That's not entirely incorrect, but it's not the whole truth. There is a very small amount of FreeBSD code in Darwin, specifically the networkiing code. The majority of the BSD layer is 4.4BSD. The far more technically accurate statement is that Darwin is based on 4.4BSD and Mach 3.0, with some code from FreeBSD and NetBSD. If you go to Apple's Darwin page, they say basically this: "Darwin integrates a number of technologies, most importantly Mach 3.0, operating-system services based on 4.4BSD (Berkeley Software Distribution), high-performance networking facilities, and support for multiple integrated file systems."
The tone of the grandparent's post indicated that he did not understand how little FreeBSD code as actually in OS X. He claimed that the release of FreeBSD 6.0 would be good for OS X because it would be the basis of the next version of Darwin. This shows a clear misunderstanding of the underlying technology, unless he's very excited about new networking code...
There is BSD code in Darwin, but it doesn't use the FreeBSD kernel. Darwin is CMU Mach 3.0 + 4.4BSD-Lite2. Apple imported some filesystem code from FreeBSD, and userspace tools from FreeBSD and NetBSD, but the kernel isn't FreeBSD's and Darwin doesn't benefit from any of FreeBSD's performance improvements.
I assumed that the reader knew that the Pentium-M is faster, clock-for clock, than the Opteron in integer and slower than the Opteron in FP. It's something that your average Slashdot reader commenting on a CPU thread should know about.
Ghz are not directly comparable between different processor architectures, and PowerPC has tended to deliver faster chips at the same Mhz as x86.
Not for the G5. Remember, the G5 is a long-pipeline design like the P4. According to my own benchmarks, my 2.3 GHz G5 is about as fast as a 1.6 GHz Opteron in integer and as a 2.0 GHz Opteron in floating-point. That means that a 1.5 GHz G5 won't be a match for the slowest Pentium-M.
What is there left to strip out? All of Apple's machines ship pretty bare-bones (512MB of RAM, etc), aside from the optical drive.
b) use cheaper components
Again, what could they make cheaper? My PowerMac uses Kingston RAM and the absolute cheapest 250GB SATA drive you can find. It uses a sub-$100 graphics card (on a $2700 machine). Yeah, they could use a cheaper case, but then the thing would be a Dell outside as well as inside.
I have no idea why everyone is so damn enamoured about the PA Semi chip. Even their theoretical 2007 SPEC scores kinda suck. Their integer performance is so low, they don't make Athlon64s that slow!
Generally, it's so you can have a document and whatever resoures you're using to type that document on screen at the same time. I almost never need to see more than a few lines at a time of the document I'm typing, but I always have several windows that I need to figure out what to type.
Morality is bunk. There are rights, and that's it. Morality has nothing to do with any of it. I refuse to have a conversation on such a nebulous, subjective, and frankly childish concept. When you're ready to talk in adult terms, well, you know where to find me.
Are we talking about the same thing? I'm not suggesting that people go out and downloading OS X. I'm talking about going to the store, buying a copy of Tiger, and installing it on your no-name PC. I don't think there is anything wrong with that. I cannot, on principle, buy into the idea that a company has any right to tell you what to do with their product (assuming you abide the law, that is). If they don't want people to install OS X on no-name PCs, they shouldn't sell OS X as a seperate product.
Eh. It's more like: if I buy a replacement Aztec transmission from the OEM and stick it in my home-made go-cart, is that okay? As far as I'm concerned, yeah, it is!
Opps. I guess the G5 does things differently than the Opteron. On the Opteron, you need a 64-bit kernel to run 64-bit binaries. Well, I stand corrected.
You can't mix 32-bit code and 64-bit code in the same process. You can have two processes, a 32-bit GUI and a 64-bit backend, but then you've just basically got two seperate apps communicating with each other.
The PII was always a better CPU all-around than the G4. The G4's floating-point performance was never very good, unless you could use AltiVec (which wasn't all the time). The G5 was never a leap over x86. From the beginning, unless you could use AltiVec (and a lot of stuff can't), it was about comparable clock-for-clock to an Opteron. When the 2.0 debuted, Opteron was at 1.8GHz, so the G5 was faster, but it was never a leap. Since then, the G5 hasn't got much faster (the 970MP is 2.5GHz), and neither has the Opteron (the dual-cores are at 2.4GHz).
It doesn't matter if they are moving to Intel instead of AMD. The OP claimed that PowerPC was efficient. My point was that the current major desktop/workstation implementations (G4 and G5), aren't.
LOL. Some Mac using moderator trying to silence the truth.
To appease the Mac faithful. I'd like to point out that I like the software on the machine very much. It's got very good Lisp development tools. However, hardware-wise, I have a much nicer Athlon X2 machine right next to it. It's (much) quieter, faster, uses higher-quality parts (name-brand RAM, Seagate hard drive), and fairly good looking to boot. Oh, and its $500 cheaper, and then only because I got an ADC discount.
Wow. You managed to regurgitate something without actually understanding it. There is no way the OS would be able to run 64-bit applications without being compiled for 64-bits. On Tiger (different from Panther), which can run 64-bit apps, the kernel is compiled as 64-bit code. Then, there are two versions of a couple of the libraries (System and Accelerate), one 32-bit and one 64-bit. What's missing is 64-bit versions of stuff like Quartz or Cocoa, which means that 64-bit apps are basically limited to the command line.
Bullshit. I remember when the G4 vs PII debate was particularly hot in the mid/late 1990s. The mantra there was that PowerPC was better, not about performance-per-watt. Well, fancy multicore oomph-per-watt chip that everyone is salvating is a warmed over version of the PII that the Mac folks found so entertaining.
With regards to CPUs, the Mac folks were, to put it simply, wrong. The G4 was never better than the PII, and the G5 was overtaken by the Opteron as soon as it hit 2.0GHz.
The dual core PowerPC brings IBM to parity with year-old AMD hardware. I have a dual-core 2.3, and a dual 2.0Ghz Opteron would be entirely competitive (win in integer, win most floating-point, lose some media-processing). The dual core 2.2 Athlon64 in the machine next to it is a step up in almost every benchmark I've tried.
You don't remember the merciless derision of the Pentium II? The "The G4 is faster than the fastest PII"? It is so deeply ironic that the Yonah chip that will replace the G4 has almost the exact same execution core as the PII that Mac users made so much fun of.
The point is that the 5% of your peripherals that are supported are very common. Looking at the Intel HCL, I know I could easily dig up several of those cards (lying around the house). Most onboard AC/97 soundcards seem to be supported, which is what is used on the Mac anyway. The only sticky point is video (only Intel 900GMA cards are accelerated), and perhaps SATA (nForce4 SATA isn't supported, most PATA controllers are). Firewire and USB are standard EHCI and OCHI, so that's all good. What more does your average user have?
Robust classy hardware my ass. I just got one of the new PowerMacs. Yes, its a beautifully-designed case, but the cooling system is loud, it uses a crappy (and loud) Western Digital harddrive, Delta fans with audible bearing noise, and cheap generic Kingston RAM.
These aren't the Mac's of yore that came with SCSI and the like onboard. These are a fairly generic Dell machine in a fancy aluminum case.
The G5, at least, isn't that efficient. I just bought a brand new PowerMac G5 (dual core 2.3GHz). It's certainly a fast machine, but for almost everything I do, its slower than the 2.2Ghz dual-core Athlon X2 that's sitting next to it. For compiling code, it's about 70% as fast as the X2 system. For SciMark, it ranges from 95% as fast (for the small in-cache dataset), to 80% as fast (for the large in-memory dataset). For nbench,if you leave out one really awful score that's probably the result of a bad compiler optimization, its about 80% as fast. These were all done with GCC 4.0, of course. The 970MP SPEC benchmarks suggest that if I used XLC (and EkoPath on the X2 to be fair), I could probably get it to be 90% as fast in integer as the X2 and 25% faster in floating-point, but considering those scores is entirely an act of intellectual mastrubation, since most stuff on OS X is compiled with GCC or CodeWarrior anyway.
Of course, I love the machine to death, because of OS X, but the way I see it, Apple is going to gain a good deal of performance by moving to x86.
GNUStep is based on OpenStep, the same programming framework OS X's Cocoa is based on. The biggest difference between Cocoa and GNUStep is that the former has had a once-over by all of Apple's artists:)
Slashdot Mirror
Um, not on this CPU :)
That's not entirely incorrect, but it's not the whole truth. There is a very small amount of FreeBSD code in Darwin, specifically the networkiing code. The majority of the BSD layer is 4.4BSD. The far more technically accurate statement is that Darwin is based on 4.4BSD and Mach 3.0, with some code from FreeBSD and NetBSD. If you go to Apple's Darwin page, they say basically this: "Darwin integrates a number of technologies, most importantly Mach 3.0, operating-system services based on 4.4BSD (Berkeley Software Distribution), high-performance networking facilities, and support for multiple integrated file systems."
The tone of the grandparent's post indicated that he did not understand how little FreeBSD code as actually in OS X. He claimed that the release of FreeBSD 6.0 would be good for OS X because it would be the basis of the next version of Darwin. This shows a clear misunderstanding of the underlying technology, unless he's very excited about new networking code...
There is BSD code in Darwin, but it doesn't use the FreeBSD kernel. Darwin is CMU Mach 3.0 + 4.4BSD-Lite2. Apple imported some filesystem code from FreeBSD, and userspace tools from FreeBSD and NetBSD, but the kernel isn't FreeBSD's and Darwin doesn't benefit from any of FreeBSD's performance improvements.
I assumed that the reader knew that the Pentium-M is faster, clock-for clock, than the Opteron in integer and slower than the Opteron in FP. It's something that your average Slashdot reader commenting on a CPU thread should know about.
Ghz are not directly comparable between different processor architectures, and PowerPC has tended to deliver faster chips at the same Mhz as x86.
Not for the G5. Remember, the G5 is a long-pipeline design like the P4. According to my own benchmarks, my 2.3 GHz G5 is about as fast as a 1.6 GHz Opteron in integer and as a 2.0 GHz Opteron in floating-point. That means that a 1.5 GHz G5 won't be a match for the slowest Pentium-M.
a) strip out components
What is there left to strip out? All of Apple's machines ship pretty bare-bones (512MB of RAM, etc), aside from the optical drive.
b) use cheaper components
Again, what could they make cheaper? My PowerMac uses Kingston RAM and the absolute cheapest 250GB SATA drive you can find. It uses a sub-$100 graphics card (on a $2700 machine). Yeah, they could use a cheaper case, but then the thing would be a Dell outside as well as inside.
Yes, because we know doing things the way we're used to is the most efficient way of doing them. Geez.
I have no idea why everyone is so damn enamoured about the PA Semi chip. Even their theoretical 2007 SPEC scores kinda suck. Their integer performance is so low, they don't make Athlon64s that slow!
Generally, it's so you can have a document and whatever resoures you're using to type that document on screen at the same time. I almost never need to see more than a few lines at a time of the document I'm typing, but I always have several windows that I need to figure out what to type.
Morality is bunk. There are rights, and that's it. Morality has nothing to do with any of it. I refuse to have a conversation on such a nebulous, subjective, and frankly childish concept. When you're ready to talk in adult terms, well, you know where to find me.
Are we talking about the same thing? I'm not suggesting that people go out and downloading OS X. I'm talking about going to the store, buying a copy of Tiger, and installing it on your no-name PC. I don't think there is anything wrong with that. I cannot, on principle, buy into the idea that a company has any right to tell you what to do with their product (assuming you abide the law, that is). If they don't want people to install OS X on no-name PCs, they shouldn't sell OS X as a seperate product.
Eh. It's more like: if I buy a replacement Aztec transmission from the OEM and stick it in my home-made go-cart, is that okay? As far as I'm concerned, yeah, it is!
Opps. I guess the G5 does things differently than the Opteron. On the Opteron, you need a 64-bit kernel to run 64-bit binaries. Well, I stand corrected.
You can't mix 32-bit code and 64-bit code in the same process. You can have two processes, a 32-bit GUI and a 64-bit backend, but then you've just basically got two seperate apps communicating with each other.
The PII was always a better CPU all-around than the G4. The G4's floating-point performance was never very good, unless you could use AltiVec (which wasn't all the time). The G5 was never a leap over x86. From the beginning, unless you could use AltiVec (and a lot of stuff can't), it was about comparable clock-for-clock to an Opteron. When the 2.0 debuted, Opteron was at 1.8GHz, so the G5 was faster, but it was never a leap. Since then, the G5 hasn't got much faster (the 970MP is 2.5GHz), and neither has the Opteron (the dual-cores are at 2.4GHz).
It doesn't matter if they are moving to Intel instead of AMD. The OP claimed that PowerPC was efficient. My point was that the current major desktop/workstation implementations (G4 and G5), aren't.
LOL. Some Mac using moderator trying to silence the truth.
To appease the Mac faithful. I'd like to point out that I like the software on the machine very much. It's got very good Lisp development tools. However, hardware-wise, I have a much nicer Athlon X2 machine right next to it. It's (much) quieter, faster, uses higher-quality parts (name-brand RAM, Seagate hard drive), and fairly good looking to boot. Oh, and its $500 cheaper, and then only because I got an ADC discount.
Wow. You managed to regurgitate something without actually understanding it. There is no way the OS would be able to run 64-bit applications without being compiled for 64-bits. On Tiger (different from Panther), which can run 64-bit apps, the kernel is compiled as 64-bit code. Then, there are two versions of a couple of the libraries (System and Accelerate), one 32-bit and one 64-bit. What's missing is 64-bit versions of stuff like Quartz or Cocoa, which means that 64-bit apps are basically limited to the command line.
Bullshit. I remember when the G4 vs PII debate was particularly hot in the mid/late 1990s. The mantra there was that PowerPC was better, not about performance-per-watt. Well, fancy multicore oomph-per-watt chip that everyone is salvating is a warmed over version of the PII that the Mac folks found so entertaining.
With regards to CPUs, the Mac folks were, to put it simply, wrong. The G4 was never better than the PII, and the G5 was overtaken by the Opteron as soon as it hit 2.0GHz.
The dual core PowerPC brings IBM to parity with year-old AMD hardware. I have a dual-core 2.3, and a dual 2.0Ghz Opteron would be entirely competitive (win in integer, win most floating-point, lose some media-processing). The dual core 2.2 Athlon64 in the machine next to it is a step up in almost every benchmark I've tried.
You don't remember the merciless derision of the Pentium II? The "The G4 is faster than the fastest PII"? It is so deeply ironic that the Yonah chip that will replace the G4 has almost the exact same execution core as the PII that Mac users made so much fun of.
The point is that the 5% of your peripherals that are supported are very common. Looking at the Intel HCL, I know I could easily dig up several of those cards (lying around the house). Most onboard AC/97 soundcards seem to be supported, which is what is used on the Mac anyway. The only sticky point is video (only Intel 900GMA cards are accelerated), and perhaps SATA (nForce4 SATA isn't supported, most PATA controllers are). Firewire and USB are standard EHCI and OCHI, so that's all good. What more does your average user have?
Robust classy hardware my ass. I just got one of the new PowerMacs. Yes, its a beautifully-designed case, but the cooling system is loud, it uses a crappy (and loud) Western Digital harddrive, Delta fans with audible bearing noise, and cheap generic Kingston RAM.
These aren't the Mac's of yore that came with SCSI and the like onboard. These are a fairly generic Dell machine in a fancy aluminum case.
The G5, at least, isn't that efficient. I just bought a brand new PowerMac G5 (dual core 2.3GHz). It's certainly a fast machine, but for almost everything I do, its slower than the 2.2Ghz dual-core Athlon X2 that's sitting next to it. For compiling code, it's about 70% as fast as the X2 system. For SciMark, it ranges from 95% as fast (for the small in-cache dataset), to 80% as fast (for the large in-memory dataset). For nbench,if you leave out one really awful score that's probably the result of a bad compiler optimization, its about 80% as fast. These were all done with GCC 4.0, of course. The 970MP SPEC benchmarks suggest that if I used XLC (and EkoPath on the X2 to be fair), I could probably get it to be 90% as fast in integer as the X2 and 25% faster in floating-point, but considering those scores is entirely an act of intellectual mastrubation, since most stuff on OS X is compiled with GCC or CodeWarrior anyway.
Of course, I love the machine to death, because of OS X, but the way I see it, Apple is going to gain a good deal of performance by moving to x86.
GNUStep is based on OpenStep, the same programming framework OS X's Cocoa is based on. The biggest difference between Cocoa and GNUStep is that the former has had a once-over by all of Apple's artists :)