Slashdot Mirror
Yellow Dog Linux 4.0 Reviewed
eobanb writes "I finally wrote a somewhat in-depth review of Terra Soft's Yellow Dog Linux 4.0. It's basically a PowerPC port of Fedora Core 2. The good? Pretty modern software, and setup is a snap. The bad? RPM sucks as always, and there are a few too many things that are broken out of the box. Linux PPC; it's a niche-within-a-niche, as I heard one Slashdot comment call it, but it may well be worthwhile if you're annoyed by x86 hardware."
Yes, actually, this is a great niche for YDL.
I have also witnessed YDL 3 turn throw away g3 macs into stable, useful desktop systems, running firefox, snappy word processors like ABIWORD, and things like XMMS and Mplayer for multimedia.
Kernel threads almost universally stay on the cpu they were originally assigned to. High performance threaded subsystems, such as the network stack, are replicated. That is, the network stack creates multiple threads (one per cpu) and those threads do not migrate because, obviously, they do not need to.
Generally speaking, the purpose of making thread migration explicit instead of automatic is to partition a larger data set across available cpu caches rather then cause the same data to be shared amoungst all cpu caches. The processors operate a lot more efficiently and SMP scales a lot better. Most people do not realize the horrendous cost of moving threads between cpus because the cache mastership change is invisibly handled by hardware, but the cost is still there and still very real.
-Matt hf
Actually wrong. iBooks start at $999. PowerBooks at $1599. The only PowerBook that is around 3k is the 17 inch one, which goes for $2799.
My employer uses it because we have many mission critical legacy programs that were written on Harris mainframes. Its that big vs little endian thing. So, we are using PPC servers with Yellow Dog. The programs are huge and cannot be converted for many reasons.
So, there are uses for it.
My reasons for use of yellow dog have nothing to do with annoyance with X86 hardware. It differs from PPC hardware in interesting ways that give each their niche.
Rather, it's because I'm annoyed with windows. While I have ~6 functioning computers around at any one time, I do the majority of my office, graphics design, and development work on a Mac: Windows is broken and Linux doesn't run Adobe.
As an active consultant and developer, I upgrade my current desktop mac every 18-24 months.* This means I inevitably have multiple old macs sitting around in closets. When I need to put together a quick server for file backup or web app testing, I grab one and throw YDL on it.
Incidentally, though, when I do have x86 hardware lying around, I use that and debian instead. YDL can be kind of a PITA.
I stole this sig from someone cleverer than me.
In order to run Linux on "Old World" hardware you need an application called Bootx
http://penguinppc.org/~benh/
In order for it to work you need a Mac OS installed on the computer. On the beige G3's I have installed it on I usually set it up like this:
OS 8.1 installed on a 100 MB partition.
Install Bootx as an extention.
Install YDL using the remaining HD space.
All is good.
Bring out the CISC-versus-RISC skeleton was not a very good decision, simply because it Doesn't Apply (TM). Face it - most processors in the 21st century are a *&$#( to classify as either RISC or CISC.
/preferred/ to program as if IA-32 was RISC-ish. See gcc -S output. Finally, the traditional characteristic of RISC processors, yes pipelining, has been available for ages the IA-32. So... is it pure CISC? Maybe not?
Lets see the current facts:
1) The clearly CISC-ish IA-32, at the current stage has half its 8086/80186/80286 (and I do mean the 16-bit) instruction set deprecated. Those *complex action performing* instructions have not had any performance edge over the simpler once since the days of the 80486, and shouldn't be used... unless your prerogative is code-size rather than performance. So, in other words, it is now
2) The clearly RISC-ish PowerPC received a healthy injection of CISC-ishness with the arrival of Motorola's Altivec in G4. Sorry, those SIMD extensions are not RISC-ish. So is this a pure RISC? Maybe not?
There are plenty of reasons to hate the IA-32. As a kernel developer (homebrew project) I find the IA-32 a register-starved architecture, systems programming for which is a pain-in-the-ass considering the oddball hacks that the Protected Mode and IBM's design of the IBM PC AT were. As a neophyte PowerPC assembly programmer (I did hack Quik to remove the stupid linux kernel size restriction though). I thoroughly enjoy the PPC ISA and I am in love with the PPC memory management facilities.
It is a stupid to claim that Apple laptops consume less power simply because x86 is CISC(ish) and THUS uses more power. Design matters, but implementation matters too. Take a look over at the Athlon. The Athlon is its own architecture that has an x86 personality (apparently emulates the IA-32) - but the design of the processor has nothing to do with any Intel design. Looking even at Intel designs, the P3 (... and Penium M) and P4 designs differ. The P3 builds on the success of the Pentium Pro/II while P4 is well... IMHO a flop... and intel should just lean on the Pentium M.
It is not prudent to claim that the latest IA-32 and AMD-64 processors owe anything more to the 8086 MCU other than a half of largely unused, despised, deprecated performance-lacking instructions as well as opcode compatibility.
There are plenty of reason why I am soon to put $1k into the purchase of an iBook - but not because PowerPC "is a a RISC-based architecture" unlike that "aging, bulky CISC" x86.
From what I understand, the main thing that differentiates RISC from CISC is that CISC code goes through a translation layer in the CPU, wheras with RISC the opcodes you use are exactly what gets executed by the CPU - thus reducing complexity within the processor, and pushing it up to a higher level (the compiler).
The comments of those saying that in x86 the instruction set gets translated to a RISC like one by the CPU are basically proving the point that the x86 archetecture is very definitely CISC.
I have no idea whether or not PPC has a translation layer, but I wouldn't say that the existence of SIMD extensions alone disqualify it from being RISC.....
However, you're right - RISC vs CISC is pretty much irrelevant these days....
Advanced users are users too!
Not an (i|Power)Book owner (yet!!) but AFAIK there is no Airport Express support in YDL. So if you have an iBook or PB12" you will have to get a USB WiFi adapter, or get another PCMCIA WiFi card for the rest of the PBs. I recommend anything based on the PrismII chipset (use wlan, orinoco_cs, or hostap drivers). Btw, if you are going to buy a Linksys WPC11 - get v.3. AVOID WPC11 v.4 - its based around some obscure Realtek chipset and the drivers suck (and are x86 only, and crash on anything older than 2.4.18). http://alumni.imsa.edu/~andyw/projects/wpc11v4.htm l
May I also suggest the good ubuntu port for PowerPC? I'm using that on my 15" Al, aside with OS X, and it seems to be built very well...
I'm fat, you're ugly. I can get slimmer, and you?