Forums | MacLife

avkills · 2007-12-25 1:10 pm

I wish x86 had died at netburst. If IBM and Motorola hadn't screwed the PowerPC desktop/mobile chips into the ground I think things would be a lot better.

I won't argue that Apple made the right decision; but from a pure technical standpoint, x86 is a smurfing cluster smurf of patches and hacks upon hacks built around a ancient ISA that should have already died. PowerPC is technically a superior product and I am extremely pissed off that IBM dropped the ball; perhaps IBM saw the writing on the wall that the PowerPC 970 was a quick fix for Apple, and knew damn well that Apple had no intention of sticking with the PowerPC architecture, hence the lack of follow through on IBMs part. It is no secret IBM can design and deliver; they did not seem to have much trouble producing CPUs for the game console makers, and they still currently still hold the top 2 supercomputer rankings.

These PC mags would not be giving Apple the time of day if Apple was still using PowerPC chips; which makes me believe that there are a whole lot of publications with hands inside of Intel's pockets. Intel can deliver the goods, but in my honest opinion, their products are only "good enough". I just recently found a Intel published PDF from ArsTechnica talking about the "Ghz myth". These smurfing asshats knew all along they were marketing lies in the stupid Mhz wars with AMD. They are all proud of themselves now for having processors that can do 4 ops per clock; welcome to smurfing 2000 geniuses. The 970 can do 5.

Ok my rant is over; and I would lying if I said I did not want an 8-core beast from Apple; I just wish that 8-core beast could have been a 980 or 990 from IBM based off of the Power6. sad

-mark

NAG · 2007-12-25 1:38 pm

Yeah, the AIM alliance really did drop the ball. I pretty much believe it (basically) died because Microsoft didn't support it for their consumer machines.

FutureDreamz · 2007-12-25 2:15 pm

Wasn't the Core Duos essentially a rewrite?

Mr. T · 2007-12-25 2:35 pm

The limits of the CISC architecture were mostly overcome years ago, with the Pentium Pro. The basic problem with CISC at the time, was that precious transistors were being wasted on redundant logic, instead of more registers, pipelining, branch prediction, superscalar, single-cycle execution, large caches, and other RISC design attributes. Recognizing this, Intel began designing RISC chips that performs a CISC-to-RISC conversion (for compatibility), and of course, AMD followed suit. Since that time, as transistors became less and less of a commodity, the RISC conversion logic became very inexpensive. Today, there really isn't any clear difference between x86 and PPC. PPC has accumulated so many instructions over the years, the terms RISC has become somewhat misleading.

fuzzynormal · 2007-12-25 4:42 pm

In 2000 I bought Final Cut Pro because I needed a good and affordable DV editing app. Then I bought a Mac to go with it.

My experience may have been out side the norm 7 years ago, but these days I think we can all agree that the media creation aspect of using a computer is really important. Kudos for Apple in realizing that practical productivity matters to people. They've developed their products to reflect this aspect of the marketplace.

Computer specs get less meaningless as the technology homogenizes, so the more important motivation for a consumer then becomes the "what can I really accomplish with this computer thing?"

I think this is how Apple really want to sell their stuff. Makes sense to me. Their satisfaction numbers in polls like this reflect this intrinsic value. It's not ALL about MHz numbers or price points.

avkills · 2007-12-25 5:13 pm

Mr. T wrote:
The limits of the CISC architecture were mostly overcome years ago, with the Pentium Pro. The basic problem with CISC at the time, was that precious transistors were being wasted on redundant logic, instead of more registers, pipelining, branch prediction, superscalar, single-cycle execution, large caches, and other RISC design attributes. Recognizing this, Intel began designing RISC chips that performs a CISC-to-RISC conversion (for compatibility), and of course, AMD followed suit. Since that time, as transistors became less and less of a commodity, the RISC conversion logic became very inexpensive. Today, there really isn't any clear difference between x86 and PPC. PPC has accumulated so many instructions over the years, the terms RISC has become somewhat misleading.

I won't argue that. But the x86 ISA is ancient. PPC's ISA is more modern. End of story. I just hope in 10 years that we still are not using x86, but we probably will be because it is always about cheap cheap cheap.

That conversion takes away performance. Why not just throw the CISC smurf out to the dumpster and just use the RISC ISA?

-mark

Pariah · 2007-12-25 6:34 pm

fuzzynormal wrote:
In 2000 I bought Final Cut Pro because I needed a good and affordable DV editing app. Then I bought a Mac to go with it.

My experience may have been out side the norm 7 years ago, but these days I think we can all agree that the media creation aspect of using a computer is really important. Kudos for Apple in realizing that practical productivity matters to people. They've developed their products to reflect this aspect of the marketplace.

Computer specs get less meaningless as the technology homogenizes, so the more important motivation for a consumer then becomes the "what can I really accomplish with this computer thing?"

I think this is how Apple really want to sell their stuff. Makes sense to me. Their satisfaction numbers in polls like this reflect this intrinsic value. It's not ALL about MHz numbers or price points.

So you are saying that because media creation has gone mainstream people need to worry less about the power of their computers??
I would disagree most strongly on that.
I am curious. In 2000, when you bought FCP did you buy an iMac or Laptop to run it on?

Last edited by Pariah (2007-12-25 6:36 pm)

Mr. T · 2007-12-26 12:06 am

avkills wrote:
Mr. T wrote:
The limits of the CISC architecture were mostly overcome years ago, with the Pentium Pro. The basic problem with CISC at the time, was that precious transistors were being wasted on redundant logic, instead of more registers, pipelining, branch prediction, superscalar, single-cycle execution, large caches, and other RISC design attributes. Recognizing this, Intel began designing RISC chips that performs a CISC-to-RISC conversion (for compatibility), and of course, AMD followed suit. Since that time, as transistors became less and less of a commodity, the RISC conversion logic became very inexpensive. Today, there really isn't any clear difference between x86 and PPC. PPC has accumulated so many instructions over the years, the terms RISC has become somewhat misleading.
I won't argue that. But the x86 ISA is ancient. PPC's ISA is more modern. End of story. I just hope in 10 years that we still are not using x86, but we probably will be because it is always about cheap cheap cheap.

That conversion takes away performance. Why not just throw the CISC smurf out to the dumpster and just use the RISC ISA?

-mark

The performance "loss" doesn't come from the conversion. The problem came from the fact that the conversion logic once constituted a large portion of the chip's total number of transistors - leaving designers less able to apply them to more important areas of the CPU. However, the number of transistors involved in the conversion logic has remained roughly constant over the years, so as transistor counts began to multiply, the wasted portion of the chip eventually became insignificant - as did the performance loss.

In other words, the only tangible difference between CISC and RISC today is the cost of development - RISC is cheaper. So if designers find a way to build a "true" RISC chip without compromising performance for x86 apps, then x86 will probably die. Will this happen? Probably. I foresee two possibilities. First is that designers may eventually have so many cores, they'll be able to fit a a couple of legacy cores on a RISC-architecture chip. The second more likely possibility is that a technology change will drive the shift (as in silicon to quantum computing): a quantum (pun intended) leap in performance will ensure that x86 apps will run well even in emulation.

Last edited by Mr. T (2007-12-26 12:09 am)

avkills · 2007-12-26 9:56 am

fuzzynormal wrote:
In 2000 I bought Final Cut Pro because I needed a good and affordable DV editing app. Then I bought a Mac to go with it.

My experience may have been out side the norm 7 years ago, but these days I think we can all agree that the media creation aspect of using a computer is really important. Kudos for Apple in realizing that practical productivity matters to people. They've developed their products to reflect this aspect of the marketplace.

Computer specs get less meaningless as the technology homogenizes, so the more important motivation for a consumer then becomes the "what can I really accomplish with this computer thing?"

I think this is how Apple really want to sell their stuff. Makes sense to me. Their satisfaction numbers in polls like this reflect this intrinsic value. It's not ALL about MHz numbers or price points.

Yes I will have to disagree as well. Go find some HDV footage and then come to me, because I can tell that HDV editing sucks even on my dual 2Ghz G5. Personally I would never *buy* into HDV, but you don't always get to edit what you want when you are in the biz. And I wasn't even doing that much; pretty much a HDV to DVD conversion. I wanted to hang myself. HDV is the worst editing codec I've ever worked with; and I say "editing codec" very loosely since it really isn't.

-mark

Czachorski · 2007-12-26 10:08 am

Sorry to post and then abandoned this thread. We had a wind storm in the Detroit Area on Sunday, and our power was out all day on the 23rd and 24th, with nasty cold weather (it was 27 deg in my house) and XMas Day Party planned at our house for the whole family. So we had an extremely busy 3 days to pull it all off. It worked, Santa came, kids happy, party successful, things back on track now.

smile

I think it is great that Apple is getting this recognition for their better than average hardware. It has been in Consumer Reports like that for about 5 years, but is a more recent phenomenon with the PC-centric magazines, probably correctly stated by someone earlier who said from the intel change over.

I will say that for the first time in 5 years, Apple slipped a little in the Consumer Reports annual reader survey this year. I tend to trust CR a little more due to their complete impartiality. I think the CR survey is a sign that here has been a little drop off in Apple quality as they have scaled up to a larger production. Not to give them a free pass, but I suppose that is not an uncommon thing as a company scales up. I would expect Apple to be uncommon and keep the high quality as they scale. Their ability to keep this thing going hinges on them continuing to put out high quality products, and I am sure they know that. All in all, great news for Apple! Keep it up!

Last edited by Czachorski (2007-12-26 10:10 am)

avkills · 2007-12-26 10:11 am

Yes but the hardware is *technically* no longer Apple designed. Only the software is. The only hardware Apple is designing is enclosures on the computer side. All this means is somewhat lower prices (although barely) and actually higher margins for Apple since they no longer have to design system controller chips and motherboards.

-mark

Czachorski · 2007-12-26 10:14 am

The things they do have control over must have some impact on quality. Five years of top marks, both before and after the intel change over, is a little hard to chalk up to just case design.

avkills · 2007-12-26 10:14 am

Mr. T wrote:
The performance "loss" doesn't come from the conversion.

I don't care what you are converting, if there is converting taking place digitally it takes time and therefore effects performance. Sure it is very slight, but it is still there.

I wonder if Apple had a hand in SSE4; since they have experience with Vector regarding Altivec. Although they probably can't disclose anything about Altivec to Intel.

-mark

avkills · 2007-12-26 10:16 am

Czachorski wrote:
The things they do have control over must have some impact on quality. Five years of top marks, both before and after the intel change over, is a little hard to chalk up to just case design.

Case enclosure design can definitely have a big impact; if the chips can't be cooled properly they don't like to work.

-mark

Czachorski · 2007-12-26 10:47 am

I have a this nagging suspicion that high marks in reader surveys like this takes more than good chip cooling.

wink

avkills · 2007-12-26 11:46 am

Czachorski wrote:
I have a this nagging suspicion that high marks in reader surveys like this takes more than good chip cooling.

Yeah, good software. My guess is that the average joe is going to rate the hardware lower if the software does not work. That is human nature.

-mark

Last edited by avkills (2007-12-26 11:47 am)

Mr. T · 2007-12-26 4:34 pm

avkills wrote:
Mr. T wrote:
The performance "loss" doesn't come from the conversion.
I don't care what you are converting, if there is converting taking place digitally it takes time and therefore effects performance.

That's certainly true for software, but it doesn't work like that at the hardware level. As long as the conversion is done in one cycle (and it is - frequently less than one), there's no performance loss (btw, I realize this may seem a little counter-intuitive, but I have a pretty good background in this. smile ).

Mr. T · 2007-12-26 4:37 pm

avkills wrote:
Czachorski wrote:
I have a this nagging suspicion that high marks in reader surveys like this takes more than good chip cooling.

Yeah, good software. My guess is that the average joe is going to rate the hardware lower if the software does not work. That is human nature.

-mark

Definitely!

avkills · 2007-12-26 4:56 pm

Mr. T wrote:
avkills wrote:
Mr. T wrote:
The performance "loss" doesn't come from the conversion.
I don't care what you are converting, if there is converting taking place digitally it takes time and therefore effects performance.
That's certainly true for software, but it doesn't work like that at the hardware level. As long as the conversion is done in one cycle (and it is - frequently less than one), there's no performance loss (btw, I realize this may seem a little counter-intuitive, but I have a pretty good background in this. ).

You just contradicted yourself. big_smile Even if it takes 1/2 a cycle, that is 1/2 a cycle that normally would not have been used for straight processing. So say you have a complex instruction that takes 1/2 cycle to break down in micro-ops (or reduced instructions); it would be faster if it was just going straight with the reduced instructions. shrug

Anyway I know it can't be that much of a hit, otherwise modern x86 chips would really suck wind.

-mark

mo' ron · 2007-12-26 6:02 pm

avkills wrote:
. So say you have a complex instruction that takes 1/2 cycle to break down in micro-ops (or reduced instructions); it would be faster if it was just going straight with the reduced instructions.

Anyway I know it can't be that much of a hit, otherwise modern x86 chips would really suck wind.

-mark

That's true, but that "complex instruction" you are referring to is one that would have taken multiple instructions anyway on a RISC architecture. So it works out the same as if it was RISC. And in HW, it doesn't necessarily have to crack the op THEN decode it, the decode stage could have a built-in cracking.

volk · 2007-12-26 6:46 pm

Are we really going to beat this dead horse again???

Apple scores great in customer satisfaction and quality...

Pariah: Yeah, but their hardware sucks because they don't offer a consumer desktop...

Czach, Sci, volk, et all: Marketshare and other reports contradict your opinion. The current machines are very nice and do most everything well.

Pariah: You're all wrong, I am right. Apple will make a consumer tower at my price point or I am taking my toys and going home.

Czach, Sci, volk, et all: Why not run Linux...it meets all your criteria...

Pariah: Linux sucks. I'd rather bitch about Apple.

....and the loop continues....

Last edited by volk (2007-12-26 6:46 pm)

ScifiterX · 2007-12-26 8:38 pm

Actually I see nothing wrong with modern integrated graphics and while I wouldn't mind being able to upgrade I'm unlikely to ever do so (and thus bought accordingly). If I did buy with upgrading in mind you can bet your bottoms dollar I'd research and get one that can upgrade without dinking around too very much.

BTW the Mac OS serves my personal needs infinitely better than Linux.

NAG · 2007-12-26 8:41 pm

I have to agree that most people confuse hardware and software frequently. Why else would people buy a new computer just because it is filled with viruses (only to fill up the new one with viruses shortly after)?

mudogramx · 2007-12-26 8:46 pm

Because its a magic box that do some sort of sorcery to make them images move on the screen!

NAG · 2007-12-26 8:47 pm

Damn ghosts in the machine are acting up again!

Forums | MacLife

#26 2007-12-25 1:10 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#27 2007-12-25 1:38 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#28 2007-12-25 2:15 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#29 2007-12-25 2:35 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#30 2007-12-25 4:42 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#31 2007-12-25 5:13 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

Mr. T wrote:

#32 2007-12-25 6:34 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

fuzzynormal wrote:

#33 2007-12-26 12:06 am

Re: Two Prominent PC-Mags Give Apple Top Marks

avkills wrote:

Mr. T wrote:

#34 2007-12-26 9:56 am

Re: Two Prominent PC-Mags Give Apple Top Marks

fuzzynormal wrote:

#35 2007-12-26 10:08 am

Re: Two Prominent PC-Mags Give Apple Top Marks

#36 2007-12-26 10:11 am

Re: Two Prominent PC-Mags Give Apple Top Marks

#37 2007-12-26 10:14 am

Re: Two Prominent PC-Mags Give Apple Top Marks

#38 2007-12-26 10:14 am

Re: Two Prominent PC-Mags Give Apple Top Marks

Mr. T wrote:

#39 2007-12-26 10:16 am

Re: Two Prominent PC-Mags Give Apple Top Marks

Czachorski wrote:

#40 2007-12-26 10:47 am

Re: Two Prominent PC-Mags Give Apple Top Marks

#41 2007-12-26 11:46 am

Re: Two Prominent PC-Mags Give Apple Top Marks

Czachorski wrote:

#42 2007-12-26 4:34 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

avkills wrote:

Mr. T wrote:

#43 2007-12-26 4:37 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

avkills wrote:

Czachorski wrote:

#44 2007-12-26 4:56 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

Mr. T wrote:

avkills wrote:

Mr. T wrote:

#45 2007-12-26 6:02 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

avkills wrote:

#46 2007-12-26 6:46 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#47 2007-12-26 8:38 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#48 2007-12-26 8:41 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#49 2007-12-26 8:46 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

#50 2007-12-26 8:47 pm

Re: Two Prominent PC-Mags Give Apple Top Marks

Board footer