature". But I guess they'll gladly support your PPC Debian systems if you ask (and pay) them to.

Re:navy

[el_womble]

LOL Its trolls like this that makes browsing at +2 Flamebait worth it.

Clones?

[gilesjuk]

Could this in theory mean someone creating PPC machines that could be used to boot PPC MacOSX?

I know the machine would need some boot code, but there's sure to be someone who could write some.

Re:Clones?

[Durf]

No! It means creating NUCLEAR SUBMARINES that can boot Mac OS X. WinTel fans have no answer for this one.

Unfortunately, the first run of "cheese grater" models are all sitting on the bottom of ocean trenches. Nicely water-cooled, but they'll be working on a version without all those holes, you can bet.

Re:You can get a PPC Linux system from IBM right n

[Hes Nikke]

If the average geek AND the average developer can't have easy access to PPC machines then Linux will stop being cross platform in a few years.

here's a cheap PPC based linux system... you might already have one in your living room...

here's an even cheaper one - though I'm not sure if it's PPC based or not...

i wouldn't worry about availability if i were you.. ;)

(Yes, those links are amazon associates links... if you really feal like sticking it to me, click here, and here for the "clean" links)

Re:navy

[DrXym]

If "it's working" here and elsewhere, where is the payoff for Yellow Dog? If people don't want to replace their systems, they're certainly not going to pop a Yellow Dog CD into the tray (assuming it even has one) and install a brand new OS.

And if as has been suggested most PPCs go into embedded systems, Yellow Dog is totally unsuitable. There are plenty of embedded solutions for Linux out there and I expect they all work quite well already.

It seems to me that YD has nowhere to go. They could offer a way to reuse old Macs (including the rack mounted ones), and there might be scope to cosy up to IBM, but what else is there? Personally I think it would be fantastic if Sony or someone licenced YD to produce a Linux for their console. The PS2 had a kit, so perhaps the same is possible for the PS3 as well - but hopefully cheaper than the last one.

Re:Yellow Dog versus Debian - me too

[NadMutter]

I had a similar experience when I picked up a used B&W G3. After trying the 'specialist' yellow-dog (and finally finding a free download rather than a buy it link), it was problematic.

The network card was incorrectly identified as sun happymeal (hme) when infact it was bigmac chipset (bmac) - a 30 min fix. Followed by editing XFree86.org to get out of 1024x768 to the native 1280x1024 my LCD screen defaults to. Change the firewall settings to enable sshd and load that in. I never got round to getting sound to work.

Found the packages I wanted were pretty old (Octave for example) after a while, thought I'd try Ubuntu.

Network autodetection and setup - check
Screen resolution ok - check
Sound - 75% ok (xmms needs working; other packages ok)
Getting packages ok - both apt-get and the gui version work (minor tweak to enable the universal packages).

For me, the free volunteer-based distro got me up and running faster than the commercial flavor.

For those with US-Navy sized budgets, prepared to pay for support and build-to-order, probably a different story.

(formatting messed up in an attempt to get round the lame lameness filter).

Re:navy

[leinhos]

And if as has been suggested most PPCs go into embedded systems, Yellow Dog is totally unsuitable. There are plenty of embedded solutions for Linux out there and I expect they all work quite well already.

YDL is a handy development platform for embedded PPC targets (at least for PPC 750s and 74xxs). I'm not sure about the other embedded PPC processors.

errr IBM make plenty of PPC kit

[zenst]

IBM make PPC kit, have done for ages, amazing realy given they invented it and prolly have more PPC kit out there than any other vendor including APple put together.

Next obvious question today will be - how do I use a keyboard to type stuff :)

Not quite true - NeXTStep WAS ported to x86

[brokeninside]

NeXTStep didn't support x86 until v3.1. Prior to that, it ran only on Moto 68k processors and before being ported to 68k, it ran on something else, which escapes me at this point in time. Version 3.1 also supported Sparc and PA-RISC.

NeXTStep itself was never ported to PPC. That was Apple's contribution in 1996/1997. Apple, with the engineers inherited from their acquisition of NeXT, ported the compiler toolchain and kernel of NeXTStep in the course of a few months. Rasphody was, in effect, the first version of NeXTStep for PPC.

The astonishing thing was that apps that targetted NeXT, for the most part, only needed new NIB files and a recompile to run on Rasphody.

YDL on XBox360 or PS3?

[argent]

By the time Apple quits shipping PPC, the XBox360 and PS3 are likely going to be out. How big a deal do people anticipate it to be to run YDL on them?

The number I read

[brokeninside]

When the rumor mill first started going about the recent change to x86, I read somewhere that Apple's orders for PPC chips amounted to 5% of the annual capacity for 1 of IBM's PPC factories.

Yellow Dog Vs. Ubuntu

[Nahooda]

I tried Yellow Dog Linux on my iBook G4 and although it installed easily, it didn't work well because the kernel was missing some extension to activate the cooler (of course, a friend told me, to download the kernel extension and recompile the kernel but... i must confess: I'm using Linux for years now, but still don't know how to compile a Linux kernel...).

Besides I didn't like the pre-installed theme and look'n'feel very much. I tried to install some stuff, but the package manager gave me error messages regularly and I couldn't compile a lot of apps since some packages were missing (good old Linux dependency hell ;-P) that usually ship with any other distribution I tried (uhmmm, well I only tried SuSE Pro but I'm just pretending to be helluva Linux expert ;-P).

In contrast to Yellow Dog, Ubuntu was another world. Easy-to-use, easy-to-install, easy-to-configure, great package management and cooler support.

My conclusion: I don't know about desktop Macs but if you want to use Linux on a portable Mac forget about Yellow Dog, use Ubuntu instead.

Regards,

Dennis B. Schramm

Re:You can get a PPC Linux system from IBM right n

[Chirs]

"If the average geek AND the average developer can't have easy access to PPC machines then Linux will stop being cross platform in a few years."

You are overlooking the veritable army of linux developers being paid by the embedded guys to make linux run on ppc. These guys will still be there long after Apple stops using the architecture.

There are literally *dozens* of ppc varents supported by linux. The PC-style ppc systems are by far in the minority.

Plastic lasts longer than wood.

[douglasq]

So what's your point?

Linux lock-in

[Donny Smith]

>This is great news for customers like the US Navy who rely on Linux-on-PowerPC for important tasks like sonar imaging systems.

Is this the way to read this sentence:
Although they use Linux, it's Linux on PPC so they're screwed and migration to industry standard (Inel/AMD) Linux would cost them a fortune.

If yes, well that's great publicity for the open and standard Linux OS.

If no, what the hell is that supposed to mean?

CARDIAC!

[janic]

You are thinking of CARDIAC - "Cardboard Illustrated Aid to Computing"

Go here for a reconstruction in Java.

Cheers!
John

Re:Yellow Dog versus Debian?

[jurv!s]

Yellow dog is just a Red Hat rebuild. YDL 4 is based off of Fedora Core 2. You won't have to learn something else for your x86 boxen. Tough luck with the SPARC and ARM boxen you have laying around...

Re:a lot of life (only when life matters)

[delire]

What good is an excellent chip if using it is so difficult?

It's not difficult. Apple just choose to do it very badly.

Cheapest way to acquire a Linux/PPC system?

[Nichotin]

I know about, AmigaONE, Genisi/Pegasos, Apple, and that you could purchase Apple-parts from ebay or something. Are there any other ways (which are within a student's price range), to have a working Linux/PPC system? It does not need to be _blazing fast_, but OK for a typical office computer.

Re:Cheapest way to acquire a Linux/PPC system?

[toddestan]

Cheap??? Where an old used G4 tower will set you back more than a new Dell system?

Re:You can get a PPC Linux system from IBM right n

[milimetric]

I don't mean to be rude, but I don't understand why 4500 dollars for that low end OpenPower system you describe is something that someone would want. I set up the following system for my dad for 2200 dollars two years ago:

Dual Xeon 2.4 Ghz
15K SCSI drive of 39 GB
10K SCSI drive of 78 GB
7200rpm IDE drive of 200 GB
2 GB ram
Supermicro motherboard with built in Ultra 320 SCSI controller
Windows 2003 Server

Besides the cache, everything else is about the same or better than what you describe, it hasn't stopped working since I set it up.

TiVo

[kris_lang]

I believe that TIVOs also use PowerPC processors, e.g. IBM PowerPC 403GCX (see http://www.9thtee.com/insidetivo.htm or the picture of the chip at that same page. It also uses an IBM MPEG-2 Decoder chip.

Re:The battle rages on. (For the Clueless)

[Bastian]

Heh, except that the folks who are looking at buying laptops aren't paying attention to the things that will make a good ruggedized computer. They're looing for nice big megahertz numbers, they're wondering why they can't get PowerBook G5s, and they're thinking about how you can get more performance for your dollar on a PC than with a Mac.

wait two days

[Joseph_Daniel_Zukige]

and it's a headline on /.

Check the specs. All the vector stuff would have to be re-written for CELL. Without the re-write, all you have is a fast G3 and some hardware doing nothing.

Now, if you ask me, the re-write would be worth it. OpenGL, for instance, could be re-implemented on the CELL SPUs, and the OS could gain some nice benefits for the visual output. But that would still not be making very good use of the SPUs.

GarageBand and QuickTime (client) could also eventually get huge gains. But there's so much that would have to be done by hand first that Apple would have to have a chip with a full G5 core _and_ a complement of the SPUs in the high-end boxes to support the dev, with little visible gain for a year or two. Then, finally, Apple would be able to use one of the Sony PS3 chips effectively in a Mac Mini.

Making the SPUs fully useable in Mac OS X would likely require a complete new API, and that's going to hurt before it feels better.

And how does one convince the PHBs at Adobe?