Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Exposing the Anti - Apple bias

Originally posted by billyboy
Oh, can you tell me what tasks I have to run to make my Powerbook lag when browsing the internet? Short of turning it off! :)

I invariably surf the web when waiting for things to be burnt and it seems to work the same as if nothing else was going on in the background. Or has my bias for Mac blinded me in some way? :).
Click to expand...
If you notice I said "while waiting for a DVD to render and burn." Key word render. I should have been more clear; what I basically meant was MPEG2 compression, which will tax the sh*t out of any CPU. I can't speak for how well the Powerbook does at this task because I haven't tried one, but on most PC architectures when there is a single process eating up 100% of the CPU your other tasks seem to slow down noticeably. The Hyperthreading makes it unnoticeable.

Maybe on your Powerbook it is an unnoticeable difference as well when you're doing MPEG2 compression while web browsing, I'm not sure because I haven't tried that (don't really get to try high-end stuff like that in the stores very often).
 
Originally posted by daveL
Curious. Where did you read that the 980 would be dual core? I don't *think* that's accurate, although it's all speculation, at this point. I had heard that the 980 would have twice the integer and fp units, compared to the 970, but a single, possibly improved, Altivec. Anyway, as I understand it, the big difference between the PowerX CPUs and their 9xx counterparts is that the 9xx is *not* dual core but adds Altivec. I guess we'll all know for sure when the specs are published.
Click to expand...
Sorry, I can't remember the source, but it was a posting in this forum somewhere. Someone mentioned that the 980 was based on the IBM Power5 which has dual cores, and it was assumed that the 980 would inherit this characteristic from the Power5.
 
Originally posted by illumin8
Sorry, I can't remember the source, but it was a posting in this forum somewhere. Someone mentioned that the 980 was based on the IBM Power5 which has dual cores, and it was assumed that the 980 would inherit this characteristic from the Power5.
Click to expand...
Power4 is dual core:

IBM launched the Power4, the first dual-core server CPU, in 2001. With the Power5, it now hopes to be the first to blend both multi-core and multithreading technologies.
Click to expand...
Found here:

http://www.us.design-reuse.com/news/news4972.html
 
It provides even more efficiancy than having just 2 real processors. Say you are running 2 threads, you use both processors, but most likely not 100% of each processor, so letting other threads use what is left over will get more work done.
 
Originally posted by illumin8
It's possible I mixed up multithreading technology with multicore. Aren't all processor's multithread capable? Excuse my ignorance, but what is the benefit of having a processor that is multithread aware?
Click to expand...

Typically, applications that benefit from multiple processors/cores also benefit from Symmetric Multithreading (SMT) or "Hyperthreading" as Intel Marketing would call it. Usually the increase in performance on multithreaded apps/multiprocessing isn't near as high as having multiple processors or cores due to resource contention issues and having to share so many aspects of the same CPU (cache, integer math, floating math, code-decode units). However the cost of implementing SMT/Hyperthreading and increase in die size is minimal, far less than implementing dual processors or dual cores. In the Pentium 4's case, it's icing on the cake.

Personally, I see great potential in this technology as given sufficient time to mature, it could provide near dual processor capability (IBM predicted 80% efficiency against dual processor) on a single chip without much increase in cost or die size. Moreover future cpus like Intel's next Xeon, the Noconas, will feature higher levels of SMT. Imagine if you can, single chips, with 4x, 8x, 16x or even greater levels of symmetric multithreading running at up to 80% the efficiency of dual processors. Assuming processors continue to develop larger caches and more levels of cache (L3, L4) such a chip could be the equivalent of several of the same cpus without SMT and would cost several magnitudes less. More importantly, it would still be a single cpu, capable of being used as dual cpus, quad cpus or having dual cores, or quad cores or more. Have no doubt about it, this technology could be the next great advancement in cpu architecture.
 
Originally posted by yzedf
This is all a load of "you know what."

You can't compare x86 to PPC with the same compiler. Get the best for each, running as near to the same routines, and see what happens. Nothing like this will happen until G5 machines are out for use by the general public (September).

Either way, it is most impressive that a Apple "desktop" machine can be seriously compared with a x86 based "workstation."

As to the remarks about wireless networking, PCI-X, Bluetooth, S-ATA and all... none of that is being used now for serious workstation stuff. Wireless (802.11a/b/g) is not secure, PCI-X is new and for the most part un-used, same for BlueTooth (as well as security issues), and S-ATA should be more prevalent than it is, once it is actually faster than IDE stuff.

That combined wth his neverending use of the "Wintel" comment, just makes me look at what he says with a grain of salt. If you are going to shoot down the detractors, don't sink to their level to do it. Someone needs to be honorable in all of this.
Click to expand...

PCI-X is necessary for editing workstations, in many cases. A friend of mine owns a graphics/editing business with high-end clientele and he requires PCI-X to use the cards vital to his business. His dual Xeon machines use PCI-X and are custom built
 
Regarding the main topic:
I'm tired of repeating myself so why don't I just kindly direct your attention to the threads below. In the first one, I explain the problems at hand, in the second one, I delve into the details and confirm some of the things in my first post as well as post some new findings. In the third post, I examine just how much SSE2 affects performance with the latest ICC compiler.

https://forums.macrumors.com/showthread.php?s=&threadid=31475&perpage=&pagenumber=2

https://forums.macrumors.com/showthread.php?s=&threadid=32832

https://forums.macrumors.com/showthread.php?s=&threadid=31775&perpage=&pagenumber=2
 
Originally posted by yzedf
This is all a load of "you know what."

You can't compare x86 to PPC with the same compiler. Get the best for each, running as near to the same routines, and see what happens. Nothing like this will happen until G5 machines are out for use by the general public (September).

Either way, it is most impressive that a Apple "desktop" machine can be seriously compared with a x86 based "workstation."

As to the remarks about wireless networking, PCI-X, Bluetooth, S-ATA and all... none of that is being used now for serious workstation stuff. Wireless (802.11a/b/g) is not secure, PCI-X is new and for the most part un-used, same for BlueTooth (as well as security issues), and S-ATA should be more prevalent than it is, once it is actually faster than IDE stuff.

That combined wth his neverending use of the "Wintel" comment, just makes me look at what he says with a grain of salt. If you are going to shoot down the detractors, don't sink to their level to do it. Someone needs to be honorable in all of this.
Click to expand...

I'm happy someone finally said this. While i'm a PC user, it doesn't mean i believed every single website out there that denounced Apple's benchmarks. In the same way, i would highly doubt any literature from a website rightfully named "applelust" -- that already gives away its condescending bias toward ALL PC users. When i see either "wintel" or "peecee" i can pretty much disregard anything said in that context and not miss out on "learning" anything new.

Likewise, I can see that with all the heat Apple has been taking that people would gratefully side with any kind of writer, i.e. applelust and its article on "anti-apple bias".

I have to agree, however, the more i think about, at just how crazy it is that a retailer is distributing 64-bit desktops. While i am and AMD fan myself, there is no way i'm spending that kind of money on existing opteron boards. I already spent a fair amount of money upgrading (well in college student terms :) ), so i will be on the wait-see approach on the upcoming Athlon 64 cpus.

To end the rant, i know that Apples will always be different than PC's b/c of architectural/OS differences. Benchmarks are the only way of "bridging", albeit the only modest and tangible way of even comparing the two. Even then, when you think about it, the comparison with even AMD and Intel was already complicated with the BABco 2002 benchmark dilemma.
 
Originally posted by Mav451
In the same way, i would highly doubt any literature from a website rightfully named "applelust" -- that already gives away its condescending bias toward ALL PC users. When i see either "wintel" or "peecee" i can pretty much disregard anything said in that context and not miss out on "learning" anything new.
Click to expand...

While I thought the article brought up some good points, after realizing it was on a site called Applelust made me feel exactly how you described. I don't mind wintel cause that is used in the pc world, but peeeceee or winblows really does nothing to further your argument besides make you look like a 12 year old. If your argument is good enough and supported with facts, then you don't need stupid commentary to help it along.
 
Originally posted by Mav451
Benchmarks are the only way of "bridging", albeit the only modest and tangible way of even comparing the two. Even then....
Click to expand...

That is precisely why Apple spent so much time emphasizing the "real-world" comparisons. Software that is cross platform, optimized for each platform...
 
Originally posted by yzedf
This is all a load of "you know what."

You can't compare x86 to PPC with the same compiler. Get the best for each, running as near to the same routines, and see what happens.
Click to expand...

I can't agree with this. GCC is a multiplatform compiler, yes, but guess what architecture has the most developers working on it? Intel. Pardon me if you are informed, but a hell of a lot of people here and elsewhere (those that bash this as a "Mac-biased compiler") seem un- or mis-informed. The Gnu developers write code for whatever platform pleases them. And for most, that must be Intel, that is and has been the most widely available *cheap* hardware around for a long time. Yes, the Gnu developers have not brought out a wickedly optimised compiler for any platform that I know of, but they have done a LOT more tweaking for x86, SSE, and MMX than for PPC/Alvitec. There's just so many more of them to do x86 stuff. And the x86 architecture is open, the PPC is shrouded in mystery and RDF. The linux geeks had to reverse engineer a kernel for the G4, apple witheld the specs from them (and from Be, which is why Be went x86.)

Does this sound like a situation where the Gnu project would have time, manpower, and know-how to publish a strongly optimized compiler when they haven't done a bang-up job on architecture they had documentation for? I don't think so.

Just because GCC isn't a great compiler doesn't mean it's a biased compiler. I use it constantly on an x86 machine, I've never used any Intel compiler. If anything, I'd say GCC IS the fair choice, possibly still a bit better at x86 than PPC. It's closest to a level field IMO. It might not be the BEST field, but at least it's the same field.

Now apple is snuggling up to OSS and the Free Software Foundation, and helping with GCC. But I'd say that just begins to make up for the years of witheld specs. If you want benchmarks done with the absolute best possible compiler for each architecture, I think IBM and Intel have already done that. Both sets are superior to Apple's SPEC tests.

Originally posted by yzedf

Someone needs to be honorable in all of this.
Click to expand...

Now THAT I can support wholeheartedly.
 
Originally posted by panphage
I can't agree with this. GCC is a multiplatform compiler, yes, but guess what architecture has the most developers working on it? Intel. Pardon me if you are informed, but a hell of a lot of people here and elsewhere (those that bash this as a "Mac-biased compiler") seem un- or mis-informed. The Gnu developers write code for whatever platform pleases them. And for most, that must be Intel, that is and has been the most widely available *cheap* hardware around for a long time. Yes, the Gnu developers have not brought out a wickedly optimised compiler for any platform that I know of, but they have done a LOT more tweaking for x86, SSE, and MMX than for PPC/Alvitec. There's just so many more of them to do x86 stuff. And the x86 architecture is open, the PPC is shrouded in mystery and RDF. The linux geeks had to reverse engineer a kernel for the G4, apple witheld the specs from them (and from Be, which is why Be went x86.)

Does this sound like a situation where the Gnu project would have time, manpower, and know-how to publish a strongly optimized compiler when they haven't done a bang-up job on architecture they had documentation for? I don't think so.

Just because GCC isn't a great compiler doesn't mean it's a biased compiler. I use it constantly on an x86 machine, I've never used any Intel compiler. If anything, I'd say GCC IS the fair choice, possibly still a bit better at x86 than PPC. It's closest to a level field IMO. It might not be the BEST field, but at least it's the same field.

Now apple is snuggling up to OSS and the Free Software Foundation, and helping with GCC. But I'd say that just begins to make up for the years of witheld specs. If you want benchmarks done with the absolute best possible compiler for each architecture, I think IBM and Intel have already done that. Both sets are superior to Apple's SPEC tests.



Now THAT I can support wholeheartedly.
Click to expand...

Unfortunately, Intel spends nearly all of it's developer resources on ICC and IFC. Intel's first serious C/C++ compiler, ICC 5.0, came out in the same timeframe as the Pentium 4 and since then, nearly all Pentium 4 optimizations came from ICC.

GCC has had very few optimizations for the Pentium 4 at all and even then, several of them, ends up actually slowing the Pentium 4 in many instances. I've already listed some examples in my previous posts, such as the Pentium 4 not having enough registers for -march=pentium4 since the target machine is a 32 register RISC chip, and GCC not being able to schedule for the Pentium 4 at all (but it can for nearly every other x86 chip and apparently the PPC970), as well as others.

Remember that Apple has also spent alot of time working on it's own version of GCC, we now know this for a fact from the recent interview at Ars Technica with proof of a optimized scheduler, correct values for the MD file, and whatever other details that are sure to come out with time.
 
Originally posted by Mav451
.... In the same way, i would highly doubt any literature from a website rightfully named "applelust" -- that already gives away its condescending bias toward ALL PC users. When i see either "wintel" or "peecee" i can pretty much disregard anything said in that context and not miss out on "learning" anything new.
....
Click to expand...
Methinks the gentleman doth protest too much.

Exactly how does "lusting" for one brand of computers express disdain for another? Is it not possible that an "Appleluster" can at least be mildly attracted to SGI, Sun, IBM, or HP? Cannot an "Appleluster" have even a smidgen of respect for AlienWare, Toshiba, or Sony Vaio? Perhaps not in your world, but most certainly in mine.
 
Originally posted by Mr. MacPhisto
PCI-X is necessary for editing workstations, in many cases. A friend of mine owns a graphics/editing business with high-end clientele and he requires PCI-X to use the cards vital to his business. His dual Xeon machines use PCI-X and are custom built
Click to expand...
Workstation is one thing, desktop machine (G5) is another. The lines are beginning to blur...
 
Originally posted by yzedf
Workstation is one thing, desktop machine (G5) is another. The lines are beginning to blur...
Click to expand...

Most definitely, but I believe you did use the word "workstation" in your original post, I could be incorrect though. That is the one point tis article does make well - that the G5 offers workstation power at a desktop price. A dual-2 GHz G5 would likely be able to match many Unix workstations while costing much less, as the guy demoing Mathematica @ WWDC said about his product. I know several people who are already considering the G5 as their next workstation because the price for performance on the dual 2GHz is very, very good. And it comes loaded even at its base price.
 
Originally posted by MisterMe
Methinks the gentleman doth protest too much.

Exactly how does "lusting" for one brand of computers express disdain for another? Is it not possible that an "Appleluster" can at least be mildly attracted to SGI, Sun, IBM, or HP? Cannot an "Appleluster" have even a smidgen of respect for AlienWare, Toshiba, or Sony Vaio? Perhaps not in your world, but most certainly in mine.
Click to expand...

excuse me sir, but did you read the article? Try reading it from a PC user's view and attempt to finish the entire article. It was biased right from the get-go, read the title "smell of fear". Obviously they have no respect for PC's if they must use that kind of language. If you are an Apple user, you probably were nodding/agreeing and probably even snickering. That it why i do not think i'm protesting too much.

To finish, here is some of the extraordinary vocabularly that the biased author uses:
"wintel weenies", "weenies", "wintel flacks", "Apple-haters", "Great Wintel FUD Machine", and best of all: "Wintel Hegemony's longstanding arrogance"

one detail i noticed :)
"To date no one at Intel has publicly admitted that AMD's offerings also exceed Intel's Pentium 4 performance either."
>> the one thing i might agree with
 
Originally posted by Mav451
excuse me sir, but did you read the article? Try reading it from a PC user's view and attempt to finish the entire article. It was biased right from the get-go, read the title "smell of fear". Obviously they have no respect for PC's if they must use that kind of language. If you are an Apple user, you probably were nodding/agreeing and probably even snickering. That it why i do not think i'm protesting too much.

To finish, here is some of the extraordinary vocabularly that the biased author uses:
"wintel weenies", "weenies", "wintel flacks", "Apple-haters", "Great Wintel FUD Machine", and best of all: "Wintel Hegemony's longstanding arrogance"

one detail i noticed :)
"To date no one at Intel has publicly admitted that AMD's offerings also exceed Intel's Pentium 4 performance either."
>> the one thing i might agree with
Click to expand...

Yes, the bias here is bad if the authors intent was truly an attempt to enlighten. His facts about the SPEC scores are interesting and could have been presented as a solid refutation, pumping the Mac platform as a solid alternative to MS and Intel without demeaning them.

I do agree with his statement about Intel and AMD - Intel has and will never admit that AMD has made chips faster - and at a better value. But this can be expected regardless. Apple never came out and publically said: "The new P4s wipe the floor with our G4s." While there was some biased testing on the G4 (DVM ran Adobe AfterEffects on both - knowing full wll that Adobe had not optimized AfterEffects very well for Altivec but had worked hard on SSE optimization), it is well known that they became less competitive over a year ago.

I think stressing advantages of the OS, the interface, and the new chip design while not completely bashing MS and Intel would have worked better. A lot of people realize that the tests Apple commissioned are some of the best done by a company in recent years in an industry known for every company skewing results.

The tests run were done, I feel, to really see where the new IBM chip was in comparison to the best Intel is currently offering. Apple found that the chip held up quite well - and showed that they could compete pricewise (although the single 1.6 and single 1.8 should be cheaper, IMHO). It is also clear that Apple is aiming at Intel and not AMD - who they appear to have some agreements with.

And I actually got to see one of these things up against a loaded Xeon last week and the SPEC reports don't tell the whole story. The Photoshop presentation was dead on - this thing moves about twice as fast as the Xeons when rendering. I'd like to check it out against an Opteron at some point, but I don't have access to an Opteron machine at this point.
 
Originally posted by Mav451
excuse me sir, but did you read the article? Try reading it from a PC user's view and attempt to finish the entire article. It was biased right from the get-go, read the title "smell of fear". Obviously they have no respect for PC's if they must use that kind of language. If you are an Apple user, you probably were nodding/agreeing and probably even snickering. That it why i do not think i'm protesting too much.
Click to expand...
Indeed I did read the article. I enjoyed it immensely. You make a serious mistake. To paraphrase Carly Simon, you're so vain you think the article's about you. It is not. The article was written to reassure Mac users that all of the crap from the Wintel Weenies is just that, crap. Apple hired a highly reputable firm to benchmark its new computers and the most readily identifiable competition as accurately as possible. For all the catawauling from Gate's Gang, this article shows that the only peole trying to obfuscate the truth are Microsoft Minions. The article is not intended to convince them, because they know the truth. They are just trying to hide it.
Originally posted by Mav451
To finish, here is some of the extraordinary vocabularly that the biased author uses:
"wintel weenies", "weenies", "wintel flacks", "Apple-haters", "Great Wintel FUD Machine", and best of all: "Wintel Hegemony's longstanding arrogance"
Click to expand...
Just because being called a "lying SOB" hurts someone's feelings doesn't mean that he isn't lying or that his mom isn't named Trixie.
 
Originally posted by yzedf
Workstation is one thing, desktop machine (G5) is another. The lines are beginning to blur...
Click to expand...

Guys... Indeed things are really getting blurred.
A desktop computer selling for 2259 euro's ??? (cheapest one) without a monitor:confused:
Sorry I do not call that a desktop price but a workstation price.
So please don't fool yourselves if you want to compare the G5's than compare them to relativly the same priced machines, in the pc world these are usually workstations.
A pc desktop would sell anything between 1000 and 1500 euro's including monitor.

The I-Mac is more priced like a desktop.

Apple can call their G5 machines whatever they want, but they are priced like workstations and have the features of workstations. so comparing these with pc workstaions isn't only logical but also the only fair comparison
 
Originally posted by MisterMe
Indeed I did read the article. I enjoyed it immensely. You make a serious mistake. To paraphrase Carly Simon, you're so vain you think the article's about you. It is not. The article was written to reassure Mac users that all of the crap from the Wintel Weenies is just that, crap. Apple hired a highly reputable firm to benchmark its new computers and the most readily identifiable competition as accurately as possible. For all the catawauling from Gate's Gang, this article shows that the only peole trying to obfuscate the truth are Microsoft Minions. The article is not intended to convince them, because they know the truth. They are just trying to hide it.
Click to expand...

Being someone who doesn't believe the comparison was done "as fair as possible" or even remotely fair for that matter, I have yet to see any of my "crap", or any of the "crap" from the techies and engineers that form the center of "Gate's Gang" refuted by this article or any other article to date for that matter.
 
Originally posted by ilben77
Guys... Indeed things are really getting blurred.
A desktop computer selling for 2259 euro's ??? (cheapest one) without a monitor:confused:
Sorry I do not call that a desktop price but a workstation price.
So please don't fool yourselves if you want to compare the G5's than compare them to relativly the same priced machines, in the pc world these are usually workstations.
A pc desktop would sell anything between 1000 and 1500 euro's including monitor.

The I-Mac is more priced like a desktop.

Apple can call their G5 machines whatever they want, but they are priced like workstations and have the features of workstations. so comparing these with pc workstaions isn't only logical but also the only fair comparison
Click to expand...
That is only for PC pricing in the past 2 years. For the 2 decades before that, if you wanted a decent desktop you had to spend 2 grand.

Blurred = performance, not price.

Desktop or Workstation? It is for the marketing droids to figure out, not me.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back