Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Yellow Dog Linux 2.0 Review 74

Posted by michael from the see-spot-run dept.
Patrick Mullen writes: "The Duke of URL has just posted its review of Yellow Dog Linux 2.0 - Linux for the PowerPC. The review covers installation, the interface, YUP (their own apt-like update/install tool), benchmarks of PPC VS. X86 and much more."
This discussion has been archived. No new comments can be posted.

Yellow Dog Linux 2.0 Review More

Yellow Dog Linux 2.0 Review

Comments Filter:

  • Yellow dog pissing me off (Score:1)

    by Anonymous Coward
    I'm in the middle of my 5th try at the yellowdog install at work, and burning time on /. as the stupid yellow progress bars fly across the screen next to me (again). GCC and network setup scripts (python) broken in the default install! This would proabably be a good distro if I could use those two things! I also dont like that you can't select individual packages in the install. Argh, oh well, back to work...
  • everything is slower

    try big file copies, try compiles, make a simple program in C, it will load slower and be slower whenever it is in syscalls. everything will be slower. os x uses mach, so the base system has one area of overhead that monolithic kernels don't. the operating system is simply slower.

    there are other reasons apps are slower than just the graphics system. the bloated system libraries for instance, but even things that don't use them have mach to help them lose their edge.

    ps. i'm using yellowdog 2.0 and tri-booting the mac oses, so this isn't entirely out of my ass. however, plenty of it is, the only benchmarks i have done were compiles.

  • Re:Looks good :-) (Score:4)

    by pudge ( 3605 ) <slashdotNO@SPAMpudge.net> on Wednesday July 18, 2001 @04:35AM (#77911) Homepage Journal
    A few problems with your reasoning:
    • Macs that can run YDL sell used for around $500 or less (I had Slash running on YDL 1.x on an iMac rev. B (G3/233, 128MB RAM)).
    • Many people have old usable Macs lying around collecting dust.
    • Many Mac users thinks Mac OS X (currently) sucks, and it is not an option. Besides, you get to use real Mac OS under YDL, while under Mac OS X you have to run your apps in Classic which, while it has obvious advantages, also has obvious disadvantages.
    As to speed: Mac OS X is slower than LinuxPPC. Period. I can't give all the reasons why. But everything I do on Mac OS X is slower than on Linux (I have my PowerBook with all three OSes: YDL 2.0, Mac OS X 10.0.4, and Mac OS 9.1, tri-booting with yaboot). I compiled perl 5.7.2 the other day on each, and `sh Configure -des -Dusedevel` took about five times as long just to get started, and took about 2-3 times as long to make.

    Is it HFS+? Is the running UI slowing things down even though this is running in a console? Are the compiler and shell utilities not compiled well? Is it all of these? I dunno. It is just slower. Everything is slower.

    I won't even bother with why I don't like Mac OS X's UI (NeXTisms) or its Unix idiosyncracies (NeXTisms).

    What I will say is that YDL 2.0 has a few glitches, yaboot was a pain to set up for some reason, but now that it is running it works well.

    Of course, I still spend 95 percent of the time in Mac OS 9.1. :-)

  • No Upgrade Path (Score:4)

    by waldoj ( 8229 ) <waldo@@@jaquith...org> on Wednesday July 18, 2001 @07:31AM (#77912) Homepage Journal
    What the review doesn't say is that there's no way to upgrade to YDL2.0. (At least, there wasn't a month ago, and I haven't seen anything on the mailing list or the website to indicate that's changed yet.) I bought the YDL2.0 CDs the moment that they were available (maybe 6 weeks ago), hoping to upgrade my YDL1.2 machine. No such luck. Frustated, I bought up the topic on the mailing list, but it devolved into a bit of a flame war, unfortunately. I was told that if I had any sense, I'd wipe my machine and re-install from scratch, that there's no reason I couldn't wait for an upgrade path, etc.

    Anyhow, if you're a user of an earlier version of Yellow Dog Linux, do yourself a favor and hold off. What would lead Terrasoft to release a 2.0 final release that lacks the ability to upgrade from previous versions is beyond me. But don't make the same mistake that I did.

    -Waldo

  • One little mistake (Score:5)

    by Erik Hensema ( 12898 ) on Wednesday July 18, 2001 @04:16AM (#77913) Homepage

    I decided on doing a make vmlinux because Yellow Dog didn't come with anything to make bzImages out of the box.

    (b)zImages are x86 specific, AFAIK none of the other platforms Linux supports has zImage. A (b)zImage is needed in x86 because of the memory modell (only 1M adressable of which 640 KB usable in real mode) and the weird boot/partition scheme (come on! A 512 bytes bootsector and partition table in one?)

  • what is he talking about? (Score:4)

    by option8 ( 16509 ) on Wednesday July 18, 2001 @04:52AM (#77914) Homepage
    "...it starts to make Linux look more and more attractive with its arguably more polished interfaces.

    With Aqua down and not getting up for a while (until Apple can revive it),"...

    these two sentences from the introduction make absolutely no sense to me. is he saying that linux has a more polished interface than OS X? what, X? KDE? Gnome? is he joking?

    and what is this about aqua being "down"? from what i can tell, aqua is still alive and kicking, considering it's in the released and currently updated product. it's not as if aqua is, say, cyberdog or opendoc, after all...

    unfortunately, this is as far as i got in this article, considering his premises are flawed, i can't see how the rest of the review can be any better

  • The Duke's normal quality of article. I could pick it apart but i'll just list one example fo the guy's clueslessness:
    Pros, Top-notch Secutity... doesn't tend to double up on applications (like installing 3 calculators). The less applications ... that there will be less holes in applications (and thus, the system) to exploit.
    Oh come on...
  • While I can agree that encoding MP3s on both platforms make for a fair performance test, compiling the kernel doesn't. To start with the architectures are different, one is CISC, the other is RISC, so the machine code geneated is not the same. Compiling to RISC machine code requires a number of compile time optimizations that compiling to CISC machine code doesn't. Another problem with this test is that the motherboards aren't the same so the required support modules will be different.

    BTW I couldn't care which is faster since I use both the x86 and the PPC - in separate computers ;)

  • OS9 (Score:2)

    by Pope ( 17780 )
    OK, on OS 9 with SoundJam, I can get 160 stereo encodes at around 5.7x on a single 400 MHz G4. I'd love to see how fast the dual 800 they just announced could do it! :)
  • As with many reviews, they are out of date by the time they hit the shelves. 10.1 has been announce. the performance problems have been fixed. I use YellowDog2 and I thought his review was good, though a little on the rah rah side. YDL is the best distro out there for PPC. Let's see if Mandrake can give them some competition.
  • The layer you are thinking of is Carbon.

    Although OS X isn't god awful slow on my machine, a dual processor g4 500, it's by no means as fast as os 9.1, or even suse running on the same machine.

    I've been using MacOS X as my OS since it's release and am presently using version 10.0.4. I'm hoping that within the next few days, apple releases 10.1 with speed increases. MacWorld Expo is happening this week, and hopefully we'll see some new hardware and software.

    My decision to use OS X is based on the functionality and productivity I get out of it. On my system the speed is acceptable, but I've seen it running on some slower machines and I don't think I could use it if I didn't have the dual processor machine. I really hope apple increases speed soon.
  • Honestly I've always liked LinuxPPC. Those guys do a great job and deserve a great deal of credit for getting Linux to the PPCs. I didn't have luck with YellowDog the first time I tried it. I didn't run desktops off of LPPC. I ran servers. I ran a very large mirror server for a number of years and had great luck with LPPC. I'm curious as to what problems the original poster had with LinuxPPC. Sound and X wasn't much of a concern to me as it would be for a person setting up a desktop machine. Perhaps that was it. It made an excellent server OS I thought. Ironically I mirroed YD and LPPC. :-)

    --

  • 45 seconds to encode an mp3!?? (Score:4)

    by bconway ( 63464 ) on Wednesday July 18, 2001 @04:17AM (#77921) Homepage
    Wow. They should check out GOGO [freshmeat.net]. It's originally based on LAME, with major portions of the code rewritten in assembly for speed. It takes advantage of SMP as well, and my dual PIII-550 can encode an average length song in 15-17 seconds using variable bitrate encoding at 128kbps or 192kbps. Granted, I don't know how well nasm would fair on a Mac (probably not at all), but it's a great tool for x86.
  • The benchmarks for the MAC look even better when you take into account that the MAC is being limited by the 4200 rpm laptop hard drive. I'd like to see benchmarks on more similar configurations. Either way the MAC is still to expensive in my opinion, but I am impressed with the performance.

  • Re:Stupidly chosen benchmarks (Score:4)

    by flatrock ( 79357 ) on Wednesday July 18, 2001 @05:52AM (#77923)
    My knowledge of compilers is limited, but I don't think that just because you end up with more, shorter instructions, that the compiler should take longer. In my opinion it doesn't really matter. People using Linux often spend considerable time compiling things on their computers, a comparison of how long it will take you to compile the kernel on each machine is a useful benchmark.

    I would thing that there is a lot of Linux software that has been optimized for x86, since it's the dominant platform, so using LAME isn't that unreasonable.

    It would be good to include another benchmark, on which the app has been better optimized to use the PPC, but I don't think the choice of benchmarks was that bad.
  • There is some problems with bencharming two different architecture.

    1) When you compile a kernel for ix86 and one for PowerPC, the compiler doesn't not even have the same thing to compile. Same thing for LAME.

    2) You're testing also the compiler. It may happen than gcc for powerpc is slower than for ix86.

    You would be better to test with applications that do not contains a lot of architecture-specific optimizations (let's say a web server, or even better, a custom application wrote specifically for benchmarking, don't if it exists).

    I've use LinuxPPC for three years, and let me say that life (with Linux) on x86 is much more easier.
    Also, there is a lot more optimizations for x86, which means that gnome, for example, is as fast on a Pentium 100 than on a PowerPC 604e 180MHz. And some libraries are quite optimized for x86 while being painfully slow on PowerPC (Imlib and Imlib2 are the most outstanding examples). And you always have some programs that have endianess problems. You only have few non-free apps: no Flash, NVIDIA drivers or StarOffice, (although OpenOffice will be a remedy). The last point is nevertheless a mixed bag: you really want free (speech) code, because that's about the only way to get an application to exists on your computer!
  • In the article it ist stated that the pm3 was encoded at 320 kbps, so you cannot compare that with encoding at 128 or 192 kbps.

    Joost
  • I would agree. I installed this on a PowerMac 7500 only to find that the on-board ethernet would loose 85% of its traffic, making it totally useless. Granted, this is more likely a problem with the kernel in general not the distribution.

    I installed this on a PowerBook 3400c, once with 48MB of RAM and a second time with 144. The first pass (which was less than the 64MB of REQUIRED memory) demonstrated that the installer can't handle activating swap as soon as it is created. So even though GNU/Linux could run in less than 48MB, this distribution can't install in less than 64.

    The second pass (after the memory upgrade) installed OK, but the X configuration is just plain wrong (size, depth). In the 16 bit configuration, colors are wrong (red shows up as green, for example). Again, this may be problems with the kernel (YDL's 2.2--the 2.4 would not boot).

    I have also tried installing this with /, /tmp, /var, /usr, etc partitions and gotten the idea that this distribution's installer only mounts / before doing the install and then quickly runs out space. Sorry, but I'd really rather have a small root partition with minimal changing directories (Hello, never had a box become really hard to deal with because /var is in the / file system and a log files gets /.ed overnight? lucky you--or you've never had to deal with inheriting a default install IRIX box. 8-) and leave my big FSCKing to non-critical non-boot partitions.

  • Forgive me, but... (Score:3)

    by kdgarris ( 91435 ) on Wednesday July 18, 2001 @04:14AM (#77927) Journal
    Is covering this topic an example of "Yellow Dog Journalism"? :-)
  • "YUP (their own apt-like update/install tool)

    You can get Debian for PPC as well, and it works great. So, don't go got YDL if what you want is apt! IIRC, despite YUP, YellowDog and LinuxPPC are still RPM-based.

  • A G4 500MHz is not using the MPC7450. It's a MPC7400. Quite a different beast than the 7450, as the 7400 has a 5 stage pipeline, while the 7450 has a 7 stage pipe.
  • So.. If compiling code is what you do for living maybe you should take the X86 then. It isn't in any way invalid.
  • Partly true as I see it. Maybe the kernel wasn't the best option to compile. But if you develop Linux/Unix apps you probably are not crosscompiling that often. A better benchmark should be to compile emacs or vi but not crosscompiling.

    If it's going to be a realworld bench that is. For just comparing speed crosscompiling probably are better.

  • Actually, he's using the Titanium Powerbook, so it's a 7410. Same design, smaller manufacturing process, so it runs cooler.
  • Wow. I'm glad that I wasn't the only one who found this article to be less than good. The one complaint I have that hasn't been posted yet is that the processor he has in the lovely laptop is not an MPC7450 (G4e, 7-stage pipeline, L3 cache), but an MPC7410 (just a G4 7400 on a smaller process).
  • It's a reference to "yellow journalism", the use of lurid features and sensationalized news in newspaper publishing to attract readers and increase circulation. The phrase was coined in the 1890s to describe the tactics employed in furious competition between two New York City newspapers, the World and the Journal. Read about it here [britannica.com] in Encyclopedia Britannica
  • I've got a Powerbook G4 Titanium notebook (G4 400 Mhz and 384 mb RAM)... I run YDL 2.0 on it right now, and it's excellent - as the author of the article mentioned, it easily keeps pace with my 850 Mhz PIII at the office.

    OS X on it is another matter altogether. It seems inherently slow, whether running applications through the OS 9 compatability mode, running native OS X apps (there are quite a few now), or running no apps at all and just the OS.

    My notebook originally came with 128 MB RAM. I upgraded to 384 almost immediately because under OS X it seemed the OS was swapping to disk constantly, with no apps running! OS X easily consumes 128 MB RAM by itself - the price you pay for the Aqua UI, which is incredibly graphics intensive.

    Personally, I love Linux on the G4 - now, I'm not a Mac head, I've got an Athlon 1200 here and a bunch of x86 machines. But the G4 Titanium is such a nice little box I couldn't resist picking one up. Price-wise, it was about $100 more expensive than the Dell Inspiron 8000 notebook I was also considering - granted, the Dell was 1 GHz, but again, my G4 seems just as fast. Plus, Linux runs beautifully on it, and the extra-wide screen makes coding a pleasure.

  • It is also invalid because he is doing make menuconfig and the saving the default .config settings. The problem is, these vary a good bit between PPC and x86, and so the G4 is likely compiling things like ADB support and the PC is probably compiling things like VESA FB support, which both the other platform does not enable by default. So it isn't even different instructions, it's different sources too!

    My experience with PPC Linux on a G4 400 was that it compiled many things very quickly, much faster than the PIII 600 sitting next to it. But the true measure would be application performance, such as Apache, gimp, blender, etc.

  • though I haven't played with OSX-Server, so I can't compare.

    Do yourself a favor and avoid it. OS-X server should not have been sold as a "Final" product but as a beta. It crashed frequently, didn't support many devices (very very few!), didn't support SCSI disconnect (read: could not eject tapes from a TBU without crashing), didn't play nice with NFS on non-OS-X clients, and had virtually no software other than the on-board software. The one backup solution was "recalled" due to reliability and copyright issues, and the AppleShare services did not support AppleTalk. Worst of all, you had to pay hundreds of dollars for the piece of junk.

  • From the article:

    Probably the oddest, part, although, is that you need a 10 MB boot partition. Coming from x86 machines, I didn't see this coming, but with the partition, things worked fine (while they didn't without it).

    Uh, how long have you been linux/unix for? AFAIK, this is considered a good idea, whether or not the distro makes you do it. Remember lilo, and your kernel having to be in the first 512meg or so?

  • yes, they've fixed that now. But for the longest time, this was a big problem. In other words, if windows was on the first partition, and linux on the second, many times it was just impossible to get linux to boot. this goes back about 3 years, but if an author claims to have x86 linux background, he should recognize something like that. Plus, keeping a seperate boot partition is a good idea.
  • As my computer architecture lecturer is so fond of pointing out, nobody gives a shit if foo is faster clock-for-clock than bar, they care if bar can scale to a clock speed that renders it faster than foo's top performer.
  • Are you downloading the correct kernel? If you poke around the Linux PPC web pages and mailing lists you will quickly discover that, yes, the official Linus and Alan trees generally do not compile. There is an active effort to keep a PPC branch up to date. The branch exists mainly because Linus doesn't like the way that the PPC people submit their patches, so they more often than not don't make it into the main tree.

    Having said all of this, I was going to post a link to the PPC BitKeeper pages. They're gone now, with a message that they will be set up elsewhere. Until then, you can find lots of Linux for PPC community resources at http://www.penguinppc.org [penguinppc.org].

  • Sorry for my generialization in the title. But this article is quite far away from being "in-depth". Some quotes: About MOL (Mac-On-Linux, allows you to run MacOS on top of Linux): "This is done sort of like how WINE (on x86 Linux)does things." Wrong! It's more like vmware of win4lin does things. MOL boots MacOs by emulating the traditional MacOS-Hardware and ROMS (and some black-assembly-magick is guess)

    "Like I mentioned earlier, I was disappointed with the speed of OS X -- including OS X Server 1.2 (which I'm told is a little faster -- but I couldn't tell much of a difference). I expected the lack of extensive Altivec support on Linux to maybe hold Yellow Dog Linux (as well as other Linuxes) back." He keeps talking about "Altivec" support all of the time (he even says something of "altivec-optimizations" in the kernel). Doesn't he get the e.g. gcc most certainly will _NOT_ use either altivec or ISSE/MMX instructions?

    "I decided on doing a make vmlinux because Yellow Dog didn't come with anything to make bzImages out of the box." Oh come on. Please. zImage ist _JUST_ for x86 (due to its stuipid design)

    I could go on like this for a while, but I think those examples are enough. This guy just doesn't know what he is talking about
  • Well, as a Mac fan, I see that AMD truly is the big kid on the block, but you have to admit that having benches like that on a 500mHz chip means that the PPC 74x0 chips really are some bad motherfuckers compared to Intel...

    Now if only Motorola can ship some chips that are up to spec.

    /Brian
  • PPC bootloaders suck because up until Open Firmware became the rule, you had to jump through hoops (i.e. kick out MacOS) in order to start something else. You can either hack the firmware (like Darwin/MacOS X does) or you can start up from a stub MacOS Classic partition, but you can't do anything like LILO because the architecture is too different.

    Besides, I think pretty much everyone uses one form or another of BootX anyway, at least where they can't reach the firmware xor are too lazy to learn Forth.

    /Brian
  • Which is why the Mac is a generation and a half behind the PC world, even though it utterly destroys a Pentium III at the same speed.

    That's what I call bittersweet...

    /brian
  • Ah. See, I have a 6500, which has OF, but it's broken...

    But thanks for the clarification. That's roughly what I meant to say in the second paragraph but I never quite got to it.

    /Brian
  • Too bad the Mac's are so expensive, otherwise i would have gotten me one myself.
    I really cant figure it though... if you bought a Mac you probably want to run OS X on it anyway. Anyone knows how many users they have ? Growing since OS X or loosing shares ?

  • Wow! And you think Apple is in a Tower of Hubris!? Look in the mirror, dude. That was the least informative post I've read in quite some time. Wishing I had moderator's rights just about now.

    Motorola engineered AltiVec which makes Photoshop go fast fast fast on a Mac. I make my living off of Photoshop and Illustrator, etc. Therefore, the AltiVec co-processor on the G4 helps me make a living. I love the little bugger! Get off your high horse! Not everyone is a programmer nor is everyone adminsitering some huge piece of iron from IBM. Calm yourself down and get just a touch of perspective!!


    ---------------------------

  • Linux + OSX (Score:3)

    by iggie ( 183722 ) on Wednesday July 18, 2001 @07:23AM (#77949)
    I've used Linux for about a year on my PowerBook G3, but haven't used it on that machine since OSX was released. Its my primary development machine for an open-source DB-Perl-Web app, which is deployed primarily on x86 Linux servers. Why did I switch? Simply, OSX is a better (much) desktop system than Linux. Yes, OSX has a noticeably more sluggish UI than Linux, but not annoyingly so. More importantly, its more powerfull and better integrated so it allows me to work faster despite being more sluggish. It is far more annoying dealing with X-Windows cut-and-paste, for example, and I do a lot of my typing by cut and paste. For the first time on a UNIX system, I've actually been using the file manager (Finder) to navigate directories rather than the command line (!). Its also nice (very) to plug in an 18" LCD monitor at work for additional (not mirrored) desktop realestate - a feat I couldn't manage in Linux despite much effort. It would be nicer still to have multiple work-spaces in addition to the dual monotors, but I haven't missed them as much as I was sure I would. Also, its very nice (critical in my case) to go on the road and be able to do live demos of a client-server app from a laptop connected to an LCD projector. Other road-related things are much better battery management (8 hours with twin batt. pack), sleep - someting not possible on PowerPC Linux as of 4 months ago, and much simpler management of multiple internet connection methods (dialup, DHCP, static IP, 802.11b - all with just a menu selection or with auto-detection). All that being said, as I mentioned above, the deplyment systems mostly run Linux (on x86). Why? because Linux is a great server platform, that's why - though I haven't played with OSX-Server, so I can't compare.
  • Since you asked here is a brief-ish account of what happened.

    It was Sept/Oct 99 and I was trying to help my friend Andy install LinuxPPC on his Mac. He wanted to try it out - so he wanted to dual-boot with MacOS (8.6 I think).

    I'd been using Linux on PCs since 94, plus I was more of a techie than Andy. But crucially I don't know Macs and Andy didn't on a techie level - though he was what I would call an expert user.

    I'd be lying if I said I can remember the exact details of everything we did but I can best sum up the problems we encountered by saying that the state of the Linux install seemed similar to some of my first experiences in x86 linux installs a few years previous.

    The documentation was scant - you had to hunt out HOWTOs and FAQs from the web. So we'd get part through the install, realize we'd miss something, reboot into MacOS to connect to the web and search for help.

    The disk partitioning was done by hand - with a command line interface. I was fairly comfortable with this but at the time x86 commonly had character-graphical tools.

    At one stage we corrupted the boot sector (or the Mac equivalent) and we were in the Mac firmware/BIOS. I was surprised and slightly daunted to realise it was written in Forth! - which I very very vaguely knew from school days. A visit to Apple's technical support webpages and we eventually fixed that problem.

    Overall I think I can't be too critical of LinuxPPC itself (and if my original post seemed so I apologise). I think it was a combination of factors, including

    - my lack of Mac expertese
    - trying to install dual-boot, always harder than a straight install

    but also I think *at the time* you did need to be fairly technical and knowledgeable to install it. Whereas Yellow Dog as described in the review seems much more end-user friendly. But then again I'm sure by now LinuxPPC is just as easy to install.

    The postscript to the story is that after a fairly difficult few days of installing, Andy was very pleased with his new Linux system. He had to learn a few new concepts but soon he was up and running. He liked the desktop (Gnome IIRC). He was impressed by Gimp - whilst it was no Photoshop it was the best thing he'd seen for free. So a happy ending.

  • Slowness of OS X (Score:4)

    by RatFink100 ( 189508 ) on Wednesday July 18, 2001 @04:18AM (#77951)
    Pretty thorough review and the guy seems to know his Macs (which I don't particularly).

    I was just wondering though if he perceives OS X as slow because there really isn't much native software yet - it all runs through the compatibility layer (forgotten the name).

    But Yellow Dog certainly seems to be a lot better than LinuxPPC which I tried to help my friend install and found tricky. But that was a year ago and things have moved on.

    I was disturbed that there seem to be a number of installation options you can't change - I think it needs an 'expert mode'.
  • Well, a lot, really. I thought the article was pretty poorly written.

    What they really don't cover, though, is the boot loader. PPC Linux boot loaders traditionally suck, from what I remember of them. To date I haven't heard of one that really "worked" like LILO does. (Though hopefully YDL 2 or Mandrake 8 beta will.)
  • Actually, anyone with Open Firmware probably uses yaboot/yabin, which isn't too bad, but the installation, configuration, and results are well behind what they should be. With OF and some firmware hacking, you *should* be able to get something reminisent of LILO.
  • I bought a Mac quite a while ago (long before OS X was really much more than a vague promise), and it won't ever run OS X well, but it kicks ass with YDL 2. Linux is a vast improvement over pre-X Mac OS's in everything but device support, which is coming along nicely.
  • Not necessarily. The comparison isn't necessarily valid not because one compiles more slowly than the other but because the thing being compiled was not the same piece of software. If, using the same config, you compiled one kernel on an x86 machine and cross-compiled the same kernel for x86 on PPC, then you might start getting closer to a valid comparison. As it is, you may as well compare compiling emacs on x86 to compiling vi on PPC.
  • Actually, often times I have used a (b)zImage to fit a large image file onto a floppy...extremely useful when trying to boot pesky old hard drives...
  • Good point. The documentation for most MP3 encoders doesn't really make it clear to people that when you enter a bitrate on a VBR encoder, you're getting an *average* bitrate, not an exact bitrate, and if your audio file is funky enough it won't even average out to 128 or 192 anyway, because it's only a target.

  • Wrong. Even granting you your premise:

    The first benchmark is valid because it shows how a frequently performed task may be affected by switching architecture.

    The second benchmark is valid because users of an uncommon platform cannot expect to find the kind of optimizations for their programs that users of the common platform are used to.

    That said, of course two benchmarks don't tell the whole story. For a bigger picture, try this collection of x-platform benchmarks [jc-news.com]. The stats are collected from many sources, so they're not always based on the same platforms, unfortunately. On the plus side, there's some altivec stuff in there, too.

  • My knowledge of compilers is limited, but I don't think that just because you end up with more, shorter instructions, that the compiler should take longer.

    Actually the binary size is larger too--you can't compile a PPC kernel that will fit on a floppy weithout gzipping it and tacking gunzip onto the front of the code (the infamous zImage.) However this could be due to the different drivers required by the platform as well.

    RISC instruction sets in general also expect the compiler to be "smarter" about optimizations. This causes the compiler to take longer.

    I would thing that there is a lot of Linux software that has been optimized for x86, since it's the dominant platform, so using LAME isn't that unreasonable.

    In general this isn't true; most programmers tend to avoid tying their code to any one architecture, because once you do that it becomes very difficult to port to another chip, and you wind up pissing off the most vocal 10% of Linux users. Any programs which compile out of the box for multiple architectures (in my experience, most of them) won't be particularly optimized for any one. LAME is a special case in that the cross-platform code was written first and then x86 opitmizations were grafted on with #IFDEFs.

  • Stupidly chosen benchmarks (Score:3)

    by plastik55 ( 218435 ) on Wednesday July 18, 2001 @04:57AM (#77960) Homepage
    The first bench mark (compiling a kernel) is invalid because, as anyone with half a brain knows, compiling a PowerPC kernel is not the same task as compiling an x86 kernel. PowerPC code, being more RISC-like, requires more instructions--making it take longer to compile on an equally fast processor.

    The second test (LAME) is invalid because several important parts of LAME have been hand-optimized in x86 assembler.

    Blech.

  • Moderator: This user's post was a pun, not a troll.

    Check out:

    http://www.thedukeofurl.org/reviews/misc/yellowdog 20/7.shtml [thedukeofurl.org]

    from the review, or more specifically:

    http://devel.yellowdoglinux.com/rp_yup.shtml [yellowdoglinux.com]

    I certainly preferred this rather than some lamer posting 'Weee 1st post!'.
  • and what is this about aqua being "down"?

    I think the poster was referring to the fact that OS X's interface (Aqua) is very slow. YellowDog Linux and KDE or Window Maker is *much* faster than MacOS X and Aqua. I had heard (rumor only) that in order to get MacOS X out (and stop any potential developper bleed), Apple had to compromise and write Aqua in Carbon instead of Cocoa. The good news is that this could certainly explain the performance problems with Aqua, and better news is that it probably means that there is room for a lot of improvement in future releases. Until then, I'll run YDL (and probably afterwards anyway).

    So, you shouldn't let a little thing like that stop you from reading the rest of the article, it isn't that bad.
  • It is just slower. Everything is slower.

    no, everything is not slower. the only thing thats slower is the gui. and as jobs showed [apple.com] today, thats being fixed and will ship in sept

    informal tests of apache show it to be faster on os x than linux, hopefully /. will post these tests when theyre complete. log in as >console and see for yourself

  • i havent done any formal benchmarking, but ive found networking and java execution to be much faster. a /. poster mentioned informal apache results and promised formal ones soon. opengl is 30% faster and games developers have stated improved subsystem efficiencies (esp memory) have given significant boosts to their games (see quake iii)

    mach-is-slow is a common myth probably borne from the ver 1.0. mach can actually be faster than a monolithic kernel, esp where multiple processes are involved. see my paper [mac.com]

  • Ol beige PowerMacs run yellow dog really well (I run one). OSX is a bit of a hog on them, but linux is more than functional. You can get into one of those *cheap*.
  • yeah, I am suggesting you run right out and do that.

  • Linux on Mac (Score:3)

    by xhypertensionx ( 229085 ) on Wednesday July 18, 2001 @12:29PM (#77967)
    Although I will be installing YDL on a partition in my new 867mhz G4 when I get it, there are some points I would like to make.

    I am not doing it because Apple isn't providing me with something stable or fast enough.

    I am not doing it because the OS doesn't provide enough features.

    I am not doing it because Linux is superior to OS X.

    Quite frankly, although these are arguably untrue today, once the tranistion is completed, and OS X becomes the dominant and leading consumer *nix distro, its superiority will be unquestioned.

    This article is a perfect example of the Linux/OS X cold-war taking place. Although the two have many things in common, and enjoy mutual benefits from the other's existance, there are many in the Linux community who are threatened by and fear OS X. Even Mr. Torvalds fired a shot, stating that OS X makes ALL the design flaws one could make, AND invented some of its own. (I'm fully aware that it was probably directed at the micro-kernel, but what was said was said).

    Who can blame the Linux community for feeling animosity towards OS X? Much of the Linux community has spent the later part of the 90's trying to convince their friends to take MS off their desktops in favor of a "free" (free as in most people could care less) *nix OS.

    All of a sudden, a company develops a *nix distro destined to be more beautiful and more usable than any other, with the goal of making it the world's most advanced OS and finally bringing *nix to the users -- within ONE year. Where Linux has failed, Mac OS X will succeed.

    To add insult to injury -- this company is Apple. A company that many curse as a cancer on the computer world. A company whose user base is considered to be sub-human by ungrateful Linux and Windows users who lash out at Apple while using their Macintosh-derivative computers and USB peripherals. It seems that the author of the article is one of these people. If he was really the Mac user he claimed to be, he wouldn't have used the same weak arguments against OS X that won't mean anything in September.

    Although I say this, I am not so preoccupied with this cold-war to not use Linux as a tool when needed. Particularly, I am installing it to learn it and to play Counter-Strike.

    Those who are trying to promote Linux on Macs are going to shoot themselves in the foot by attacking OS X. The best way to promote it is to target the older Macs, those who want free (beer) software, those who want to learn, and those who want games like Half-Life.

  • Too bad the Mac's are so expensive, otherwise i would have gotten me one myself.

    I wouldn't go that far. It's nice to see the Mac community getting a Linux they deserve, rather that some of the second-hand ports which have been available. Buy a Mac though? No thanks - I'll stick with my Athlon for the moment. If only, as you mention, for the cost, and the (lame) fact that I'm used to having a PC on my desk.

    cmclean

  • am I missing something, but as of right now both lilo and my kernel are located well beyond 512megs, as in or around 11gigs

  • RISC v. CISC (PPC v. X86) (Score:4)

    by daniel_isaacs ( 249732 ) on Wednesday July 18, 2001 @05:29AM (#77970) Homepage
    Just to get this out of the way (and quell the predictable debate about PPC v. X86) anyone posting a thread conerning this aspect of the story should read this [arstechnica.com] first.

    That's http://www.arstechnica.com/cpu/4q99/risc-cisc/rvc- 1.html for the Goat-phobic.

  • umm... It would be extremely hard (I have no experience with assembly for mac, though) to port assembly code from x86 to another platform.

    Also, the article says that the MP3 that is enbcoded is over 7 minutes long, and encoded with 320 kbps. I can't really see that your comparison does any good (and mp3enc is not that much slower, I think I used about 30-40 seconds to encode a song last time I used it P3-500)
  • I run OS X on a 300 MHz beige G3 (aka Gossamer), the oldest officially-supported Mac. (According to the Apple History [apple-history.com] site, the machine was available 11/97-1/99, so it's almost a four-year-old model.) It seemed slow, even with 576 MB of RAM, but with each update it has improved. I do know from experience that, for the apps I use, the difference between the G3 and G4 is a lot more than can be accounted for by the MHz gap, even in OS 9, aka Classic. Going from OS X to Classic is not a swift process, but again, it has been improving with each release.

    I think one thing that people tend to overlook is that OS X is still in the Beta stage. I can't really blame them for that view, since it's been commercially available for some time, but Apple just started shipping it with new computers in the past couple months, and even then the systems are set up to boot to OS 9.1 as the default. Personally, I think OS X is pretty slick, and given my machine's age, I am ready to get a new system ... after I buy a house this fall. (First things first.) We'll see if iCEO Jobs introduces dual-processor G4s today. I hope I hope I hope.

  • In other news... (Score:4)

    by Balinares ( 316703 ) on Wednesday July 18, 2001 @04:35AM (#77973)
    The Green Ostrich for VAX and Pink Shoe for Spectrum Linux distributions should be made available soon.
    Microsoft's Blue Screen distribution is expected for later this year.

    -- B.
  • But ofcourse it does, its linux isnt it?
  • One thing that really bothers me is that i can't just download a new kernel and compile it. With every try i have i get disapointet by getting some sort of error. On my pentium machine i just compile the new kernels (i like running new kernels). But here i just run the suplied kernel.
  • Thanks for the review. Anyone who produces information is a good thing...

    I think it should be pointed out wherever possible that Motorola Bastardizes the PPC, and that IBM's Power III and upcoming Power IV are the salvation of PPC. Apple is an tower of hubris, they lost market share and are 15% owned by Gill Bates because of one thing: overpriced garbage.

    We must look to IBM to provide us with decent PPC hardware! The PPC is a thing of raw power and beauty, and it really does shine on IBM's power workstations and whatnot. With the Intellificastion of the Alpha [(I know someone on the Alpha design team - here is the scoop, Intel can use Alpha MA, EV7 will be finished at Compaq - Compaq will offer 4 to 256 CPU based servers, the EV8 design team will work for Intel in one year, Intel believes VLIW/Merced/McKinley was a mistake and will charge the EV8 design team to come up with a new CPU with the only requisite: it must be IA-64 compatible)], and the deathly low margins of Athlon CPUs, we really need PPC to stop being bastardized by Motorola. Screw Apple and Motorola.

    For the record, OS X is the most useless thing I have ever seen creep out of Apple. I am an ardent fan of BeOS, and Apple would have done much better to have embraced BeOS, offered Intel based "IMACS" for the cheap stuff, and hardcore POWER-3 powered high BeOS Macs.

    The error that they made will live in the annuls of history at this point.

    I call upon the PPC community to dethrone that pompous ass referred to as Steve Slobs and his vile Microsoft's bitch company, Crapple.

    By the way, Intel may not be as badass, but its CHEEP.

    PS. CON TE PARTIRO, Crapple CUBE.
  • Most of the information was provided in the article, that being that PPC is slower.

    Why would such hardware command a premium. I'm not here to engage in warfare, I'm just dying for a way out of the Intel gammit, and Apple isn't paving a golden path. Of all the zealotry in the world, the worst is this AltiVec. Its almost meaningless, I went to all sorts of sites, and the advantages of AltiVec are that is some doofy name to put on a box and market. Its constantly pointed to and said, there it be, the goldne ALTIVEC. Yet it fails to produce much interest. Maybe I have a myopic approach to looking for information, and I want Apple to fail Thats what you think. I happen to have exposure quite frequently to "new" MACS, they are boring and overpriced machines.

    You have not convinced me you have tried the alternatives. You have not. You stick by what you know and thats fair. I feel confident in saying that I know what I have tried, this ranges from commercial Unixes, BSDs, Linux, MAC OS/OS X, and the Microsoft junk. After the novelty of "the color computer!" wore off, Apple has in my estimation failed to trailblaze much of anything for our industry. And if Motorola was so great, why did Apple and Motorola have some of the bloodiest corporate fights?

    This is just an opinion, and CIO's agree with me, Apple certainly isn't the lesser of two evils. Market share goes to show that.

    Apple users make it out like they have some uber workstation, like having and Indy 5 years ago, and there is all this power they have no one else does.

    Trust me, Power-III CPU based workstations start at $20,000 for a reason, superior architechture is at the top of the list.

    What is with all your emphasis!! In your! Post! You seem to have quite a bit of emotion attached to this discussion, which leads me to believe you make emotional conclusions and not those based on fact or experience.

    You would mod me down because I have said what you did not want to hear, not that my arguments were flat out wrong or flaming or trolling.

    You would mod me down because I have rendered an opinion with which you disagree. This is a sad revelation to make to our community about your nature - censorial.

    Now, as far as getting a perspective, you product something a little more convincing that "AlitVec rules!! I know! I use Photoshop ALL DAY MAN!

    Prove me wrong, show me some web site that says, yes, here is a barrage of valid cross platform cross operating systems tests that shows without a doubt that PPC kicks the crap out of any Intel CPU.

    I have searched for such evidence and have found none. Only Apple's heavily disclaimed* ** *** claims on thier website say such things.

  • Since Yellow Dog is so new, security is obviously very good.

    This is not obvious. Furthermore the security of an OS is not causally related to age of the release.

    Note to all IT professionals: Security through obscurity NEVER WORKS.

  • Wow. They should check out GOGO . It's originally based on LAME, with major portions of the code rewritten in assembly for speed. Granted, I don't know how well nasm would fair on a Mac (probably not at all), but it's a great tool for x86.

    It'd be interesting to see what a talented PPC assembly hacker could make of GOGO. The blurb on the Freshmeat page says it takes advantage of MMX, 3D Now!, and SSE.. given that AltiVec is better than those (similar power to SSE, but with 32 SIMD registers instead of 8, and a wider range of instructions), it could be an absolute MP3-encoding monster. Of course PPC assembly hackers are probably a lot rarer than x86 ones. :-(

  • Too bad the Mac's are so expensive

    On the desktop, sure, where you have to compare them to the frankenstein boxes that all us geeks love to build from parts.. but one of those new iBooks would be a fine, fine Linux laptop, if they could just get the audio working. But it's being worked on [xiph.org]..

  • "Classic" is the mode where Mac OS 9.1 boots up inside of Mac OS X. The Finder (the Mac OS shell) is not started in Classic. Instead, non-Mac OS X programs are started from the Mac OS X shell, are ran in Classic, and appear just like any other Mac OS X program (with the same GUI they would have in Mac OS 9.1, afaik) The programs runing inside Classic and those running in Mac OS X interact just like they would if they were both Mac OS X or Mac OS 9.1 programs.

  • YDL 2.0 rocks! (Score:3)

    by agupta_25 ( 468946 ) on Wednesday July 18, 2001 @12:07PM (#77982)
    Yellow Dog Linux 2.0 is by far the best distribution I have used that actually "works" and is the easiest to install yet. I have tried LinuxPPC 2000 Q4, Mandrake and Debian. All of them suck in one way or the other (well ... LinuxPPC is probably the closest after YDL). YDL 2.0 on the other hand installs like a charm. Of course, there is still a lot of room for improvement, but given the state of other installers, I think YDL is the best so far. I am using YDL 2.0 with a 2.4 SMP kernel on my dual G4 450MHz PowerMac. It blows Mac OS X right out of the water. KDE looks awesome and it runs all my favourite games using SNES and MAME. My only gripe is that I am stuck with a 1280x1024 resolution (I have a Apple Studio 17" monitor) and YDL will stubbornly refuse to let me set 1024x768 ... but hey ... who cares. KDE looks great at 1280x1024 with anti-aliased truetype fonts. Konqueror is awesome and is FAST on PPC. I even have accelerated X server support for the Rage 128 video card! Since I started using YDL 2.0 on my PowerMac, I have virtually forgotten my x86 installation of Linux ... All I will ever need runs great on the PPC and much faster at that. This really makes me think how much the x86 architecture sucks ... YDL 2.0 rocks!

Slashdot Top Deals

This restaurant was advertising breakfast any time. So I ordered french toast in the renaissance. - Steven Wright, comedian

Close