Slashdot Mirror
G5 vs. x86 and Mac OS X vs. Linux
demonbug writes "Anandtech has an article up comparing performance of dual G5s to AMD Opteron and Intel Xeon workstations. The article also takes a look at performance under Mac OS X versus Linux. It provides an interesting look at some of the strengths and weaknesses of the different CPUs." From the article: "This article is written solely from the frustration that I could not get a clear picture on what the G5 and Mac OS X are capable of. So, be warned; this is not an all-round review. It is definitely the worst buyer's guide that you can imagine. This article cares about speed, performance, and nothing else! No comments on how well designed the internals are, no elaborate discussions about user friendliness, out-of-the-box experience and other subjective subjects. But we think that you should have a decent insight to where the G5/Mac OS X combination positions itself when compared to the Intel & AMD world at the end of this article."
Wow, double flamewar. :)
Slashdot editors are impressive
I have discovered a truly marvelous proof of killer sig, which this margin is too narrow to contain.
Linux 2.6.5 - That's rather outdated... Maybe more recent kernel snapshots offer better performance in some regards?
:%s/Open Source/Free Software/g
YTARY!
A Homos vs Dirty Hippies throwdown.
Comparing OS X to Linux is like comparing a brand new Lamborghini to an old Charger. Yea, they may both be fast, but only one is refined, and only a certain breed of people are going to pick the Charger over the Lambo, given the option.
:wq here and 'sudo' there just to get a damn X server running...
That's what gets me about Linux. OS X is the power of a BSD with REFINEMENT. Linux? If I have never have to use vi to set up a simple routing configuration again, it will be WAY too soon. If I can't point and click my way to a basic setup, it's not a useful system, and comparing it to something that can do that is ridiculous. I shouldn't have to 'alias' this and 'rm' that and
And, of course, like the guy in the Charger, the Linux people will extoll the limited virtues of their decision while gleefully ignoring the fact that they're still driving around in something that smells like pee.
Alito: A vote for Alito is a punch in the eye to put that bitch back in her place!
Anyway..here's the article summary:So, forget OS X in the server room, but have fun if you want a desktop OS.
This comparison is flawed. A more direct comparison that would have resulted in better information would have been Mac/OS X vs. x86/BSD.
What performance is he measuring? The hardware or the OS? Comparing both with no baseline control for each is about as informative as pulling numbers out of my ass.
Do not taunt Happy-Fun Ball
It's difficult to take the article seriously when the author has the spelling of a 4th grader and isn't technical enough to use a spell-checker. How hard would that have been? Doesn't anyone proof their work anymore?
Darwin vs. Linux on PPC!! This is a more useful comparison, IMHO, than Linux on a P4 and/or AMD vs. Darwin. Then you can better gauge the kernel latencies, etc. Are there any differences in Mac OS X Server's kernel? The article concludes Mac OS X is ok for Desktop use, but not for server use. I found this article disappointing.
The Macs just aren't that fast.
At 2 GHz a G5 barely keeps pace with my 2.1GHz 3 year old Athlon - using the byte portable benchmarks.
The G5 woops when it comes to floating point, and stays just behind in everything else. AMD of course takes top honors in almost everything. The find out that OS X kernel doesn't do so well on the server when it comes to multiple threads created while using MySQL and other possible open source software, so they conclude OS X a good desktop, but Linux is better on the Server. They will look into Linux on PPC to see which is better next time, PPC or x86 when it comes to a Linux server.
-----BEGIN PGP SIGNATURE-----
12345
-----END PGP SIGNATURE-----
I guess that the first 50 posts above 1 will be useful, the next 50 will start with the "ford pinto to mercedes" comparison (all modded up Insightful or Interesting), and from there it will just go downhill. I also predict 17 references to the "i've been sitting at my 6300 which should be a much better machine, waiting for this file transfer..." and at least 3 "BSD is Dying" posts.
Tell me how well i do.
Hikeeba!
do() || do_not();
How about OpenDarwin x86 vs. Mac OS X on Apple Hardware?
How about Linux on x86 vs. LinuxPPC on Apple Hardware?
jeesh
Anandtech has an article up comparing performance of dual G5s to AMD Opteron and Intel Xeon workstations.
Ok, Rule #1 - its a performance comparison...
It is definitely the worst buyer's guide that you can imagine. This article cares about speed, performance, and nothing else!
Calm down, did we forget Rule #1 already?
No comments on how well designed the internals are, no elaborate discussions about user friendliness, out-of-the-box experience and other subjective subjects.
OK... Rule #2, no more posting news for you.
I wonder if he uses a mac or pc....
are opterons are super super fast and AMD kindly, and without NDAs, provides technical documentation on them. that's why i buy them
vodka, straight up, thank you!
I'm not a Mac Zealot, lets start with that.
But they are running a test and are identifying the thread creation as being really slow on the Mac and that that is the cause for the Mac's slow performance on the MySQL test.
Come now, if you are running software that is slow because you are creating threads all the time then you need to change software.
Use some kind of threadpool and *kaping*, problem is gone.
This is more revealing for MySQL than it is about Mac OS X.
The Internet is full. Go Away!!!
Until I saw the blatanly placed & scantily clad woman with the words "Root Me" written with MS Paint on the desktop.
Read the rest of the paragraph... They say they're just parroting Apple's marketing. The rest of the article tells a much more unbiased story.
Don't forget the
"The people who buy Macs are creative professionals" partyline that we've been hearing since Joel was still on the S.O.L.
do() || do_not();
"he PowerPc 970FX is a very wide, deeply pipelined superscalar monster chip, with excellent Branch prediction and fantastic features for streaming applications. And let us not forget the two parallel FPUs and the SIMD Altivec unit, which can process up to 4 calculations per clock cycle."
The PPC970FX/G5 looks really hot, even up against Intel and AMD's top CPUs. I'd love to see such a direct comparison as this report with an extra couple of columns for nVidia and ATI's top-end GPUs. Sure, they don't run Darwin or Linux (yet), but as GPGPU gains momentum, I'd like to see what kinds of horsepower gains are available for GPGPU designs, vs. RISC designs. Then I can pick the HW best suited for the task, perhaps even in the same machine.
--
make install -not war
Did you read any further? He clearly states he is parroting Apple's marketing.
They end up coming to the conclusion a G5 is not a good server CPU, but fail to do a balanced test to see if the issue is OS X or the CPU. They should clearly have tested:
Linux x86 vs. Linux PPC
OpenDarwin x86 vs. Mac OS X
The entire paragraph reads
"It is a professional 64 bit Dream machine with supersonic speed! It is beautiful. It is about the ultimate user friendliness. It is about a lifestyle. It is a class apart. You guessed it - I am parroting Apple's marketing."
And... you can guess where it is headed after that... The article goes on to slam Apple.
The Internet is full. Go Away!!!
I read through the whole comparison/review. The article points out that the main factor that MySQL is slow on OS X is how threads are handled in darwin. It's a speculation based on good observations. However, I think that the author should have done a more controlled test to prove his point, such as running yellowdog linux and OS X on identical hardware to compare MySQL performance. Instead, the mahcines that ran linux were opteron and xeon machines, which made it hard to tell whether the hardware or the kernel contributed more to the performance difference.
MySQL runs just fine on the BSDs, Linux, and even Windows. Every project on the face of the planet that uses threads has to be re-written for the sake of Darwin/OS X?
Don't forget five complaints about folks not reeading TFA, at least one list of steps to "Profit!", and three comments about old people in Korea.
Oh. And somewhere, someone is thinking this is all Microsoft's fault.
I'm not tense. I'm just terribly, terribly, alert.
As some people have pointed out (but not completely), you should be comparing:
Linux forks 5 times faster than BSD, but that's been known for years. You didn't need a new benchmark/ad for that. Finally, the article doesn't have a benchmark that uses Altivec to its full potential, so it might be a hack piece as well.
"Thirdly, hardcore gamers are not the ones buying Apples, but rather, creative professionals.
So, we focus on workstation and server applications..."
How could anyone who has ever met a "creative professional" think they care about "workstation and server applications" like MySQL and Apache??
Sorry, guys, but being a sysadmin does not make you a "creative professional..."
-- Mark
I happened to be passing London today and popped into Apple's flagship store. All very nice, but I idly hit F12 in order to bring up dashboard on one of the Powerbooks - 15" or 17" I think - and it was sloooooow.
Now, this could be down to a dozen factors, and it only slightly deflated my ambition to own a nice shiny mac - perhaps a macmini - but it wasn't a good thing, that's for sure.
~~~~~ BigLig2? You mean there's another one of me?
... Will we see PCIe @ WWDC? Will ATI even release an AGP card that accelerates H.264?
Steve, AGP is a dead end.
"I am no operating system expert, but with the data that we have today, I think that a PowerPC optimised Linux such as Yellow Dog is a better idea for the Xserve than Mac OS X server. "
I'd like to see these benchmarks rerun with the G5 running the same OS as the other CPUs: Yellow Dog + kernel v2.6.5. The Darwin vs Linux competition makes deriving real info about just the hardware impossible, though an interesting aspect of the review.
--
make install -not war
At the presentation, they mentioned the G5's potential, but noted that it was closer to the Intel architecture in the sense that each CPU shared a memory controller (but it's not hampered by the bus). The Opteron's HyperTransport model is simply more scalable. Apple got the point, but whether they will address this deficiency in their Xserves (particularly the Cluster Nodes, where it makes sense for massively parallel systems) remains to be seen. All I know is that *I* love the performance of our G5 systems, Xserve and desktop alike.
There exists no way of exchanging information without making judgments. --Bene Gesserit Axiom
I'm not sure I trust the benchmarks. They proudly show their benchmarking code which has a blatant mistake in it. They spotted it, but wrongfilly blamed VB.NET and threads for it and then proudly announced they had to write a super-duper workaround for it to make it work.
Worthy of a WTF.
If you'd like to see it for yourself: look here
Wouldn't it have been better to use compilers that are tuned for each platform? Say, Intel's compilers for the x86 systems, and IBM's compilers for the PPC systems. These compilers could perform better prefetching, for example, and you might get a more accurate idea of what the systems could do with binaries that are tuned for that system.
Most of the benchmark data is bottlenecked by gcc, as the review mentions. That's fair, because that's what so many of us use to compile on these kinds of platforms. But I do think that Apple would do well to throw some of their programmers at the GCC project, at least adding their expertise to some of the Altivec modules. It would show off their platform, and return some value to the gcc project surely used extensively by Apple.
--
make install -not war
when I saw they were going to benchmark Apache. Maybe they used Gigabyte ethernet, but otherwise, there's nothing interesting in the comparision with a 486.
...so you don't have to read it: Apple = slow, Linux = the shit.
;)
Now, had they gone x86 BSD on the G5 versus OSX on the same G5 then that would be a bit different. But nobody ever does that. I'm glad I'm not the only one who saw that "comparison" was flawed a bit. But it is still nice to see the myth of Apple being all-powerful being demolished by cruel charts.
Really comparing Linux vs Mac OS X is very much pointless... Now, Darwin (XNU) vs Linux would be more interesting... especially Darwin running Xorg and KDE (or GNOME) vs Linux running Xorg running KDE (or GNOME)... that would be more fair, I believe.
Insanity: doing the same thing over and over again and expecting different results.
Another thing is I've used the Mach api and it shows different Mach thread ids for the Posix threads I create with PTHREAD_SCOPE_SYSTEM so I'm a little mystified as to why they claim only one Mach thread is being used. A lot of the Posix thread libraries used to default to user threads with the default PTHREAD_SCOPE_PROCESS.
but chances are "energy waste" didn't come up...
If my AMD64 can get me 30fps in ut2k4, compile the kernel in under a minute or two and render porn at acceptable jerk rates...
WTF DO I CARE!
Its doing all this while taking a quarter the power of the G5. All I know is my AMD64 doesn't have a windtunnel in the case to keep from melting through the board.
Efficiency people, not raw numbers.
If you can do X amount of work with Y less power in a comprable amount of time... that's a good thing as Y increases.
Tom
Someday, I'll have a real sig.
I would really like to see a comparison using better compilers though, Intel on x86 and IBM on Power.
Personally, I don't know any people using MySQL who aren't both creative and professional. But maybe I just don't get out enough....
Linux? If I have never have to use vi to set up a simple routing configuration again, it will be WAY too soon.
vi? Where do you *have* to use vi? Is it meant to stand for [any plaintext editor]?
Granted, editing text configs *can* be less friendly in certain situations (it can also be a lot more flexible and straightforward); but I guess invoking the name of vi (which has a reputation for being arcane) makes textual config sound more complex than it actually is.
I use and like vi in preference to Emacs (vi IMHO is less friendly on the surface, but more straightforward than Emacs once you know the basic keys). BUT.... we're discussing its reputation here, and it seems this is being exploited to make your case.
Don't like vi? Use a different editor, but don't try to rub vi's alleged arcaneness off onto text editing in general.
"Slashdot - News and Chat Sites Deviant". (Click "homepage" link above for details).
I think they missed the most interesting statistics possible when comparing Mac OS X against Linux... how they rate when both are run on the same architechure. Since Linux PPC is available, it should provide a method of getting a more "Apples" to "Apples" comparison.
Also, given that both Apache and MySQL should be functional with just Darwin (without all the other Mac OS X junk), they could have also done a comparison using Darwin x86.
Instead, they left Darwin tied specifically to the PPC and Linux tied specifically to the x86. Personally, I think the results of Linux PPC vs. Linux x86 would have been interesting. There has to be something worth while about how Linux performs on the PPC if IBM is willing to state that it will replace AIX by the end of the decade and also award prize money to those that port to it.
It would have also been nice if they could have borrowed a full IBM p-Series system to also compare.
Sorry, but this completely invalidates any metric including the word "performance".
IBM's C compiler should be used on the Mac side (OSX now uses GCC 4.x BTW), Intels C compiler on the AMD64 side.
Do that, and try again.
Repeat after me - "GCC is crossplatform - performance sucks on all eequally".
- Adam L. Beberg - The Cosm Project - http://www.mithral.com/
OSX's posix thread performance does not seem very impressive, but:
1) General performance comparisons should not be made on the basis of a single application.
2) High performance server software uses more sophisticated thread pooling
I use Oracle running on an XServe under MacOSX and its performance is very impressive.
As someone who's hand-optimized PPC code for almost a decade, I can easily point to a fatal flaw in this benchmark. Even after all of Apple's recent tweaking, GCC's PPC codegen is still horribly flawed, and greatly inferior to its x86 codegen. The compiler of choice for high-performance PPC codegen is still - and by far - Metrowerks Codewarrior, with IBM's xlC sitting somewhere in the middle. This is not about automatic vectorization (particularly on the 970, where the Altivec unit is afterthought-ish in many respects.) It's about day-to-day necessities like vector coloring and scheduling. I routinely get 15-30% faster code from CW than I do from gcc on hand-optimized C code in complex performance-sensitive inner loops.
So, take the results and tackle on a 15-30% confidence interval due to their compiler choice. Or throw them out altogether.
I want to hear about the techniques used by "the major database vendors" to deal with the thread blocking issue. Maybe programs like MySQL can take advantage of these enhancements, too.
This doesn't appear flattering for Apple, but it's apparent that they have been scrambling to get the user experience right in OSX, at the expense of sub-optimal kernel development. Hopefully they will be able to refocus on the kernel and the compiler and get the performance up to what Linux people expect. Thread blocking will become much more of an issue as multi-core CPUs become mainstream.
Linux is a good example of what can happen here. They got crummy benchmarks, the kernel guys identified the bottlenecks, experiments were written to overcome the bottlenecks, and eventually the fixes made it into the kernel and everyone benefits. Notice how Microsoft doesn't brag about performance any more?
Sunshine is the best disinfectant. - Tip O'Neill
I would like to see a comparison of hardware only. Linux on PowerPC vs Opteron
And in some cases, you really do want to do that. 80% of the time, you'll be pounding threepenny nails into pine blocks. So this particular benchmark is going to tell you which is the best hammer for you.
It's worth noting that all of these benchmark reports must be taken with a grain of salt, but more importantly, anyone planning to make decisions based on them should get a really strong idea of what they plan to do, then use them to figure out which hardware/OS/software combo will give them the best performance for their primary task.
If you're going to do all sorts of stuff, then a "general purpose" benchmark may be for you. But if you're mostly doing 3D modeling and rendering with Maya, then what is really useful is to know which processor/OS combo ekes out the highest scores on Maya tasks, not which processor is considered the best "workstation" based on benchmarks for software you don't use.
Every set of benchmarks makes it easier to eventually hunt down information that's relevant to the task you need benched, so I don't want to discourage journalists from posting these kinds of articles. In fact, I'd like to see more. But I'd like to see more application-based tests and less results based around arbitrary measurements like floating point operations.
- Greg
Start a happiness pandemic
There's no OpenLinux, FreeLinux or NetLinux. I think that proves that *BSD forks at least 3 times faster than Linux.
Those Apple things really do suck.
My investigation has uncovered a series of hippy drum circles arranged in a flower shaped pattern on this map (that you cannot see).
My research clearly shows that we are very close to the start of a hippy music festival. It could begin at almost any moment. In fact it may already be too late.
Appended to the end of comments you post. 120 chars.
everyone keeps complaining that these tests were not fair because it wasn't linux to linux or bsd to bsd etc etc. but isnt the point of a test like this to compare the typical working packages that one would have in an environment? people are going to buy an apple machine to have OSX. people are going to run linux on x86. by all reasonable accounts, apple does not want us to seperate their software from hardware. it is a mac. end of line. isnt this is a fair comparison? the computers as you would find them in real life, not finagled to get differnt specs on operating systems that most likely are not going to be used, like ppc linux? (i know i am sure someone out there uses is.. but really). the computer here is a whole product, and for servers macs apparently fall short.
Hmmm... where did you read that? Even in the fefe test, freebsd and linux have very similar performance characteristics, and that's a two year old benchmark.Quote:
"FreeBSD looks like it would scale O(1) if I could create more processes with it, but as long as I can't confirm it, I can only give it the second place.
"
Check the graphs ... and the corrections (author did not read man tuning, sysctl, the handbook... well, the documentation in general at first, so did not know how to set kern.maxproc).
You're welcome (to this information) :)
Even the FPU performance is questionable-
They state that the simple flops.c is designed specifically to isolate and test the FPU-
Ok- then they go on and do this:
The really funny thing is that the new Xeon Irwindale performed better when we disabled support for the SSE-2, and used the "- mfpmath=387" option. It seems that the GCC compiler makes a real mess when it tries to optimise for the SSE-2 instructions. One can, of course, use the Intel compiler, which produces code that is up to twice as fast. But the use of the special Intel compiler isn't widespread in the real world.
Well- so WTF is really being tested here?
"twice as fast"?
This then becomes more a test of gcc's optimization for a particular proc does it not?
Does Apple have their auto-vectorizing modded gcc out yet?
What's AMD's compiler technology looking like?
Comparing 3 different CPUs in such a rudimentary manner with only a single compiler, without taking in to account the compiler's own strengths and weaknesses is kinda pointless IMO...
I guess the next step would be to hand-optimize a short FPU routine for each differnt CPU...
I still love my Mac, but c'mon...
I browse at +5 Flamebait- moderation for all or moderation for none.
"If I can't point and click my way to a basic setup, it's not a useful system"
:)
Whoa! Homer Simpson reads Slashdot!
Ah, another post about how there needs to be some more benchmarks to hide OS X getting thrashed.
I keep thinking Darwin is available to people who are willing to pull the code and compile it...
Lamborghinis are ugly and impractical. The Charger is beautiful and impractical. Easy decision. You can take some of the money left over and reupholster it if you are offended by passengers suffering incontinence from 440/6-pack acceleration trauma.
If your application is having its performance destroyed by a 2-5x multiplier in thread creation time, this application is not written very well.
Why I remember when your computer came with circuit diagrams, machine language opcodes and technical details about what every region of ROM memory did! And pretty much every manual came with an ANSI chart, whether the product actually required you to have that information or not! Why I oughta... get off my lawn you damn kids!
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
My roommate and I work together developing websites. He runs a powerbook and I've got Gentoo Linux running on a Dell 600m. The main advantage that he has is a reliable PC emulator on which he can test apps for IE. The real tie-breaker though is what KDE provides me in a text editor.
Far from vi, Kate (the editor embedded in most KDE apps) has an unbelievable ability to make work go faster. It has a native understanding of html/css tags way better than Dreamweaver or the like, supports Perl, Javascript, and PHP with inline variable-name completion and auto-syntax. I can't name all the things it does well, but a system running the latest KDE 3.4.1 can take on an OSX system any day in refined web development.
World Changing - News for Humans, Stuff about our planet
Mod parent troll, and advise him to get try a recent Mandrake release.
I'll criticize the entire article after reading it, but rigt after the first page I can tell this is a crap test badly done, here's why:
They say Macs are bought by creative profesisonnal so they will test open source solution such as MySQL and Apache!!!???
Since when those are the tools of creative pros?
They compare Xeons and Opterons to the G5, it should be Athlon64 and P4 against the G5, lets compare what target the same type of user base, else why not simply make a test that pit the G5 against a cray machine or Blue Gene?... I mean I know the Xeon isn't that powerfull and I don't have the test result but the base config is flawed in terms of comparison. Even if it wouldn't I mean Apache... common, at least pit the Gimp against the Gimp if you wanna look like you don't give the test gift-wrapped to Linux and x86, even then it would be flawed comparison since Gimp doesnt get the same amount of devellopement on G5 than on x86... and people say Photoshop benchmarking is flawed... geez at least the app truly IS optimized for both patforms...
Anyway on to the reading of the second page...
I usually like their reviews, but this is terrible - it's a bunch of tests that skip anything that would make a useful comparison even though it keeps wandering around and doing more stuff.
1) They should've compared several compilers. I suspect that gcc on OSX is much less optimized on i386. They showed that gcc doesn't speak vector almost at all. I also suspect IBM or someone has a better optimized G5 compiler. While I suspect there's no way to make a really fair comparison, giving us some idea of the range of compiler difference would've helped us know how singificant it might be - and it's a lot.
2) If, like they say, they're trying to compare the CPUs, they should've compared Linux on G5. They basically say as much at the end of the review - that they were really comparing OSX to Linux 2.6 across different platforms. I would've liked to see 2.4, 2.6 & OSX all on G5.
3) If OSX's has big threading disadvantages because of it's similarity to BSD4, they should run a benchmark compared to BSD4 - and another one to BSD5, which will presumeably give something of a "view into the future" of what OSX's performance will soon be.
Looking for freelance Actionscript (Flash/Flex) or ColdFusion work and/or freelance developers. Email me, put Slashdot
Here's a similar comparision/benchmark. And Tech Report posts a summary/commentary of the tests:
PC World has posted some benchmark results suggesting that Apple's Power Mac G5 isn't the world's fastest desktop PC after all. The PC World tests compare dual G5 systems to a Pentium 4 3.2GHz, Athlon 64 3200+, Athlon 64 FX-51, and Opteron 246, and the results make Apple's claim to the desktop performance crown look rather foolish. The dual 2GHz Power Mac G5 can't even manage a win in Photoshop, where the dual Opteron system turns in the fastest performance.
I suggest you check out the benchmark results.
Let's see the Apple zealots explain this one. Will it be "OS X is much more than just kernel threading - it's much prettier!", or will it be "it's not a fair test, even though it's comparing the native OS on the Mac to Linux."
Come on. Just admit it. OS X is worse than Linux in server rooms, and the G5 can't hold its own against the latest generation x86s. There is a reason why the x86 is the dominant architecture, and why Linux and Windows are considered the two serious competitors for servers. The AIM alliance has lost just about every round of the processor wars to x86, and Apple's still got work to do if they want the Mac OS to be taken seriously in fields beyond those "creative professionals."
They can start by rebuilding the threading system.
You forgot to mention that you tried to copy a 17 Meg file from one folder on the hard drive to another folder ...
The largest performance penalty was found in the usage of fork().
But there are also different (often more efficient) ways to create threads.
fork() often copies the whole C-stack.. (and sometimes environment variables too).
But a thread-object like in some C++ implementations and some OS's start a cleaner stack.
Likely that would really make the difference.
have you READ the article? It was not apologist at all.
Just because Thurott didn't come in and say that the Mac was a worthless POS that only a pink Miata driver would use...
Ok that was funny
If I had mod points, this post would be one higher.
"But this one goes to eleven! That's one more!"
Well, I'm very very drunk now, but at the time I had only had a pint of carlsberg and probably the crapest cheese and ham toastie in the world - (c) the mitre pub in hatton garden - so I definitely hist F12, not Fn+F12
~~~~~ BigLig2? You mean there's another one of me?
As a matter of fact, energy waste did come up.
But I guess who cares? This is slashdot, isn't it?
We suffer more in our imagination than in reality. - Seneca
I own both a dual G5 and an AMD64 system, with both Windows 2003 and Linux running on it. Have fun arguing about which platform is best.
And since you're curious, I use my G5 most of the time.
it's a good thing the linux boxes crushed the macs in server tests.
The reviewer is using 3.3.3! Iirc apple put alot of work that was implemented in 3.4/3.5/4.0
The linux kernel version does not make as much difference(same version throughout), even though it is upto 18/20% slower in some tasks than earlier and later kernels (Search slashdot for Linus asking for constant monitoring of performance as there are very large fluctuations in performance between different kernel releases!)
Cudos to this guy going through all the trouble so potential PPC Users don't have to. :-) .
The G5s exeptionally bad real life performance with Apache and MySQL is an eye opener for me, as I am considering XServe as a plattform for the stuff I deliver.
However, the G5s highest LW Raytracing performance is comforting, as I just bought an LW 8 license for Mac OS X the other week
Very informative article indeed.
We suffer more in our imagination than in reality. - Seneca
Jeez. I've got an idea. Why don't I take this opportunity to tell you that you should be using the One True Bracing style when coding so we can really flame on into useless obscurity.
Run "Top" so that it refreshes every second. (I believe that's default.)
.
On the Mac Mini, it uses 7% CPU
I have a friend who just got a dual 2.0 G5. Had him run Top and I was stunned at how much CPU time it took. I didn't write down the exact number so I'm not going to mention it here.
On any x86 system I've used, Top barely uses any CPU
Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.
Like everyone's already said: Linux/PPC would have been a good thing to add in. Like someone else mentioned, Apache 2.x and a PostgreSQL database would have been good tests along with the MySQL+Apache 1.3 ones. I haven't seen anyone mention the gcc compiler version.(they used 3.3) 3.4 is more wholesomely made. 4.0 is the latest, woulda been interesting to see if that made any significant changes(and doesn't panther use that defaultly?) The test lacks the ability to show whether the issues in server based MacOSX are CPU based or OS based. Linux/PPC would have been helpful.
It really is a shame that Apple didn't build their new OS around Linux rather than BSD/Mach. It would have been great for both OS X and for Linux.
I know it's more complicated than that (eg. it might have been bad for Apple long-term if they helped Linux), but it's a nice thought anyway.
You forgot one:
PPC Darwin vs. x86 Darwin
They are cooperating (freebsd and apple) it seems on many issues.
It would probably be more accurate to say that the FreeBSD developers are hacking a whole bunch of code and Apple is incorporating it into OS X / Darwin.
I've yet to see any indication that Apple is funding or helping FreeBSD in any way. While that is their perogative under the BSD license, it is a bit disingenious.
Why didn't they compare the fastest G5 (2.7ghz) to the fastest Opteron (252, 2.6ghz)?
You're not the first person to ask this, and Johan responded to the same thing over at Ace's Hardware (where he was before heading over to Anand). Basically you can only do so much while having a reasonable article:
Johan's response
Sorry, I'm unfamiliar with this app???
I don't want to sound like a Apple Apologist but...
Really all this caring about differences in performance is becoming a rather old argument, and is becoming less and less valid every day. Sure back in the early 90s the speed difference of say 5% was important it would be the difference from a slow system to a smooth system. But now most tasks can be run well on a lot of systems. I am using a 667mhz Powerbook and still most of my tasks and buy all specs even with PCs the same age of the system it will loose on all benchmarks. But it still does what I want it to do for most of the tasks I want to do and it does it just as well as systems 5 times its speed. Because for most of the work I do which is mostly typing/programming, networking and some light graphics work. I rarely notice any bottleneck except for me.
So unless you are using some tools where you can notices times where a 5% speed increase can save you 20 minutes on your task. Most of the time you are just as better off buying a system that can take a lot of ram a big harddrive and is the color you want.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
I didn't realize that homosexuals and hippies were a race of their own...
I love OS X, but there are some definite rough edges in Tiger, and unfortunately those are Spotlight and Dashboard, the two most anticipated features, or at least in my mind. Dashboard takes sometimes up to 15 seconds to come up and fill in the info on my 12" 867 with 640mb memory. Subsequent cycles through F12 work a lot faster though, unless I do a bunch of other work and let it sit for a while, so I think this is a memory issue. Spotlight also has some memory issues. I moved about 23,000 files from my laptop to another computer and it was *amazing* how much faster it ran. I can only attribute the speed increase to spotlight not having to hold info about those extra 23,000 files. All in all though, I still think OS X is much better than XP. (Digression: I'm interested to see what Longhorn has to offer, but I'm not holding my breath, especially since even if it smoked OS X in eyecandy it still wouldn't be *nix.)
Your subject, comparing apples and oranges, is invalid:
m e1/v1i3/air-1-3-apples.html
http://www.improbable.com/airchives/paperair/volu
http://www.darwinmag.com/read/0502/apples.html
Space and Computers.
This is what I'm getting on my DP 2.5. I typed in "top -RF" as noted in the other post.
I once had a print job get stuck in a process, or something like that, anyways it was eating up almost 100% of just one of my procs. Once I figured it out, I killed it and haven't had that issues since then. Maybe your friends DP 2.0 had something running in the background like that?
Why limit your options? There is software that runs on one and not the other, or runs so much better on one than the other, but if you own both you can't complain.
And if we're keeping score, I too use my Mac most of the time.
annoucement at the WWDC in SF....The dual core mobile Intel chips rock and this will ensure a future it seems IBM can't deliver!! WOW if it is true http://news.com.com/Apple+to+ditch+IBM%2C+switch+t o+Intel+chips/2100-1006_3-5731398.html?tag=nefd.le de
OS X on Dell here is comes!
"However, the G5s highest LW Raytracing performance is comforting, as I just bought an LW 8 license for Mac OS X the other week :-)"
The G5 wins in floating point performance, so that's easy to believe.
The server performance is dependant on the kernel, and Linux gets an easy win there.
I rarely criticize things I don't care about.
Homo means Apple for some reason. Hippy means Linux for some other reason.
Transcend Humanity. Please.
User level threads aren't necessarily slow, it depends on how well they are implemented, how much support the underlying system gives and what you are trying to do... User level threading packages can switch threads *very* cheaply, without going into the kernel at all. They are sometimes limited to only run on one CPU but this is an implementation issue. Purely kernel-based threads (like Linux 2.6's model) have their own problems (eg. scalability with those more expensive thread context switches).
The Acura is a better car. It drives and handles better anyway. I've test driven both, but own neither (multiple times). Now the McLaren F1 was a sports car.
I think it is about time the comparsion between cars and computers is dropped. The part about stripping being down to cheap cars of course is true.
It's not. OS X's threading model is not at all similar to FreeBSD 4.x's all-in-userland threading; the authors of the article were deeply confused on that point.
otherwise, spawning threads would be really important to you on the desktop.
The reason these tests ARE relevant is that the vast majority of users do not run Linux on their Macs, nor do they run BSD on their PCs.
The tests are pitting the common OS on each platform against each other. That is a fine comparison, because it represents the basic choice that people face when they want to choose a platform.
You just have to be careful how you interpret the results. Since neither the hardware nor the software are held in common as a "control" variable, there is no way to compare System A's software against system B's, or System A's hardware against System B's.
The best conclusion you can draw from a system level comparison like this one is that, given the test environment, System A was faster than System B overall.
And in this case, it looks like the G5-OSX combo is "System B"...
Isn't GCC 4 supposed to be slow (for now). Did they use it because it's the only one that compiles 64-bit code?
I'm interested to know just how the threading implementation on MacOS X stands up to others. Anyone with a G5 and some time mind benchmarking with Mac OS X, Linux, and maybe NetBSD? That would settle if it was just OS X using pthreads and pthreads being slow.
The conclusion of this article seems to be that MySQL and Apache don't run well on OS X. While Apache certainly is pervasive, most database servers are Oracle or J2EE based. And I believe both would perform well on XServe or PowerMac G5. Most multithreaded servers use thread pools of a fixed size. During peak loads, they can grow, but generally threads are expensive to create (or assumed to be).
Oracle database typically doesn't use threads in UNIX (they do in Windows). They have two modes: dedicated (1 process per session), or shared (a pool of processes). I can't see how Oracle would be affected by slow thread creation (process creation is generally expensive on most platforms). I know there's been some talk recently on comp.databases.oracle.server that Oracle 10g RAC is very fast on XServe G5 with XServe RAID, with a much better price/performance during informal tests, and that Oracle themselves are starting to adopt Mac hardware for their own data centre (at least the XServe RAID arrays, don't know about G5 yet).
-Stu
I meant to say "most database or application servers" are Oracle or J2EE based.
-Stu
What makes you think PostgreSQL would be faster?
Hey, what's the deal with the article ripping on fusion cuisine so much? That style can lead to some really pheonomenal food.
You don't remember OpenLinux? It was brought to you by the fine folks at SCO -- well, Caldera.
what about nt4 performance on G5's vs. windows xp on 1.8 celeron w/ 256mb o ram
i am typing this on such machine(1.8) and damn my 300mhz laptop with 160mb of ram running debian testing whooops it.
let's all get angry!!!!
wooo hooo fuck it!!!!!
we got ignorant lemmings provoking the angry lemmings, but we all just need to be wasted and thrown off the nearest cliff.
you first!!!!!!
bwaha ha hha h ah a hah
bhwhw ahah ha h ahahha
lol.
rotflmao!!!!!!11111
I checking out pc prices for next fiscal year's budget. I notice that Powermacs are the cheapest of the dual processor systems. Pretty much, if Xeon/opteron are better, you will be paying extra for it.
You don't have to be smart to use a Mac, you just have to be smart enough to buy one
No, actually, I remember reading an article linked to /. that compared Linux with Solaris, BSD and Win2k, probably more than 2 years ago though, that showed that BSD was 5 times slower on average at fork(). It's not POA, I just can't find it...
Well, the explanation for this is OSX may have a different forking/threading model than the latest stock FreeBSD. OSX may have started out with way older code than what you can find in FreeBSD today, as would be logical if you were trying to adapt the thing to Mach and didn't want to mess with patching the changes back in until you had something to release.
Some of the benchmarks in the article actually used OS9, by the way, so they were really running around the map with it. They seemed to be more interested in generating controversy than actually getting comparable results.
Say what ?
Cocoa is written in Objective C, which is very much a compiled-to-machine-code language. It has Java bindings as well, but they're not the prime target. Objective C has runtime binding to its objects' methods (selectors) but there's no bytecode involved, as far as I know!
aside: ObjC is a gorgeous language - sufficiently simple that C programmers "just get it" immediately, and without the horrendous baggage that is the cruft of C++. It comes with a nice standard library (the NS... class set) that mirrors the java ones in many respects, but it compiles to almost-as-fast-as-C++ code (the runtime binding takes a toll). About the only thing that puts people off is the odd syntax for method declarations and method calls. Even that becomes second-nature after a while.
Considering the language is so old now, I'm surprised it hasn't caught on more - it does make development very rapid (akin to Java), but without the speed issues...
Simon
Physicists get Hadrons!
eWeek benchmarked the same Xserve and said this:
"Performance-wise, the dual-processor Xserve G5 compares well with Linux-based x86 servers. This is not surprising, considering that Mac OS 10.x is based on FreeBSD UNIX. Using the included Apache server, we ran the Xserve G5 through our standard WebBench tests. It did quite well on the static WebBench test, outperforming a competitor's dual processor x86-64 server running Apache server on SuSe Linux-64."
So who do you beleive? I'll take eWeek over some college guys.
since Apple is apparently ditching PPC for Intel.
So they'll have a crappy OS running on a crappy chip. Glad I sold my AAPL stock...
"Lack of technical competence coupled with the arrogance of power, as usual, leads to no good end."
FreeBSD (especially FreeBSD 4) has always had a slow and unstable implementation of threads. This is why they had to rewrite everything in FreeBSD 5 and in DragonFlyBSD.
No surprise that regardless of the platform, Linux + NPTL has way better MySQL performance.
Please never, ever characterize OS X performance as "BSD" performance, as it is extremely different from Free/Net/OpenBSD (all of which are more similar to Linux than to OS X in design and performance characteristics).
Linux forks 5 times faster than OS X/Darwin, it's about the same as FreeBSD. A quick verification seems to confirm that it should be in the right ballpark (these are on different machines with different performance, but I've tested identical machines before and according to my memory Linux is about 5x OS X and within 20% of FreeBSD - which is faster has varied depending on the specific versions):
OS X 10.4.1 (G5 2.7 GHz): 1600
Linux 2.6.11 (Athlon 64 3200+): 7900
FreeBSD 6-current (Athlon 64 2800+): 6400
k, I don't know whether this "between 2 and 5(!) times slower" stat is true, but even if it is, it should hardly matter. The time it takes to create a thread should be insignificant compared to the time it takes for the thread to do its work (at least, in a good program). Not to mention that in the case of a multithreaded server (like Apache 2), thread pools are used so that less time is spent creating and destroying threads.
The bits on the bus go on and off... on and off... on and off...
yes... all this... but I think they should have added compiler tests for Java as well... I'm curious to see how Linux on the G5 compiling Java would be compared to Tiger or just Darwin doing the same. Also, I think it is obligatory to throw in a few machines running at the same clock speed as the G5... just to see... But why can't benchmarkers get it right? Control control control! If only they could eliminate the variables, and do the right tests, then we could really see what we are looking at!! (uh... you know what I mean)
The Admin and the Engineer
..due to Apple's planned/hinted switch over to an Intel chip.
For every present, there is a past
If not four you're sig, I would not have bothered to point out that Kudos is the correct sperring...
It was a good artical up until the tried to explain the poor server performance with MySQL and Apache. They totally blew it there. For starters, fork and exec have nothing to do with pthreads (and probably have nothing to do with the performance issue at hand either). Also, (as has already been pointed out), each POSIX thread gets its own Mach thread (actually task in MACH terminology) -- the POSIX threads are not implemented in user mode.
What I have heard is bad about OS X threading is that the mach threads use the POSIX mutexes and the POSIX mutexes cause priority inversion problems, but that would be the same with all POSIX thread implementations.
I would guess that the more likely cause of problems is the way I/O drivers are written as mach tasks that run in the kernel process (or was it that all drivers ran in a single kernel thread -- something like that). There might also be a big hammer lock around the entire file system. The downside of messaging systems is that they tend to bottleneck when one thread does too much work. The same as the way a locking system behaves when the locks don't have a fine enough granularity. Neither architecture is really "the answer". Neither architecture makes doing MP easy...
An engineer who ran for Congress. http://herbrobinson.us
How often are you going to find a dual Opteron *desktop*?
> I shouldn't have to 'alias' this and 'rm' that and :wq here and 'sudo' there just to get a damn X server running...
I don't either. I just do `emerge xorg-x11`, wait 2 days and I have an X server!
And if they had used the IBM compilers on PPC, then they would have also gotten a huge increase there as well. I think it would have been better to use the Intel compiler on x86, and the IBM compiler on PPC, and then compare those results. If you're looking to see what the platforms can do, you might as well use the best compilers for those platforms. Seems to me you might get a more meaningful comparison. But, like you said, I might just be wrong again.
Your insight is appreciated; do you have more details?
I nonetheless stand by my point...
If the authors THINK that that's the problem... they should've actually tested it.
Looking for freelance Actionscript (Flash/Flex) or ColdFusion work and/or freelance developers. Email me, put Slashdot
their cheese and pickle toasties are better I think...
The article states that gcc isn't very friendly to the G5 and then give the command they use to compile the test:
gcc -O2 flops.c -o flops
They wonder why few of their tests seem to use the Altivec, yet they leave out the flag -mAltivec, which seems to me the key to getting gcc to work well with the vortex engine. Am I wrong? Shouldn't the command be:
gcc -O2 -mAltivec flops.c -o flops
hmmmm
That's the fly in the ointment. The vast majority of people are easy marks for rigged benchmarks (and marketing types know it all too well).
For instance, a tiny detail like the Mac memory timing is CAS3 whereas the Xeon memory is CAS2.5 slips right by most casual observers and can make fair bit of difference. If that's the way the Macs are always shipped, that's one thing, but if the people doing the benchmark deliberately chose that configuration on purpose to make the Mac look bad then that's quite another.
That's why I prefer having a control for experiments. It makes those sorts of details harder to slip by. If my computer is slow, I want to know why so I can make informed purchases, upgrades or changes.
I agree that most people will just want to know the emotionally charged and page-hit generating "if the Mac is faster," but things change so quickly that the value of the simplistic finding will be of use for a very short time (and override more important user concerns). By the time the public perception accepts a fact, it will no longer be true... So, it is better for people to conclude that benchmarks with simple answers are too prone to be a front for marketing, and that there is no substitute for being informed about what one is getting.
So while everyone's coming out of the woodwork to declare themselves both creative and professional, I figured it was a good time to point out how this term grew out of the marketing of products like the Macintosh. I can't say this is how the term originated, but it's certainly where it's found its niche.
Several years ago, Apple had a problem finding a market for their desktop systems outside the domain of graphic artists and musicians, for whom the brand still had significant cachet.
They pursued a couple of strategies. The first, which most people are aware of, was finding a home market for audiovisual technologies like video editing and music, and bringing their low-end price point down to meet at least the affluent home user's preferences.
The second was to try to find ways to leverage the popularity of Macs among artists to the rest of the business community. "Think Different," while not so effective, was one stab at this. Another was to start referring to their systems' core market not as "artists" but as "creative professionals."
This had a couple of effects -- "creative professional" connotes not just an artist, but a financially successful (or at least self-sustaining) artist, which was a more favorable connotation from Apple's point of view.
Second, though, it was a sufficiently ambiguous term that it invited everyone, even those not in Apple's core market, to redefine themselves as a potential Mac buyer because they were both "creative" and "professional," even though from the context it was clear that Apple was using the term to mean graphic artists, filmmakers, musicians, writers, and publishers
I was not trying to knock sysadmins down a notch by saying "being a sysadmin does not make you a 'creative professional...'," but instead was pointing out that neither Apple (nor the authors of the comparison that this thread talks about) uses the term in a general sense to mean everyone who's a little bit of either.
Apple, of course, would love for people to say "Well, I'm creative, I'm professional, I guess I must be the target market they're talking about!" And that's precisely what several people have done in this thread. And that's exactly why talking about "creative professionals preferring Macs" is powerful marketing.
-- Mark
I have compiled a set of benchmarks for scientific computing applications which are broadly consistent with the posted benchmarks:
http://jsekhon.fas.harvard.edu/macosx/
Y'know, if all you do is watch porn and download bootleg music files in your mom's basement, I'm sure the mac kernel seems fast to you.
I said, "OSX hacks a BSD kernel into a Mach microkernel, and thus performance is nearly as bad as Mach despite the existence of the mature, standardized interfaces of a BSD."
and you said "This is completely wrong."
So, you claim that OSX is NOT comprised of a BSD kernel hacked into a Mach microkernel? And that the performance is worse than Mach? Or maybe that performance of the hybrid kernel is better than a monolith?
Or is there some new meaning of the word "completely" that I haven't heard yet?
I don't get it.