Slashdot Mirror
Perfect Pair: PowerPC And Linux
grubby writes: "A member of my local LUG NCOLUG has written an article in LinuxJournal about what the PowerPC and Linux could do together. He brings up many good points about the history of the WinTel PC and what he hopes the future may bring. I have personally had numerous conversations with him about his ideas and would like to know what the slashdot population thinks about them. Check it out, it's a good read." This piece takes a somewhat broad view of things -- which makes sense, given that radical changes can take time to grow in the background before they actually make public waves. Also of interest on the PPC front: kilaasi writes "Looks like IBM is getting back to it's PowerPC which is/was/will be co-developed with Motorola. IBM has some tricks that will increase speed and at the same time decrease power consumption." Here's CNET's brief story on upcoming PPC developments."
Getting closer. Kindall was out. His wife would not sign the NDA.
The goal of this prodject wouldn't be the fastest desktop avaliable. Infact, I would imagine that stability would a much more useable feature.
Imagine a small linux box, sold for, say $300, and with optional subscription support... A machine that can be your companies Webserver, Firewall, Internet Gateway, Mail server, File server, Print server... And it's all useable with a simple web interface from (only) the internal network.
Obviously, this little machine couldn't take care more than a small buisness, but it would be enough to get a small buisness on it's feet on the internet, as so many are using the current "model" of having somebody do all the hosting/webserice for you. (Blech, most of those suck, BTW.)
I guess what I want to see is an inexpensive linux box, you can take out of the box, plug in to your network, plug into your internet connection (weather it be modem or ethernet), and get the basic funtionality out of the box. The instructions should have the user use a web browser to setup his initial settings (Donaim name, Services on/off, passwords, etc.) and that should be about everything.
And, now getting back to the article, the PPC chip would do this just as well, or better than an intel chipset.
I guess I'm just wishing here. I'd love to make something that's that easy to do.
Pathway
Forgive my spelling.
The ONLY problem with PPC is that it's very closed hardware. I.e. If you buy a PPC based system you will likely have a single source for upgrades and replacement parts. As is typical in such situations this single source charges an arm and a leg for everything.
What this translates to is that you should not be considering low end or midrange PPC systems for anything that doesn't absolutely depend on that architecture. High end is different in that sometimes you just have to have the fastest no matter what. As for stuff that's bound to PPC. If you have it you have no choice. Linux isn't one of those things.
Everything I have said here applies to Spark, PA-Risk, S390 and some very proprietary, name brand PCs. For most jobs generic x86 is the best choice because after you choose it you can still decide where to buy each and every component, thus forcing the vendors to jump through hoops to get at your money.
--
Quidquid latine dictum sit, altum viditur.
Whatever is said in Latin sounds profound.
--= Isn't it surprising how badly I spell ?
The article mentions it, yes.
I'm just pointing out that, in my opinion, this is the key reason that PowerPC hasn't caught on as an architecture like it could.
Topher
Restricted and limited in that I have to buy a system preconfigured by Apple, according to what they want to sell me.
;-)
There isn't a lot of choice. . . I'm forced to buy what they're willing to sell me, and that's it.
Now, in a PC, if it starts becoming obsolete, I can go out and buy a new motherboard and CPU, slap it in, and bang, my computer's ready to live another few years.
What do I do with my becoming obsolete Mac? Doorstop, anyone?
Or, what happens if I want a desktop machine with SCSI disks? As I recall (and I may be slightly wrong on this), Apple no longer offers non-IDE drives in their desktops. I don't want to buy a machine just to rip out half of it to replace it.
The fact that they finally moved to the industry standard PCI bus is a great thing (Although, I don't know that I've yet seen a Mac with more than 2 slots), don't get me wrong, and the USB is nice (firewire is nifty, but still poorly supported by devices, though it's getting better). However, there's a lot more to flexibility than that, and I don't like being limited to only what one company is willing to sell me.
Topher
Laptops are a completely different world from desktops. When dealing with laptops, Apple machines aren't as bad of a buy.
However, as impressive as the PowerPC processor is, you will *not* get the same performance for the buck out of a Mac that you will out of an x86 machine. It won't happen.
Period. This isn't FUD, this is reality. When you are dealing with a single vendor who has to subsidise their research and development costs through hardware sales, that's what happens.
And don't pull that "Total Cost of Ownership" crap. TCO is based very heavily on the previous experience and knowledge of the users. If they have Mac experience, they'll have less trouble with Macs. If they have Windows experience, they'll have less trouble with Windows. Same thing with administrator experience. What you know will cost you less.
Topher
First of all, the RS/6000 is a very different beast from a PC or a Macintosh. It's an enterprise level Unix workstation/server comparable with a Sun UltraSPARC, Compaq/DEC AlphaStation, or an SGI workstation.
Additionally, if you want to be strict, the Power CPU used in RS/6000's is not exactly the same as the PowerPC processor. It's a more advanced 64 bit version.
So, there is a *huge* difference in what you get betweeen an RS/6000 and a Macintosh. They have different purposes, different uses, and shouldn't really be directly compared.
Now, as to my second point, I never said I would buy an RS/6000. In fact, I was trying to make the point that for general use, they are much too expensive. Your average person cannot afford to spend that much on a machine.
Lastly, I don't 'hate' Apple. I dislike many of their business practices, as well as their Operating System (pre-MacOS x), but I don't hate them. Honestly, I don't care enough about them as a company to hate them.
Topher
You do the PPC injustice. The benchmarks here show that while you'll save a couple bucks with an x86, a properly tuned G4 system is superior.
That's very amusing. You provided a link to a page which back up my argument, and failed to counter it. If you read what I said, I stated that you "you will *not* get the same performance for the buck" with Mac vs. x86.
I am *not* saying that the PowerPC is a bad processor, in fact if you read an earlier comment, you'll see that I specifically state that if I could buy PowerPC CPU's and motherboards directly, without buying a Mac, I'd be the first in line for one. Now, here's a direct quote from the URL you gave:
This is exactly what I said above. You get more bang for the buck with an x86 system than you do with a Mac.The fact of the matter is that TCO is a valid argument, which is probably why you are trying to derail it. Go on eBay, and you can find three-year old Macs being sold for 50% of what they were purchased originally. How does your x86 stand up to that?
Total Cost of Ownership (TCO) is one of the most subjective and misleading statistics ever created. It's sole purpose is for computer/software/OS manufacturers to use to convince you that it's better to pay more money for their stuff than someone elses, because you'll save money in the long run.
The fact is, TCO depends as much, or often times more, on the people you have and the experience and knowledge they have, than anything else. You can't give a single TCO for a system, and have it be realistic for every buyer of that system.
As for your statements about buying Macs on eBay, that is unrelated to this argument. All that does is prove that Apple has a restrictive stranglehold on the Macintosh market, and that they artificially inflate prices in order to create revenue.
Also, remember that Apple/Mac and PowerPC are not necessarily synonymous, and should be treated as separate entities. I love the PowerPC processor, however I dislike many of Apple's business practices, and don't like how they inflate hardware prices to cover research and development costs, as well as software costs. They should either make up their mind and choose to be a software or a hardware company, or they should separate their products more and open up to allow for outside R&D.
Topher
Yes, actually, I do. ;-)
Many things have changed since the mid '90s. Computers have become ubiquitous. Back then, they were still a very expensive and somewhat uncommon item, even if they were gaining in popularity. You didn't find them in millions of homes across the US, as well as throughout the world.
The PowerPC processor never had a real opportunity to break into the general PC market, because of lack of support. When it was first introduced, Apple was the only major non-embedded user, and shortly after, IBM. Availability of PowerPC CPU's and motherboards was never at the same level as Intel compatible parts. Additionally, that was a different world of PCs. Five years ago, it was almost unheard of for someone without a technical degree to build their own computer, while it's become very common today.
Also, on the Operating System front, if you built a PowerPC system, what OS would you run? Until Windows NT was released, with PowerPC support, there was no Microsoft compatible operating system. There was MacOS, if your BIOS would support it, but then why not just buy a Macintosh? Also, shortly after NT 4 was released, Microsoft stopped suporting NT on PowerPC.
Today, there is a very viable Operating System in Linux that can be run on it, and two lesser known alternatives, BeOS and NetBSD (and OpenBSD) which can be used on PowerPC systems. None of these existed in a viable form five years ago.
Five years ago, if I were building a PowerPC system myself, I would be worried about what OS I would put on it, what applications I would run, and where I would find them. With Linux and the plethora of available, that question has an easy answer.
Five years ago, if there were PowerPC CPU's and motherboards available to build a PC with, I wouldn't have cared much. I didn't have any use for it. Today, if they were available, I'd be among the first in line to purchase one. So yeah, again, I do think things would be different today. ;-)
Topher
What I think we really need, before we'll see the PowerPC really take off, is the ability to build PowerPC Systems.
For example, if I want to build an Intel (or compatible) system, I buy a CPU, a motherboard, memory, any devices I need, I put it together, and boom. A system is born.
I can even do that with Alpha systems, though it's a little more work.
However, has anyone ever seen PowerPC CPU's and motherboards available anywhere?
I love PowerPC systems, I think they're absolutely amazing. However, I don't have the $10k to drop on an RS/6000, and I have no interest in purchasing an over-price Macintosh computer. They're too restrictive and limited.
As soon as I can buy a PowerPC CPU and motherboard somewhere, though, and build a system myself, I'll be the first in line to do it. Linux runs amazingly well on PowerPC processors.
Topher
Not really. Sure, there's a complex instruction set, but these days, it's essentially a RISC core, and the x86 opcodes get mapped to internal RISC ops for execution. Furthermore, PPC (and SPARC and the others) have been moving further and further away from true RISC, and their instruction sets are anything but reduced now -- although admittedly, they're nowhere near as bloated as the x86 ISA. The only true RISC chip left with any sizeable market share is probably ARM. Personally, I miss the Motorola m88k. One of the nicest instruction sets I've come across since the 6502...
"The invisible and the non-existent look very much alike." -- Delos B. McKown
Heck, you don't even need a laptop.
The [NEC SimpleEm], available from Dynamism, is truly cool.
Flat panel, touchpad, keyboard -- all connected by IR. No cables. Sweet.
And very much eye-candy!
--
--
Don't like it? Respond with words, not karma.
--
Apple would be highly unlikely to help take potential Mac OS X seats away from themselves in such a way.
Apple makes a lot of money off of sales of their PowerPC-based hardware.
They don't make dick off of Mac OS, X or otherwise.
Very few people have as part of their PC-buying decision the mental exchange "should I use Linux or MacOS?" The markets don't overlap that much.
It would be in Apple's economic best interest to continue to support PowerPC Linux development, and to actively market their hardware to Linux users when they feel the OS is mature enough on their platform.
The only real change they need to make in the hardware to support this is the addition of another flippin' mouse button or two.
However, every Unix person that I know that has played with OS X is impressed... It's a sharp OS.
Every Unix person I know is impressed with OS X, too; and still running Linux.
Two of the three web sites listed in the article are dead. So where was this wonderfully huge market that was tauted?
Methinks a little more market research and a little less empty speculation might be in order.
by Mike Buddha -- Someday the mountain might get him, but the law never will.
Intel and Apple have both admitted it in their digital hub rhetoric. Desktop pc's are losing their sex appeal. Rich Gold from Xerox Parc's "Ubi-Comp" is the way things are going. For recent terms...that's ubiquitous computing. "puters are all around us. Cell Phones, Palms...Microwave Ovens.
Do they run on x86....hell no. the only ubiquitous computing machine that could benefit from that 20+ year old architecture is a toaster...'cause the heat could do some good there.
Own a tivo?....guess what...you have a Linux PowerPC box in your house that you rely on more than your x86 Linux box.
That's how this is going to go down. You're going to buy all these gadgets that have nifty comm features. the PowerPC takes a whole lot less energy and MHZ to be just as powerful as an x86. Guess what Linux will be in your embedded systens....the one who sucks less.....power that is.
The other end of this fork is the lifespan of hardware. x86 PC's usually are useful for 3 years before bloat renders them useless. Macs usually take 6 years to be annoyingly slow.
Seriously...I bought the Mac I'm typing this on in 1996. It was only a month ago that I had to switch to SuSE on it to keep it tolerable.
so yeah. If your goal is world domination of Open Source...then x86 Linux is a pitfall. The landscape is moving away from PC's and into nifty gadgets running PPC or ARM.
Stop wasting your time with that packard bell in the dumpster and focus on todays cool tech. embedded Linux on PPC...and some cool desktop hardware as well
"Let him go, Ralph. He knows what he's doing." --Otto Mann (simpsons)
Comparing prices of lots of 1000 processors, back in 1997 when it's 2001... Not especially useful information.
And i don't know what exactly you're talking about in reference to the PCI bus, but taking sound off of it spares 150KB/sec out of 132 MB/sec per CD quality channel... basically a drop in the bucket. And videos' been off of the PCI bus in x86 computers for years now... AGP, remember?
before anyone posts a rant saying they cant get hold of stuff
apples are Power systems and your local electronics shop are able to order them from MOT and IBM (at least they where here in the UK via RS electronics ) and thats with a ATX board
IBM has released some nice docs and BOM's on building the hardware you can find it on the semi page
MOT seem to want to push the Power Arch hard and have seen various roadshows and freebees to entice developers
you can find alot of people useing it e.g. TIVO
and lineo likes it alot
what I would like is for GCC to get sorted because the code output is not as optermised as it could be personaly I just cant wait for GCC 3.0
but IBM keep the powerPC code under bitkeeper wich is a bit (haha) strange at first but works well try it out
just a couple of veiws
regards
john jones
How can you say that end customers killed the product by not buying it, when there never was a product? At worst, you could say they killed it by being predicted to not buy it.
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
[Wild guess] 20 bit addressing, maybe? "640K is enough for anyone" doesn't sound nearly so stupid when your competitors are limited to 64K.
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Actually, PPC is a tactical choice too. Look at recent Macs: no fans, and they only sip the juice compared to my AMD-brand guzzlers.
PPC may be a good choice if you wanna build a box that is going to be powered on 24x7, such as a server.
All we need now is for something to happen that makes people think about power and cooling issues. But what are the chances of that happening?
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
...might actually materialize from the years of vapor.
bplan just a couple of weeks ago released images of their prototype PPC ATX motherboards. Part of the reason for this board is to appeal to us Amiga nuts so we can run AmigaOS clones like MorphOS on it, but Ralph has said that the board will be fairly generic and that Linux will also run on it. (Which is a good idea if bplan actually wants some serious sales volume.)
So maybe some day, a sensibly-priced PPC box will be available (well, actually iMacs aren't such a bad deal).
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
I just have to ask: Why?
The powerPC is a nice chip, but I don't think the 'linuxPC' concept would wash, at least not as a powerPC based system. It's been hard enough to get popular commercial software ports to intel Linux, much less different architectures. Examples: Acrobat reader, Flash plugin, Arcserve client. 'All' it would take is a re-compile to run them on PowerPC, but good luck getting commercial vendors to do it.
Look at some of the existing attempts to package linux appliances on non-x86 CPUs: Remember the sidewinder and netwinder box that Corel used to make? Looks like their new products were switched to a transmeta (ie: x86) CPU. How about Cobalt? Gone are the MIPS based models, now they run x86 chips.
Why would the average customer at a discount retailer like Best Buy purchase the PowerPC based linux system (running at, say 2GHz), vs. the Intel/AMD/Transmeta x86 based system (running at, say 6GHz)? (speed ratings inflated assuming this doesn't happen for a few years)
The RISC vs. CISC point is almost moot nowadays, given that the more advanced Pentium and Athlon chips are essentially RISC chips that emulate the x86 CISC instruction set.
Now if you'll excuse me, I'll hunker down in my anti-flame bunker.
"But actually trying to use m4 as a general-purpose langage would be deeply perverse" --ESR
It's been a while but as I recall the Z80 only had a program counter that was 16 bits wide so it would only handle 64K of memory. This was great when all we ran was CP/M on the Z80. The 8086/8088 had something like 20 bits for address so a whopping 1MB could be accessed although some of that memory space was reserved for BIOS and stuff so that left 640K to work with.
As for IBM choice of the 8088 rather than the 8086 I heard they wanted an 8 bit data path since 8 bit devices were much more common and cheaper at the time than 16 bit devices were.
No, he doesn't.
Why would he want the results of executing $DISTRO there?
You may want to try reading the sh man page before speaking next time.
Glückwünsche, haben Sie Slashdot ermordet, indem Sie zum korporativen Druck beugten und Subskriptionen einlei
what about on their notebooks? I'd love to get a powerbook, but I know the one mouse button would start to irriate me(another problem is the damn touchpad --who likes these things?)
------- Oh damn.... the Sigfile escaped... -Great OM
We do a ton of embedded PPC stuff where I work (and we have an board design that is totally in house). We are using IBM.
Why PPC? Power consumption.
Why IBM? Power consumption.
Whats missing is support from IBMs notoriously slow microelectrionics division. While the uP division is churning out PPC every 6 months, it takes the uE division another year to come up with a PLB/PCI bridge to support it.
You do realize that IBMs LATEST and GREATEST CPC710 (which is only sampling) doesn't support IBM's latest SHIPPING uC.
IBM went looking for an OS afterwards.
-- Improve Windows - Buy a Mac!
mem*, str* would gain from it. both in the kernel and in glibc.
-- Colin
...or it's like saying the Ford Mustang and the Plymouth Voyager are one in the same since they were both Iacocca's babies.
--mike
(Self moderation down to -1 for Off-Topic. Don't waste your points on me)
Uh.. that ain't it (as far as I know).
The version of the story I always heard was that Intel had licensed bubble memory technology (yikes!) and in return IBM got the rights to manufacture the 8086/8088 without paying royalties (or at least substantial royalties). Originally the IBM engineers wanted to use the 68000 series, but the 8086/8088 could be had for next to nothing so they went for it.
spoo
What are the strengths of the PPC? Embedded systems (=cheap/low power). What are the strengths of Linux? Cheap servers.
Why not unite the two strengths? Make super cheap/lower power/tiny foot print racks mounted units with the *embedded* version of the PPC running Linux! You can optimize the server for its purpose while still having the flexibility of Linux.
I don't know of anyone who is running Linux who doesn't keep Windows as a dual boot for running games. That alone will doom the efforts of a "consumer" Linux PPC.
Nobody wants to run production on Intel, and why would I dump HP-UX or Solaris in favor of Linux, voiding my support contract in the process? This article has a lot of vision, but what Linux needs is the warrantied support of a chip manufacturer.
(note: If the author is mainly refering to personal computers, I would question his reference to "the IT Industry" which the last time I checked, included but was not defined by PC sales.)
Or not.
------
If Apple can use its knowledge and influence to help convert annual production of about 25 million WinTel boxes into LinuxPCs, then, together with its own use of roughly 5 million PowerPCs, that 30 million piece annual volume would give the PowerPC a fair chance at gaining further market share against the roughly 95 million box market that WinTel boxes would then have.
Um, yeah, and why would Apple want to give away 25 million PowerPC chips to support an OS that it's trying to compete with? Shareholders and users alike would be infuriated.
Make no mistake, Apple supports (or at least doesn't discourage) Linux on PowerPCs, but they are a business with a goal of dominating the market with Mac OS X just as Microsoft is a business with the goal of dominating the market with Windows and Red Hat, Caldera, SuSE, etc. are businesses with a goal of dominating the market with $DISTRO. Apple would be highly unlikely to help take potential Mac OS X seats away from themselves in such a way.
--
I use Macs for work, Linux for education, and Windows for cardplaying.
You're just brainwashed by Jobs - PowerPC performance "at half the clockspeed" is a myth unless you're dealing with highly optimized and narrowly defined benchmark suites. Why not compare two Linux systems side by side? The G4 Mac you buy for $1500 is easily outclassed by the Athlon T-bird you get for the same money (and the Mac doesn't even include the monitor!). Compilation performance alone shows that the PowerPC is a dog. And as far as possible I'm comparing like systems - both using 512M of PC133 RAM, both using identical 7200 IBM 75GXP drives, same kernel revision, etc. etc., only the clock speeds are different - that Mac has been a big disappointment. People, if you're thinking of a used G4 just for Linux - make sure you get to test it before paying for it. I'm not saying that it's useless, just that there are more cost-effective options.
Well this is halfway off topic (as usual) but what I would like to know is when is someone going to focus on doing something really cool with SGI based hardware. They have some pretty hardcore boxes, and with the way the company is going (or so it seems) is they'll be going going gone soon, which means your likely to pick up some SGI boxes for dirt cheap.
Yes I know SGI makes a Linux based box, but surely someone could create a special port of Nix focusing on maybe some hardcore graphic design packages for that machine... Anyone using an SGI running anything other than (*cough*crap*cough) Irix? And not an SGI Indy... gimme some hardcore O2, Origin freak respond to this with their pimpification of an SGI
Want Root?
Sheesh, Henry. Lay off the 'shrooms and buy a laptop, already.
Not from the article:
Amen.
k.
--
"In spite of everything, I still believe that people
are really good at heart." - Anne Frank
"In spite of everything, I still believe that people are really good at heart." - Anne Frank
The article says, "...In the late seventies, Intel won the chip war with Motorola and Zilog by offering certain features in its 8086 chip that favored MS-DOS over then existing competitive OSes...". What were the features?
Motorola Computer Group does make PPC ATX system boards. I don't think there's a version of Mac OS that runs on them, but you can run Linux on them, though it's not fun to set up.
Mac OS not running on these systems isn't Motorola's fault. Apple doesn't chose to support these boards, other vendors do support them. Apple sells computers, not OSs, so they aren't interested in supporting their OS on other platforms.
The ATX boards are Motorola's MATX series. They also have CPCI and VME boards.
www.mcg.mot.com
OS/2 For The PowerPC was finished and version 1.0 did ship. There was a time when you could buy a PowerPC-based computer from IBM with OS/2 PPC pre-loaded. The proof? You can still buy today an application for OS/2 PPC: The Graham Utilities for OS/2. Scroll to the bottom of that page and you'll see:
--
Lord Nimon
And the men who hold high places must be the ones who start
To mold a new reality... closer to the heart
- j
Let's see a successfully mass-marketed x86 PC with Linux preinstalled (and I define "successfully" as "I can go to Best Buy or Circuit City and pick one up"), before we move ahead with a different system architecture...
1) The PPC isn't 'lost' to BSD. To call it lost, makes it a win, lose game, rather than a win-win. Too bad vast, vocal parts of the Linux world sees things as a win-lose game.
/. story
/. story
/. story As we know, that didn't happen. And, the more time that passes, the more likely its not gonna happen.
In the 'not helping linux on PPC' department:
1) LinuxPPC Co-founder resigns.
2) Some feel the PPC version of the kernel isn't merged with the std X86 centric kernel.
3) back in Aug 1999 a story was floated 'will PPC become the leading linux platform
Linux will be an option for PPC based machines, just like BSD and AIX is.
Just like GNU/Linux doesn't rule the X86 world, its not going to rule the PPC world.
If it was said on slashdot, it MUST be true!
When will this FUD end? The iBook does extremely well against comparably priced notebooks. More features, plenty of new-standard (no legacy crap) ports for expansion, same price points. Desktops do, too, even without looking at Total Cost of Ownership arguments.
I have zero tolerance for zero-tolerance policies.
Constitutionally Correct
You can't buy a Linux box at Best Buy or Circuit City because those places sell nothing but crap. Buying a computer there is like buying shoes at K-Mart. You think that just because you can't get Nikes at K-Mart that they aren't successfully mass-marketed?
I have never been a real apple fan since i started using x86's, but considering the performance you can get out of a chip with half the clockspeed, I think they ARE pretty effecient. Combined with linux, it sounds like a good deal. I still remember running an old shell server off a converted ppc machine. Too bad it had a broken chipset and melted down after two months ;)
I am !amused.
Why? The article starts by talking about the "concern... of letting one company, Intel, supply all the processors and thus control the IT industry" and then goes on to suggest that it's a good thing that this guy's latest and greatest product will only run on PPC. The whole thing reads like a long justification for a personal obsession with PPC, and contains all sorts of inaccuracies to make the point.
Rubbish! Motorola remained a force in the microcomputer market by developing initially the 6502 which competed with Zilog's Z80 processors in the home computer boom of the early 80s, and then releasing the 68000 series which were used in the Apple Macintosh, Commodore Amiga, Atari ST etc.
OK, maybe I'm being picky now, but I thought most Linux fans grew up on Apple ][s, ZX Spectrums, Ataris, Amigas etc. etc. none of which had any kind of Intel chip in them (well, maybe some SSI chips....).
But "AMD" and "Cyrix" could do a very good job of getting China out of the "Tel" clutches too! I don't see why selling them overpriced PPC boxes is going to benefit them. Nope, sorry, you'll have to try harder to convince me that a wholehearted Linux Community 'Push' behind a PPC-based Linux platform is a good idea. Personally, the only way I can see PPC-based Linux boxen becoming common is if a really decent rack-mounted server were produced that consumed less power and had better performance than an Intel equivalent (we already know that Althlons run to hot...)
I could go on, but there seems to be a marketing-style totally loaded statement in every paragraph (reminds me of a recent Microsoft press release..) and I can't be bothered to quote them all... I think that was more than $0.10...
You haven't looked at OS X, have you? It's a SLICK Interface on a UNIX. The system has quirks. It is definitely not a replacement for my Linux workstation yet, but it's DAMNED close. When Office for OS X comes out, those of us running two machines to use Unix on one and regular productivity software on another, are going to be NUTS not to look at OS X.
I got a Cube with OS X 10.0.3 on my desk for evaluation, I love it, I just wish that I had some time to play with it and get it set up.
However, the Mac faithful aren't deciding between Linux and OS X, they are mostly running OS 9 and whining a lot. However, every Unix person that I know that has played with OS X is impressed... It's a sharp OS.
That said, I think that Apple would be smart to ship a stripped down PPC system for hackers. Let people buy a G3 machine w/o monitor. The Cube/G4-Workstation are priced a bit out of a hobbyist range (someone that likes to play with machines, and wants a computer to put new pieces in, not be useful on) and the iMac is no fun to play with and comes with a crappy monitor.
Apple should ship some G3 Workstations... hell, pull the Beige G3 design or B&W G3 designs out of the closet and run the suckers off like there is no tomorrow.
Or sell the motherboards. Let people play around with them.
Alex
Alex
IBM was shipping $5000-$10000 PCs to their mainframe customers. The idea was to create a HOME computer so people could work at home. The goal was to wed these machines to the Mainframe business. IBM laughed at the PC market, and never dreamed of them being cheap toys OR work machines. You would use the mainframe at work, but you could dial-in, etc. from home.
IBM was under anti-trust investigation for bundling their OS with their mainframes, and wanted to avoid antagonizing the DOJ. Apple was scoring big with their machines because of Visicalc. IBM wanted to stop that quickly, and needed to get a machine out the door. The guys in Boca grabbed some "off the shelf" ICs and put a system together.
Intel was in the right place at the right time with a cheap 16-bit machine. IBM wanted to cut costs, so they went with the 8088, which was the 8086 grafted onto a 8-bit bus. Remember, a simple bus is that many "wires," so 8-bit is a cheaper mobo to manufacture. They built a BIOS, and Gates gave them what seemed like a sweetheart deal. IBM thought Gates just wanted to push BASIC sales, and therefore was licensing the OS for nothing. When Compaq reverse engineered the IBM BIOS... well, MS-DOS was born.
Remember the old days: IBM: BASICA (Advanced Basic, included support for disk drives so you weren't stuff with cassette tapes like BASIC), and MS: GW-BASIC.
Similar, a few different quirks, etc.
I think I have my PC-DOS 2.1 disks and MS-DOS 3.3 disks somewhere around my parent's place.
Alex
A friend of mine was in Motorola when this went down. He wasn't working in the chip division, but he was at the company. He told me this at the time after taking a tour of some stuff that they were doing in the Boca Raton facility (I actually ended up working 2 blocks from the old IBM site a while later, but this was after IBM shut it down and it's a useless tangent so I'll shut up).
IBM had rooms filled with PPC Computers, but they all ran NT. IBM REFUSED to ship them (despite the NT port), for two reasons:
1) Embarassment: they couldn't get OS/2 shipping, and it was always REAL SOON
2) Dumb corporate policy: Until about 2 years ago, IBM refused to allow two divisions to compete with themselves, and NT-PPC Machines would compete with x86-OS/2 machines, so no NT-PPC machines.
Remember, there was little PPC/NT support, and it would be running DOS/Windows applications, which OS/2 did.
Apple didn't kill Open PowerPC, IBM's management did.
Ironically, about 3 months after NT-PPC was dropped (largely because IBM, the one pushing PPC, wouldn't sell NT Workstations with it), was when IBM decided that they needed to sell NT Workstations... and they did so with x86 chips.
This was a Management decision that I am certain they don't regret. Remember, Win95 didn't successfully kill the x86/DOS arena, and IBM wouldn't have gotten good application support for NT-PPC.
Alex
Can anyone tell me if the platform specific parts of the linux kernal for PPC been written to take advantage of the Altivec instructions?
Drag n' Drop DVD Recommendations
I'm talking about the Motorola/Everex CHRP board that was fruitless to market because not only did Apple raise the MacOS price, they 'decided' not to support CHRP. You're talking about their earlier PowerPC systems designed as platforms for NT and Motorola's UNIX.
The reason Wondows users didn't buy PowerPC machines is because they were limited-production and very expensive. Motorola's ATX board was to be at a price point comparable to the then available x86 boards.
And even a 15% market share is more than triple the share PowerPCs have on the desktop today!
Steve Jobs was scared. There were warehouses full of unsold Mac 6500s, and the FUD about Apple's obsolete system software was cutting into the overall growth of the Mac market. Once MicroS**t invested in Apple, and MacOS8 hit the streets, the FUD stopped (even though the OS hadn't changed that much). But the ability to commoditize PowerPC-based computers was forever lost when Jobs killed the clones, and thus the Motorola ATX CHRP board... didn't I already say that?
blessings,
"Only in their dreams can men truly be free 'twas always thus, and always thus will be."
--Tom Schulman
A brand new Apple doesn't count as a reasonable Linux platform, their hardware is the antithesis of an open system, and their prices are extremely absurd. IBM's attempt last year to make an 'open system' complete with a free schematic failed.
Despite the cheerful '2.0 ghz' (future, always sometime in the future) press release from IBM, the PowerPC lags considerably in cost and speed. Let's not try and get teary-eyed with unfounded hope of a cheap, high-performance, open PowerPC platform. It has found its home in the embedded market.
blessings,
"Only in their dreams can men truly be free 'twas always thus, and always thus will be."
--Tom Schulman
The Mac arena is one place where BSD is making a definite win over Linux, with Mac OS X promising to add a couple of million BSD sites this year.
This is not a signature.
... QNX RtP, MorhpOS and AmigaOS 4.x (ETA the end of this summer). All runs PPC as well.
These links might be interesting:
Pegasos PPC-board
Eyetechs AmigaOne PPC-boards
What really would make my day would be to see FreeBSD running on PPC.
Bjarne
Boy, that would have been a kick if IBM went with the 68000 and Apple picked the 8088 back when the choice presented itself.... We would all be laughing at the Apple users with thier 1.8Ghz cpus that couldn't keep up with our 500Mhz machines, instead of the other way around...
Something to think about
Come on Abit, VIA, Soyo, etc.. How about a cheap PPC motherboard.
I'm sick of buying a huge heat sink with a big fan & 3 case fans to keep my crappy x86 from melting. Give me a reasonably priced motherboard for a RISC processor.
What could be done is to just take one of the BSDs, shave it down to a microkernel (wait, isn't this how Avie and co. created Mach way back when?), and slap a Win32 interface on it. Use the Qt framebuffer, and you're good to go...
/Brian
Depends on whose definition of PC you use. Are you talking about a personal computer (the current Jobsian definition, which is not too uncommon and includes pretty much everything desktoppish and related/developed from), or a PC, which is indeed x86-powered.
The PC in PowerPC indicated that it was a POWER chip designed for personal computing platforms like the Mac.
/Brian
Maybe the big problem is a lack of snazzy, clear cases on the open market - IBM might not want to go public with the news that the G4 won't work unless its in a spiffy clear case or a cruise missile.
--
In spite of the suggestions and all the tests that I have made, I have not cavato a spider from the hole.
This is very true. The last couple of games like e.g. Black and White show a trnd to move away from card-targetted games like quake to more intelligent world games like B&W. Oni was going to be just like that but apparently they never got further than decent character animation, probably because of the strategic buy of Bungie by MS(read: xbox). In any case, B&W is setting a new waypoint for gaming, and you better dig up your athlon 1.33 or intel 1.5 because they are going to smoke out loud.
With great power comes great electricity bills.
Umm, What did intel add to the 8086 that made ibm take it over the z80? I thought big blue went for the 8086 because they didn't want the new pc to compete with their existing line of computers. If I recall Bob Cringley's book Accidental Empires, he said Bill G insisted that IBM go with a 16-bit cpu. So big blue picked the lamest 16-bit cpu they could find. I guess they should have used the m68k like apple did; then we would all have ppc's in our linux boxes.
Don't open that link if you wanna keep it G rated
You sir, need a 3D accelerated graphics card. Now I do agree that games use a lot of CPU power but the most strain is put on the graphics card. I have a Pentium Pro 200 with a Matrox Mystique (for 2D) and a VooDoo2 card. This system is very very outdated for gaming standards.
I especially like strategy games, and there CPU power is nearly not used. It plays strategy games without a problem. Well, lots of RAM seems to enhance performance.
While First Person Shooters are just *unplayable* when NOT using the VooDoo2 card, I can assure you that they work very smooth and fast at 800x600 resolution.
Note: I talk games like Quake, HalfLife, Unreal and the like. I didn't touch new games made in the last year (?) because of lack of time, but as far as I heard they are still based on the same engines, so probably they will run.
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
Ehm, as far as I know, SCSI was the standard hard-disk-interface on all pre-fancy-colours models. At that moment the started to use IDE disks. I know it's a cost decision (IDE is cheaper), but then I love SCSI and I hate IDE (and am willing to pay for that love)
There is by the way no problem to get a SCSI card for your G3 (or G4), I saw that numerous times.
But then IANAMU (I am not a Mac user)...
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
It was a joke to get running compared to something like Mandrake or Redhat.
With the price of x86 hardware, I can't see why anyone would bother... that is of course, with the exception of some kinds of embedded systems where PowerPC might make sense.
The x86 is the commodity chip, and therefore will be the cost effective solution for general applications. For operating systems like Linux that wants to be the commodity operating system, it makes sense that the x86 is the chip of choice. That does not imply, however, that it makes any sense to compare other processors using such a simplistic method as clock speed.
ISA for the personal computing mass market.
:)
x86 has shown one thing, having a poorly designed ISA is only a small bump in the road towards performance. Unless you are an assembly hacker you shouldnt care about the ISA, it just makes sense to concentrate the market onto one ISA... we would not have the present price performance ratio's without it.
Its a pity it had to be x86 but its a done deal now, for very highly parallel SMT machines small register sets might actually be an advantage in the future
You can keep your One World Fragmented Market, I rather have the One World One ISA one where effective competition is not only possible but present and working.
What is your point? These processors do what they are designed to do... x86 arent what they are because of the ISA, the underlying architecture with any wide superscalar machine has bloated so much that a little extra ISA translation on top wont do jack diddly.
PPC is fine for embedded, but dont kid yourself it aint the be all and end all of low power... its only a little better than x86 compared to whats out there. Id never imagine using it for portable applications, theres just way too many far better alternatives.
I wouldnt mind seeing some good competition for win32 OS's (Linux+Wine aint it).
It's more than just a simple hardware availability issue at this point. What software will you run on your non-Apple PPC? Sure, you could run linux apps that you cross-compile (LinuxPPC is always behind the x86 linux, so this will mean in software as well). If you like games, oh well, yer skrewd. No games and the games you already bought are useless.
Nobody makes games for PPC systems, fer Cthulhu's sake. Yea, some will get all whiney and scream how wrong I am...except there are sh*tloads of games for the x86 (and doze, of course) and a mere pittance for the Mac on PPC. None at all right now for MacOS X.
You go PPC, you are choosing to go into a software wilderness inhabited by a very few scattered far and wide.
In Bushworld, they struggle to keep church and state separate in Iraq as they increasingly merge the two in America.
So you prefer to waste CPU clock cycles on sound rather than doing the proper way with a dedicated sound processors? Does this mean that you also consider the proper way to do modems is to have the CPU handle that too?
The proper way to do this is with dedicate DSPs, leaving the CPU to do important and better things. Leave the graphics heavy lifting to the graphics chip, the sound processing to the sound processor, and the modem function should be handled by a dedicated DSP too, not the CPU.
In Bushworld, they struggle to keep church and state separate in Iraq as they increasingly merge the two in America.
The whole argument about having to upgrade/alter your motherboard is bogus. If software vendors dedicated as much work to making PPC apps as they do to making x86 apps, then PPC systems would have to be upgraded every coupble years too.
Because software is targeted primarily at x86 - because it is virtually the entire desktop/workstation market all around the world - and because the software competition is so hot, they push the envelope. Game developers are particularly into this. You give them X amount of processing power and they will consume it as fast as they can. Then they will go on to surpass it, anticipating an improved chip next year, etc. You want the latest and greatest software? Then every couple years you need to update your processing power to match it.
This isn't an inherent weakness of the x86 design, it is reality regardless of the design. If the roles were reversed, then we would all be upgrading our PPC cpus every few years so we could keep framerates up and make use of the latest productivity suite version.
PPCs are NOT orders of magnitude faster than their x86 rivals. They are on par but they lack the focus of the VAST majority of software developers.
For embedded systems, there is no reason to upgrade CPUs or motherboards. You can use the same, slow, simple CPU next year as you did last year, and the year before. A microwave oven is NOT a clock-cycle hog. A car is NOT a clock-cycle hog.
In Bushworld, they struggle to keep church and state separate in Iraq as they increasingly merge the two in America.
It most likely would be Openfirmware.
But you forget that there are lots of different purchasers of PCs and the cost/benefits are not the same for all of them.
I am typing this on a Dell computer, not because it was cheap or because I like Dell, but because my employer likes to buy computers from Dell. Being a well-established business with a reputation for good support is worth a lot of money.
If IBM is marketing a PPC-based linux PC, $1k either way won't make a difference compared to the reputation and connections of IBM when a business is looking to buy PCs.
Hehe, yup.
:)
At work, we use Intel 80188 processors on our development boards as host processors to our DSP *because* it is so godawful slow and has a very limited pin-count. In effect, by making things work with a 80188, no matter what cheap processor our customers decide to use, the worst they'll have to is insert some delay cycles
Tim
I would certainly hope they put a better BIOS than what is on the current intel boards. A Sun style one maybe?
Most Linux fans are not old enough to remember that Intel did not always dominate the PC processor market. In the late seventies, Intel won the chip war with Motorola and Zilog by offering certain features in its 8086 chip that favored MS-DOS over then existing competitive OSes. Subsequently IBM selected the 8088 for the first PC, knocking both Motorola and Zilog out of the emerging PC market.
I thought IBM chose MS-DOS because they refused to deal with the CP/M, and had Bill Gates & CO steal^H^H^H^H^H write their own DOS for IBM.
Running Linux on Power PC will no more make it survive than did MS-DOS on Intel. It is a matter of power vs. price, although more for the price. That's why AMD is kicking Intel with steel-toed boots in the speed category.
Yeah, right.
PPC chips are not cheap, compared with x86 alternatives. Likewise, as x86 is the default development target, PPC will always be a step behind its i486 cousins.
Not that it's a bad chip, mind you. Just that x86 seems to be the main focus of things right now.
Dancin Santa
The unfortunate matter is that Nvidia's OpenGL drivers aren't GPL, however it would help the practicallity of a Linux/PPC Desktop if Nvidia would simply compile their drivers for PPC (this wouldn't take very long at all). This would also allow new Macs with LinuxPPC to utilize their Nvidia GPU.
Anyone else remember back in '93 when the 601 was released? They did have some momentum. But IBM killed it. They said 0S2 would be out for it any day now. Well without a decent workstation OS nobody wanted it. IBM programmers couldn't get 0S2 out the door and it lost. Billions and Billions have been lost to intel because those guys at IBM couldn't finish OS2.
Just to correct a bit things; there are actually some games for Linux out there with ppc binaries included. Take a look at Loki Entertainment, they are distributing PPC binaries with at least some of the games they publish - e.g. CivCTP, Heroes III, RR Tycoon, Myth II, to name a few.
Not to mention the multitude of the already-existing games for MacOS, though there are always some titles I am missing... But if I really wanted to play, I'd get a PS2 instead of a frigging PC: at least the PS2 doesn't freeze up or something, in the middle of a game.
still running a x86? dinosaurs do exist!
One thing that is not mentioned in the article is pipeline length. Intel ramps speed by extending the pipeling- P4 is at 21 (or is it 28, I forget) while the G4 PPC is at 7. People worried about this because the G3 has a 3 or 4 step pipeline. Shorter pipelines emphasize more work done per clock step, while long pipelines do less per clock tick and try to do a lot of pre-emptive execution, which can lead to total execution failure (side effect that makes 1.4GHz P4 slower at Word than 1GHz P3). A short pipeline reduces pre-emptive failure, plus can achieve a 1:2 up to a 1:4 ration for work/clock tick. This means that a 500MHz PPC can do the same as a 1GHz x86, or up to a 2GHz x86 depending on the task. Of course this is theoretical, but I hope you see my point about the relative merits of pipeline depth.
Vote monkeys into Congress. They are cheaper and more trustworthy.
The only real PowerPC platform is Apple. Other than that, companies have little to gain from trying to produce their own PowerPC boards.
Then there is the issue of actually trying to get PowerPC chips from IBM or especially Motorola. Apple has the lock on all they can produce. There is often shortages for Apple.
Linux + PowerPC is a nice idea, but it will never become popular. The only reason why I would ever want to run a PowerPC would be for MacOS X anyway. My Athlon runs Linux just fine.
I think that many posters fail to understand that we're talking a looooong time in the future. The primary reason to make the switch then is that the x86 intel clones are all CISC proccesors, which is an archaic and limited architecture, destined to grow bigger and more power-hungry over time. Apple made the switch to RISC with the original PPC 601 chip, which was a great idea. What started off with a compromise on lower chips (Wozniak spent the money without Jobs' permission, and they went with the then-slower motorola 68000s) has led to an early adoption of a superior technology. AMD and Intel don't have any RISC chips in the wings, and Motorola and IBM are producing speed upgrades all the time. In the future, their experience and knowlege of RISC design will let them gain greater market share, and Linux users, as early adopters, can no doubt appreciate the gains that come with experience.
P.S. Contrary to some earlier posts, it IS possible to buy apple motherboards, and PowerPC daughtercards. I built a frankenmac that runs 4 604e proccessors, out of an old Daystar motherboard and some daughtercards that cost me $50 apiece. I call it the Beast, and it rocks my world. I'd love to drop Linux onto it, but that couldn't utilize the multiple processors. Oh well!
My Karma is so good, I'm the Dalai Lama...or something.
The PowerPC chip is a niche player. That's not necessarily a bad thing, as long as the various ecologies in which it's dominant continue to be viable.
But - promoting diversity for diversity's sake is just plain silly. Diversity is a strategic choice, not a tactical choice, and it's very difficult to push strategic choices down the throats of users, especially when cost is involved. It's going to be practically impossible to get PPC stuff down to the commodity pricing level of the x86 world, and that's fine, because they're for different markets. Is there a value proposition for PowerPCs at the current price points? Yes! Does that value proposition make sense at the commodity level? Obviously the answer is no, because, well, x86 is good enough.
When it comes to non-standard computing environments, however, PowerPC chips are much better. Low power, low heat, good performance with few compromises, all combine into an attractive and compelling package. In restricted environments, the heat dissipation characteristics alone are compelling.
So remember, it's just a chip. There are more important things to get worked up about, like licensing schemes and such.
Are many of us here on Dope here including IBM???? PPC is way ahead of IA32 and SuSE Linux has had this port for some time....simple for all you RedHat idiots....just load the SuSE Linux PPC and cahnge: /ETC/ISSUE = "Welcome to Redhat"
Instead of Welcome to SuSE Linux
damn, what is it that we are turing into "follow like good little cows" reminds me of the uprising of NT....doesn't anyone care about engineering of linux at all anymore?
Regards,
Jon
As one who was writing S/360 assembler code 10 years before the Z80, 8085 et al., I must tell you that you're wrong about why MS/DOS (and later, WinTel) succeeded: they won because they made something the managers approving purchases could be confident would be around for a while. 30 years later, I insist on W2K on every new machine not because it's superior, but because if something doesn't work I don't need any gurus to fix it, just an MCSE or two. Think of Fedex and their trucks: a geek would buy BMWs or even design a whole new "Fedex Delivery Unit". A responsible (i.e. successful) manager would buy Chevrolet - and if one broke down, scrap it a buy another.
No. Well...maybe. Actually, yes. It really just depends.