Slashdot Mirror
'Black Lab' Linux For G3 Clusters
ChristianC writes "'Black Lab' Linux, a relative of Yellow Dog Linux, has been released for PowerPCs (including G3s and iMacs). The distribution offers Beowulf and Cluster computing - 20 G3s at once!
"
← Back to Stories (view on slashdot.org)
Hmm. They might be terrible running just the MacOS, but play Q3T or throw Linux on there, and you will fear their performance.
Doing mathematical equations (generating SSH keys) my G3 is twice as fast as my PII, both at 400 MHz (granted, the G3 has more cache).
Oh, I agree completely -- it's a great architecture. But I'm a businessperson, not a research scientist, and Apple boxes don't give me enough real-world advantages to compensate for their disadvatages. I'm open to persuasion, but your arguments have to show me how I'll (a) save time, (b) save money, and/or (c) get a better product from an end-user standpoint.
Not everyone's a business person. Not everyone works from the premise of "getting" a new computer to run stuff on (see 68k NetBSD, OpenBSD, and Linux projects if you don't believe me). Computers are tools. Not everyone has the same criteria; it varies on their needs. This is obvious.
If you went with cost, you might have gone with an AMD or Celeron, not even a PII or PIII. But there are people who buy PIIs and PIIIs; they aren't stupid just because AMDs and Celerons are cheaper. They might have different criteria, e.g. they needed the little extra speed. There are some people that might want to build Linux servers cheaply.
But there are Linux ports beyond the x86. Why? People had the hardware. If I have an Alpha, need a web server, I might try running Linux on it. It would be stupid to purchase a new machine unless the Alpha was insufficient.
There are Mac users who desire or have reason using MacOS but also have use for running Linux. With PPC Linux, they can use both for the price of one box, not two. What, run MacOS? Plenty of people like it. Plenty of environments exist where it's simply easier to get a MacOS computer over intel.
Thus, likewise, there are people that already have two or three G3s and 604s. Why should they have to go out and buy equipment? Utilize what they have. To research scientists, who you admitted you are not, who might need a clustering option, Black Labs allows them to not spend $10,000-$20,000 on getting intel equipment. $10,000 not spent means more money for regeants or that back-burner experiment that they might not have had funds for. And since a lot of research money is NIH aka taxpayer's money, I am in general a more happy camper that funds are being used efficiently.
i quote from the beowulf mailing list faq v2 at: http://www-unix.mcs.anl.gov/~utha yopa/bwfaq2.html
2. Where can I get the Beowulf software? [1999-05-13]
There isn't a software package called "Beowulf". There are, however, several pieces of software many people have found useful for building Beowulfs. None of them are essential. They include MPICH, LAM, PVM, the Linux kernel, the channel-bonding patch to the Linux kernel (which lets you 'bond' multiple Ethernet interfaces into a faster 'virtual' Ethernet interface) and the global pid space patch for the Linux kernel (which, as I understand it, lets you see all the processes on your Beowulf with ps, and maybe kill etc. them), DIPC (which lets you use sysv shared memory and semaphores and message queues transparently across a cluster).
As for the licensing info, presumably the patches to the Kernel will be GPL'ed. Check out: http://www.beowulf.org/software/softw are.html. I couldn't find any specific information on liscensing, although they do refer to the software necessary to implement Beowulf as:
implemented as an add-on to commercially available, royalty-free base Linux distributions
hope this helps
alex
Well, I've installed and experimented with MkLinux, PHTs TurboLinux, and LinuxPPC (still waiting on R5). I've also installed and used BeOS. Eventually, I've always gone back to MacOS just because the apps I really need (or want) are there.
Why do Mac people also use Linux?
1) As stated before, if you held out against the Wintel duopoly during the recent dark years, you're not a "just go along" type of person. (Yeah, the whole "Think Different" thing is pretentious, but there's a grain of truth to it. It's a stubborn streak that suits both Mac and Linux users well).
2) It's easy. MacOS coexists peacefully with another OS on the same hard drive. Most of the alternate O/Ss provide easy-to-use dual booters, and choosing between different startup drives has always been trivial on the Mac. You can boot from just about anything on a Mac: external HD, floppy, Zip, Jaz, whatever. So it's easy to experiment with little obligation.
3) This one is subtle, but I think the most important. Both Macs and Linux reinforce the idea that computers are *logically designed and predictable systems that can be mastered*. With Linux there is predictability and mastery, but it requires much greater initial knowledge (i.e., a steep learning curve). You've got to want to do something, then learn how to do it.
With the Mac there is a shallow learning curve, but you can keep going as far as you like. If you're Grandma and you just want to e-mail and file recipes, you can stop there. But I've found the Mac actually encourages "serendipitous" learning -- sometimes I've ended up doing new things just because it was so damn easy and predictable on the Mac. I got a cable modem, then before I knew it I was using my machine as a Web server and router, with three or four other Macs on a LAN with it. And it was *easy*. I've seen novices (like I was) go from using software to installing it, from attaching peripherals to swapping components, from using higher end apps to coding. Even when Macs crash, it can be *logically* narrowed down to an extensions conflict or offending app.
Compare this to the Wintel side of the world, where even experienced technicians just reinstall Windows to solve a problem. Whenever I use Windows I feel like the system is fighting me every step of the way. Sometimes things just don't work. Sometimes it just crashes. No one is ever able to explain it. It's all just freakin' voodoo. Although its not as bad as it used to be, how would anyone ever learn cool stuff by chance when the computer is this inscrutable device that doesn't do simple stuff right? If the computer is not logical and predictable, why even expend effort trying to understand it?
Even if you do make the effort, it's just one kludge after another. Right down to the processor and its assembly language. No offense to the Intel users out there, but x86 assembly language is just a mess. Let's not even discuss PowerPC, because I don't want to make this a Mac-vs-PC thing. Look at MIPS assembly language. Instructions logically grouped, all the same length, with consistent syntax. It's understandable. It's *knowable*. When you look at it you see logic behind the design.
But I'm getting a bit off topic here. Basically, I think the shallow learning curve of the Mac and its consistent design encourages users to learn more. It tells them that computers (and technology in general) are rationally designed things you can understand and master. Do that for long enough and you will build up the necessary knowledge and confidence to tackle Linux.
my bet is: the G3s sux big time for this.
Whatever--to both of you.
G3s are comparable to a PII system in FPU and better in integer. I wouldn't call that "sux big time" and I sure wouldn't say that it kicks butt.
If you want FPU, you go Alpha. You want bang for buck and you're starting from scratch, Celerons or even AMDs might do you well. However, all of these assumptions go from a premise of the need solely for higher performance (using alphas) and/or starting from scratch with a budget (using Celerons).
The idea of beowulf is to build a specialized high performance cluster (with high MFLOPs, but slow interconnectivity of the nodes) from cheap off the shelf components
Eh? While true, that's only part of the picture. "Build" doesn't equate to " newpurchase."
There are, lo and behold, groups that have a bunch of G3s that were purchased for other reasons. Like who? University's who have public computing labs. Science/research labs, in general, like Macs. Some of these people might have an interest or developed a need for clustering and doing so on their current investment of computers is beats the pants off new purchases, aka makes it almost free.
Black Labs brings these people the simple option of using their current investment and/or money already pointed for Macs for other reasons (e.g. some chemistry software is preferentially run on macs). If you've looked at other clustering options for G3s (you did before you shot your mouths off, right??), you'd realize the AppleSeed project had some of these exact reasons in mind. And even here, there are pluses and minuses between using AppleSeed and Black Labs.
We have around 300 G3s and 4 iMacs at our school(the government gave the school discounts). We've had macs since the opening of the school years ago. Anyway, the performance of G3s and iMacs is terrible. Now, I may not be a big mac fan to begin with, but the performance of those computers is lacking. I think linux would run well on them, but it's amazing how slow stuff will run, when they claim there's a 300+ Mhz processor in the boxes. It's good to see that different versions of linux are being ported for macs, maybe apple will open it's eyes and adopt the open source theory. Wait, no they wont. :)
droool...
-Isaac
I am not a lawyer, and this is not legal advice. For Entertainment Purposes Only.
The iMac runs the G3, which is the decendent of the 603. If I remember correctly, the 603 chip was designed as a "consumer level" chip, and therefore isn't designed to do anything much beyond single-processor stuff.
:-)
I didn't bother reading the report (Postscript..no printer happy...too lazy to use a reader), but I would suspect that the 604 would yield much better results. The 604 was made to be a more versitile "industrial" strength processor and therefore includes many features that are absent on the 603. The G4, which is supposed to ship before too long, is a decendent of the 604, and likewise may yield better results than the G3.
Then again, I could be on crack and be completely wrong....since I didn't read the report. So if I am wrong, please correct me kindly
An Australian research team clustered a bunch of iMacs, and the results were sad. Intel and Alpha boxes trounced the iMacs. The iMacs weren't designed for that sort of application, either, but it points out some major problems with using PPC (currently) for that sort of application -- at least on the iMac.
The article is at:
http://www.dhpc.adelaide.e du.au/reports/065/abs-065.html.
-- haaz.
Old, but interesting. Check it out.
I like the yellowdog people a lot, but yellowdog 1.0 was released so far before it was ready, and their support (on-line at least) is horrible. AFAICT, it's one brave guy (many, many kudos to you, Dan!) who has way too many hats to wear. Yellowdog 1.1 looks like it'll be a really good product, but if blacklab is anything like that, well... wait. Wait quite some time. I'm sure you'll see good product from them eventually.
~luge
IAAL,BIANLY
(a) save time, (b) save money, and/or
(c) get a better product from an end-user standpoint.
One way IMHO is hardware maintainability. Power Macs (and most 68K Macs) are smart enough to select a bootable hard drive, boot directly from CD when requested, use default boot settings (just in case your newly-compiled Linux kernel isn't up to the task), etc. Compared to Intel machines, Macs are a breeze to maintain. So much so, in fact, that if it weren't for the lag in porting the latest versions of Linux S/W (apps as well as drivers, mind you), I'd say there's no reason to buy an Intel box.
Side note to haaz: Whither LinuxPPC 5.0?
I've doctored Intel machines running under WinXX, Win3.1, OS/2, WinNT, and dealing with the hardware was always a pain. Maybe that's changed somewhat, but the Mac was always easier to futz with.
-----
".sig,
I'll apologize out front, but I really take offense with some of your comments. To say that it is not worth making linux for a mac is entirely out of line. Your other comments displays a level of ignorance or bias that is absolutely pathetic. I'm almost speechless.
For one thing, there are some extremely talented ppl porting linux to the PPC. Consider this point. These ppl, who are a small minority of linux developers, have somehow managed to get linux working on the PPC architecture. Are their efforts are a waste of time? Why did they choose to develop for the PPC?
Perhaps you only want an Intel version of linux? Or maybe only an Intel, Sparc, or Alpha port of linux? The great thing about linux is that it is cross-platform! You are obviously too young to remember when Unix was a totally splintered OS. Linux represents a way to reunify Unix, which is a really important point. This requires cross-platform support otherwise you get HP-UX, Apollo-Unix, AIX, AUX, SCO, Ultrix, SunOS, ad nauseum.
As for a MacOS "emulator": Have you heard of SheepShaver or Mac-On-Linux? Not entirely there yet, but they are real close!
I have been a Mac user for over 10 yrs so I must be a retard. My first experience with Unix was with an IBM RT (try to figure out what was this POS). This at times was a tough machine to work with, but it did have a C compiler and introduced me to a better computing environment. I still like this computing environment and still use a Mac.
"The platform has grown up, partly because Apple has close control over what goes on."
Oh, I agree completely -- it's a great architecture. But I'm a businessperson, not a research scientist, and Apple boxes don't give me enough real-world advantages to compensate for their disadvatages. I'm open to persuasion, but your arguments have to show me how I'll (a) save time, (b) save money, and/or (c) get a better product from an end-user standpoint.
--Tom
Tom Geller
I won't bore you with the details, 'cause I'm sure you know them. Price, component availability, a community of support, yadda yadda yadda. Although I really wanted to use a PPC-based machine, I had to ask myself: What advantages would I gain?
Speed? Processor speed is virtually irrelevant in Internet serving, as bandwidth and disk access are the gating factors. Striking a blow against an evil company? Apple's not so clean, and certainly can't claim better corporate morality than AMD or the screwdriver shop where I bought my Linux box. (Needless to say, there will be no Microsoft code on it. :) )
Having said that, I'm glad to see Yellow Dog continuing with its plans. Until now, there have been three PPC Linux vendors that I know of: TurboLinux, mkLinux, and LinuxPPC Inc. (not the same as LinuxPPC.org). Frankly, none of them have approached the market with the resources, experience or commitment needed to make an impact. mkLinux is the side project of a book publisher; TurboLinux does PPC support as an afterthought (and devotes $0 to PPC Linux promotion AFAIK); and LinuxPPC Inc. has problems best not discussed in public. Yellow Dog seems to be making the first real stab at the market: I hope they can eventually convince people like me that running Linux on PPC is the best deal.
-- Tom Geller
Tom Geller
Apparently you've missed the relatively exciting things apple has done the the Mac architecture lately. The platform has grown up, partly because Apple has close control over what goes on.
Besides, what's the point of developing a unix for 286's etc. (minix)...for one because it can be done, but also because the more platforms an OS supports the more useful it can be -- even for that doorstop 8086 lying around (let's see you boot M$ Windows 98 on that.
I'll quickly admit,though, that i'm not a Mac advocate; i don't use one.
Subculture wrote:
he's right. and mac linux is not as good as linux for an x86 or a computer with alpha archs. IMHO I don't think it is even worth making linux for a mac,...
I don't see how it matters whether it's a better platform or not -- part of linux's appeal is that you don't have to choose which architecture to worry about, simply the OS.
What i love about the GNU/Linux phenomenon is that it favors giving the user control (over hiding everything behind set rules like M$ does). That control should extend to what hardware they choose as well. Not everyone is looking for the biggest, best, or fastest computer.
There are several wine-ish products out there; i haven't used any of them so i have no idea of the quality, but they do exist.
M$ was wise not to ignore the Macintosh as a platform and i think we would be just as smart to concentrate on porting the wonder of linux to every system we can...because when it comes down to it, what's the loss if only 40 people end up using it? Those are 40 people using a better OS. Obviously the people porting it feel that it's worth it, if only for their personal experience.
Also, linux is not exclusively about business. If it was it probably wouldn't be where it is now. It's an OS and you do with it what you choose, on what hardware you choose (ideally at least).
Hi guys,
Maybe someone can explain this for me. I'm not sure if it is because of the differences between RISC vs. CISC processors, or if the PowerPC is better at floating point operations, or just what.
I'm running the SETI@home screensaver on two machines. The first is a Macintosh PowerBook (several years old) 3400c, using the PowerPC 603e at 180MHz and running Mac OS 8.6. The second is a new desktop AMD K6-2 350MHz machine running Win98.
After completing a few data units, I noticed that the PowerBook is completing data units about TWICE as fast as the AMD. I can't figure out why this is!
--globalnap.net, product of pure caffeine--
(Background: I've been through a number of Macintoshes and I'm on my fourth PC right now)
... the fact that they use not just a diff. manu but a different architecture is good for everyone. Here's a whopper of an analogy: what if paper was somehow constructed so it could only be written on with one sort of pen? That would seem silly, since there are lots of other choices. But software tend to be written to OSes themselves entirely dependent on their host hardware. Linux helps buck this trend (not as much as the BSDs, maybe, but still) by providing an OS which runs on multiple platforms.
...
Like the other respondents to this point of view, I don't have an irate reply, but would agree with those other respondents as well that the total universe of Mac users is pretty broad.
Why would anyone want to run Linux / other Free OS on a Mac? That's not exactly the question here, but it seems to be lurking beneath the surface, and that same question was asked in a thread one level up from here.
I can't answer this for everyone, but here are the reasons I think that Linux and the Mac make a great combination, practically and normatively.
- Macs tend to have nice human engineering. The gulf is not as wide as it was 10 years or 5 years or maybe 2 years ago compared to the PC world, but well-labled parts, legible icons, attempts at friendlization still IMHO work better on Macs, but this comparison of course ignores that PC vendors vary tremendously in this respect.
- Macs use non-Intel chips
- They look cool. That factor sells a lot of PCs, however unsatisfying that fact may be.
- Ubiquity. Not as many macs in the world as PCs, but the lopsided numbers / market share I think are misleading, since it's hard to go ten minutes in any US city and avoid seeing either an ad or an actual iMac / G3, not to mention older and still humming Macs.
- The Mac OS begs for a replacement, or at least whines a little. I use one at work (an iMac) and am actually fairly happy with it (Most of my complaints I list on my web page and will skip here) but it crashes all the time! I would love to be running an underlying *nix, whether it still looked like the basically-well-conceived Mac OS or like my home linux machine.
I don't understand why Apple pulled the plug on clones -- how about because they were making Apple look slow by releasing faster, cheaper machines? -- but I wish they hadn't. Then we could perhaps be running on a G3 PowerComputing box with SCSI, firewire and USB by now
anyhow. I don't have a home mac, but if I find a cheap one I'd like to run some free OS on it.
timothy
jrnl: http://tinyurl.com/c2l8yr / foes: http://tinyurl.com/ckjno5