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

Something is Coming...

Why has no one pointed out that boxes shipped to computer stores hold not one but numerous machines, particularly in the case of smaller systems such as notebooks? The idea that anyone would see small boxes and large boxes and assume notebooks and desktops, respectively, is therefore laughable. For that writer to actually see notebook-sized boxes, the boxes on a single pallette would have to be held together by shrinkwrap (which is possible), but with *no paper covering the notebook-sized boxes*. If that were true (and knowing Apple, it never would be), someone could immediately see what product they hold.

I've also said this before, but Steve is on the record saying that WWDC is not a venue for new hardware. This might change, of course, but expectations here are *way* too high.

elo
 
Originally posted by elo
Why has no one pointed out that boxes shipped to computer stores hold not one but numerous machines, particularly in the case of smaller systems such as notebooks? The idea that anyone would see small boxes and large boxes and assume notebooks and desktops, respectively, is therefore laughable. For that writer to actually see notebook-sized boxes, the boxes on a single pallette would have to be held together by shrinkwrap (which is possible), but with *no paper covering the notebook-sized boxes*. If that were true (and knowing Apple, it never would be), someone could immediately see what product they hold.

I've also said this before, but Steve is on the record saying that WWDC is not a venue for new hardware. This might change, of course, but expectations here are *way* too high.

elo
Click to expand...

expectations are way too high, yes... but boy it sure is fun! screw reality, we have wild speculation and we actually have rumors to back it up.... if you can use rumors to back anything up that is. but hey, why wouldn't any of it be announced?
 
Re: "64-bit consumer"

Originally posted by AidenShaw
I like 64-bits - I use 64-bit systems every day. Most of them have at least 16 GiB of RAM.

What I'm arguing is that if Apple shouts "64-bits" and sells you boxes with 256MiB of RAM, that's like putting a tachometer that goes to 12,000 RPM on your Trabant.

Nice tach, but what good is it?
Click to expand...

Well, I see your point now. Still, copious amounts of RAM will soon become useful, and anyway, I know few "consumers" that stay with their original amount of RAM. Anyway, of course 64bit doesn't do much in the way of speed as such, unless you write code to deal with very big chunks of data (like Video, e.g.). I is useful there.
 
Re: CNBC

Originally posted by pkradd
CNBC streamed the recent Apple event that introed new iPods and iTMS. No reason to believe they won't do the same for WWDC, even if Apple only streams to stores as they did recently...
Click to expand...
Remember that most, if not all of the stuff at WWDC is NDA'd. If Apple did broadcast the WWDC keynote, they'd have to cut out stuff that they'd otherwise be presenting to NDA'd developers...
 
Re: Re: "64-bit consumer"

Originally posted by dekator
Still, copious amounts of RAM will soon become useful, and anyway, I know few "consumers" that stay with their original amount of RAM.
Click to expand...

One key thing will be to look at the max memory that the systems will support.

If it's the same old 2 GiB, people who now use 64-bits will laugh at any Apple ad that shouts "64-bit". 2 GiB would probably be the max for a laptop (x86 laptops have that, there are 4GiB SPARC laptops but they're big and heavy).

If the max is 4 GiB...yawn. 16 to 32 GiB would be very interesting. And it's OK if Apple says "4 GiB now, 16 GiB when X GiB DIMMs are available" - headroom is important.

unless you write code to deal with very big chunks of data (like Video, e.g.)
Click to expand...

Actually, video has very small chunks of data (frames), just huge numbers of them! ;)
 
Re: clearly a fake

Originally posted by grayleg
The "Heart of Europe" is very strange. Not likely used by marketing ppl. The pic is obvs. Photoshopped. Sorry Guys, u'll have to wait a couple of days.
Click to expand...

What's strange about the "Heart of Europe", man ?
Maybe it is for someone from Sweden, a country that refuses to take part in Europe. For others it's not strange at all. Odd you would say so.

If I'd say: "In Topeka, the heart of America". That wouldn't be strange either, would it. Or would it ?

Also, I think there is a good chance the invitation isn't fake. Anyway, Cancom, a reseller has put up a pic on their shop page, showing a wrapped-in Powermac. Text: NEUE Apple Desktops
ab Montag 23.06. abends (= New Apple Desktops, starting Monday evening 23rd of June). So, the PPC 970 *is* coming.
http://shop.cancom.de/shop/MediaAndBusinessSolutions/wa?showtemplatecomponent=t_startseite
 
Heart of Europe is not too far fetched, as you can think that it comes close to the center of the EU (kind of). The link here, Cancom store shows that they are quite confident of new machines. Hmm.
Of interest, if you save the pick of the packed PowerMac to the Desktop, the filename is (translated) "G5-packed-grey-small.jpg"... heheheh!!:cool:
 
Re: Re: clearly a fake

Originally posted by dekator
What's strange about the "Heart of Europe", man ?
Maybe it is for someone from Sweden, a country that refuses to take part in Europe.[/url]
Click to expand...

Huh? <cynism>You're American, right? </cynism>

Sweden is a memebr state of the EU. Maybe you mean Switzerland? That is admittedly not a member, but geographically very definitively in the heart of Old Europe (TM).

But I agree, if we're talking about the entire continent (as one should when talking about 'Europe') then Berlin is a good candiate to be its heart. It's multi-cultural and geographically close to the former Eastern European countries (it was halfway part of "Eastern Europe" and more or less halfway on the north-south line as well.

Andreas
 
Re: Fight the "64-bit myth"

Originally posted by AidenShaw
And you're in the camp that thinks that 64-bit is some magic elixir that can improve everything. ;)

Look at http://www.anandtech.com/guides/viewfaq.html?i=112


That article bebunks your other claims as well....

Did you know that x86/G4/970 all support exactly the same floating point formats? Same size and precision? No 64-bit benefit here.

Did you know that the P4 and G4 already have internal datapaths up to 256 bits in width? They have 64-bit wide external memory busses. They can load 64-bit (fp) and 128-bit registers (sse/altivec) in a single cycle. No 64-bit benefit here.

Did you know that C compilers (VC++ and gcc) have support for 64-bit integers already? They generate extra instructions to accomplish this, but the program has the same precision 64-bit integers whether the code runs on a 32 or 64 bit CPU.

Separate the 64-bit truth (more memory) from the hype. The 970 is a fast CPU that happens to be 64-bit. It is not a fast CPU because it is 64-bit.
Click to expand...


*yawn*

Yes, but the datapath to/from the CPU it 32-bit in all your examples. You neglect to acknowledge the simple fact that doubling that data path halves tranfer rate with a 64-bit bus vs. the same bus with a 32-bit CPU ingress/egress.

Did you know that current compilers that can produce 64-bit precision on 32-bit chips not only do it with extra instructions, but that those extra instructions *disappear* when compiled 64-bit clean on a 64-bit CPU?

Did you know that the register size is irrelevant, on the current G4/P4s, as the bottleneck we're discussing -- and thus one of the benefits we'll see -- is the point at which data enters or leaves the CPU?

You're full of sound and fury, signifying nothing. Please finish fretting your time upon this stage.
 
Originally posted by AidenShaw
And you're in the camp that thinks that 64-bit is some magic elixir that can improve everything.
Click to expand...
Stop thinking backwards!

By 2010 all consumer computers will most likely be 64-bit. It's the next logical step. Someone at Microsoft (I don't think it was Gates, but it definitely wasn't Steve Jobs) made the claim that nobody would ever need a 32-bit chip. So would you like to try running iTunes on a Lisa?

BTW, the 970 processes more Gigaflops than the G4. Much more. And gigaflops are a major factor in computing power. Possibly even more important than clock speed. And, speaking of clock speed, let's not forget that the 970 alone will have better speeds than any G4 Moto has tried to crank out.
 
Re: Re: Fight the "64-bit myth"

Originally posted by mcl
*yawn*

Yes, but the datapath to/from the CPU it 32-bit in all your examples. You neglect to acknowledge the simple fact that doubling that data path halves tranfer rate with a 64-bit bus vs. the same bus with a 32-bit CPU ingress/egress.
Click to expand...

You do realize that both the Pentium4 and the G4 have 64-bit external data busses, don't you?


Did you know that current compilers that can produce 64-bit precision on 32-bit chips not only do it with extra instructions, but that those extra instructions *disappear* when compiled 64-bit clean on a 64-bit CPU?
Click to expand...
Of course, this is obvious.

And you realize that integer instructions are single cycle in most cases, so a 3GHz chip takes 166 picoseconds extra for that extra instruction. (Remember that the integer ALU on a Pentium 4 is double-pumped, and actually runs at 6GHz on a 3GHz chip.)

64-bit integers are fairly uncommon, so a few hundred picoseconds now and then doesn't matter for the overwhelming majority of applications.
 
Re: oxymoron - "64-bit consumer"

Originally posted by AidenShaw
This is really funny - "64-bit consumer"!

What "consumer" needs 4 GiB or more of RAM? 64-bits isn't any faster or better than 32-bits, unless you've hit the "31-bit" wall of 2GiB of memory.
Click to expand...

There is more to 64-bit OS's than just addressing, *if* the processor running it has, for example, wider than 32-bit registers.

Other architectual details can have a lot of effect on performance...more than just the total address space.
 
Re: Re: Re: Fight the "64-bit myth"

Originally posted by AidenShaw
You do realize that both the Pentium4 and the G4 have 64-bit external data busses, don't you?
Click to expand...

You do realize that part of my post -- the part you chose not to quote -- explains in small words that the point I was making regards the point of ingress and egress between the bus and the CPU, don't you? I don't care if it's got a 2^128 bus: If you double the datapath from the bus to the CPU (i.e., go from a 32-bit to 64-bit CPU), you'll reduce the time necessary to get data on and off the bus, and thus increase performance.


And you realize that integer instructions are single cycle in most cases, so a 3GHz chip takes 166 picoseconds extra for that extra instruction. (Remember that the integer ALU on a Pentium 4 is double-pumped, and actually runs at 6GHz on a 3GHz chip.)
Click to expand...

I happen to work on systems doing real-time processing where that extra 166 picoseconds matters. Your point?

You may also want to deal with the fact that it's floating-point, not integer, where the 970 shines against the current crop. I notice you've been studiously avoiding discussing FLOPS.


64-bit integers are fairly uncommon, so a few hundred picoseconds now and then doesn't matter for the overwhelming majority of applications.
Click to expand...

Again, you curiously avoid the obvious benefit of 64-bit floating-point calculations, which aren't uncommon.

Your misdirection's not all you'd wish.
 
Re: Re: Re: oxymoron - "64-bit consumer"

Originally posted by ssamani
I *think* it was actually one Steven P. Jobs referring to the mem in the original Macintosh in 84 and its lack of upgrade options compared to the IBM PC.
Click to expand...

Actually no, for the longest time, you had to use horrendous hacks to get PCs to use more then 640 KB of memory. The origional Macs had a 128K and 512K limit respectively, not 640K. After that, no Mac shipped with less than 1MB.
 
Re: Re: CNBC

Originally posted by cgaraffa
If Apple did broadcast the WWDC keynote, they'd have to cut out stuff that they'd otherwise be presenting to NDA'd developers...
Click to expand...

Last year the entire keynote was specifically not under NDA.
 
Originally posted by elo
Steve is on the record saying that WWDC is not a venue for new hardware.
Click to expand...

When did he make a global statement like that? Apple has denied new hardware at specific events (not this one), but if 970s are going to ship in a few months it makes no sense for them to hold back because "WWDC isn't for hardware".
 
Re: Re: Re: clearly a fake

Originally posted by jobberwacky
Huh? <cynism>You're American, right? </cynism>

Sweden is a memebr state of the EU. Maybe you mean Switzerland?
Andreas
Click to expand...

No, I'm not American and I'm well aware that Sweden is a member of the EU. However, AFAIK, they haven't joint the single currency yet. The latter is what I was referring to. However, I wasn't really serious. I just wanted to diss back ;-)
 
Re: Re: Re: Re: clearly a fake

Originally posted by dekator
No, I'm not American and I'm well aware that Sweden is a member of the EU. However, AFAIK, they haven't joint the single currency yet. The latter is what I was referring to. ;-)
Click to expand...
Yep, you're right about Sweden not (yet) having joined Euroland. However, why single out Sweden? The UK of A (as the country was quite appropriately called years ago in "Spitting Image") hasen't joined either. (Nor has Denmark, but it's long been known that there's something rotten in that state ;-)

Andreas
 
Sorry, I didn't mean to ignore you - I just didn't realize how confused you were.

Originally posted by mcl
If you double the datapath from the bus to the CPU (i.e., go from a 32-bit to 64-bit CPU), you'll reduce the time necessary to get data on and off the bus, and thus increase performance.

I happen to work on systems doing real-time processing where that extra 166 picoseconds matters. Your point?

You may also want to deal with the fact that it's floating-point, not integer, where the 970 shines against the current crop. I notice you've been studiously avoiding discussing FLOPS.

Again, you curiously avoid the obvious benefit of 64-bit floating-point calculations, which aren't uncommon.
Click to expand...

What's 64-bit floating point have to do with a discussion of 32-bit and 64-bit processors? Both P4 and G4 have 64-bit wide internal datapaths to the 64-bit registers already. No news here.

If the 970 has a superb floating point unit - fine. But that's unrelated to the bit width of the integer/pointer registers - which is what defines a 32-bit or 64-bit CPU.

Did you know that the G4 already has:

64-bit wide datapath from memory and/or L3 cache to the chip
256-bit wide datapath between L2 cache and L1 cache
128-bit wide datapath between L1 cache and vector registers
64-bit wide datapath between L1 and floating registers
32-bit wide datapath between L1 and integer registers

So, with that information in hand, what's your point about ingress/egress and 32 vs 64 bit CPUs?

Check out: MPC7450 RISC Microprocessor Family Technical Summary (http://e-www.motorola.com/brdata/PDFDB/docs/MPC7450TS.pdf) (particularly the diagram on page 4) at http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC7455&nodeId=03C1TR046708718653
 
Re: Sorry, I didn't mean to ignore you - I just didn't realize how confused you were.

Originally posted by AidenShaw
What's 64-bit floating point have to do with a discussion of 32-bit and 64-bit processors? Both P4 and G4 have 64-bit wide internal datapaths to the 64-bit registers already. No news here.



If the 970 has a superb floating point unit - fine. But that's unrelated to the bit width of the integer/pointer registers - which is what defines a 32-bit or 64-bit CPU.



Did you know that the G4 already has:



64-bit wide datapath from memory and/or L3 cache to the chip

256-bit wide datapath between L2 cache and L1 cache

128-bit wide datapath between L1 cache and vector registers

64-bit wide datapath between L1 and floating registers

32-bit wide datapath between L1 and integer registers



So, with that information in hand, what's your point about ingress/egress and 32 vs 64 bit CPUs?


deId=03C1TR046708718653[/url]
Click to expand...


My point is that you keep pointing to things that are either on-die or on-chip, and cliaming that's the be-all and end-all of enhancements, while you continue to ignore the fact that, no matter how big the internal datapaths are, they're bottlenecked if any data must pass to or from the chip to the rest of the system by the 32-bit access path.

A 64-bit chip doubles this access path.

As for "64-bit wide datapath from the memory and/or L3 cache to the chip", you're incorrect. From L3 cache, yes. That's on-chip (but not on-die). But there is no 64-bit wide path from the chip to and from the memory bus. it's 32-bit. Period.

However, you seem intent on ignoring even fundamental facts about the 64-bit advantages, jumping up and down and pointing at relative register sizes as though it eliminates the advantage of a double-sized interface between chip and system busses. So perhaps you'd better just end here, as will I. I obviously believe what I believe, you obviously believe what you believe. You're not persuading anyone, nor am I.

But feel free to have the last word, as I'm sure it's a pre-requisite for your ego.
 
Motorola and Apple can have the last words

Originally posted by mcl
...they're bottlenecked if any data must pass to or from the chip to the rest of the system by the 32-bit access path.

A 64-bit chip doubles this access path.

As for "64-bit wide datapath from the memory and/or L3 cache to the chip", you're incorrect. From L3 cache, yes. That's on-chip (but not on-die). But there is no 64-bit wide path from the chip to and from the memory bus. it's 32-bit. Period..

But feel free to have the last word, as I'm sure it's a pre-requisite for your ego.
Click to expand...

http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC7455&nodeId=03C1TR046708718653
36-bit physical address space for direct addressability of 64 Gigabytes of memory
High-bandwidth 133 MHz 64-bit MPX Bus/60x Bus
Click to expand...

http://e-www.motorola.com/brdata/PDFDB/docs/MPC7450TS.pdf
2.7 System Interface
The MPC7451 supports two interface protocols—MPX bus protocol and a subset of the 60x bus protocol. Note that although this protocol is implemented by the MPC603e, MPC604e, MPC740, and MPC750 processors, it is referred to as the 60x bus interface. The MPX bus protocol is derived from the 60x bus protocol. The MPX bus interface includes several additional features that provide higher memory bandwidth than the 60x bus and more efficient use of the system bus in a multiprocessing environment. Because the MPC7451 performance is optimized for the MPX bus, its use is recommended over the 60x bus.

The MPC7451 bus interface includes a 64-bit data bus with 8 bits of data parity, a 36-bit address bus with 5 bits of address parity, and additional control signals to allow for unique system level optimizations.
Click to expand...


http://www.apple.com/powermac/specs.html
Full 128-bit internal memory data paths
Click to expand...


http://developer.apple.com/techpubs...rMacG4Sept02/2Architecture/Processor_Bus.html
Processor Bus
The processor bus is a 133 or 167 MHz bus connecting the processor module to the U2 IC. The bus has 64-bit wide data and 32-bit wide addresses. The bus uses MaxBus protocols, supported by the U2 IC.
Click to expand...
 
Re: Re: Motorola and Apple can have the last words

Originally posted by mathiasr
You did not mention the funniest:
http://www-3.ibm.com/chips/techlib/...2AE087256C5200611780/$file/PPC970_MPF2002.pdf
PowerPC 970:
- Two unidirectional buses
- 32-bit read, 32-bit write :D
Click to expand...
I was saving that irony for later! ;)

But first, I was going to mention the fact that the memory DIMMs are 64-bits wide - and ask if it was just a coincidence that 8 bytes times the 167Mhz memory bus gives the same 1.3GB/sec that Apple claims for memory bandwidth?


<kidding>

Hmmm.... PPC970 has 32-bit busses.

The 32-bit G4 computer has 64-bit busses, and the 64-bit PPC970 computer has 32-bit busses.... That would make the 970 4 times slower than the G4, right, because it needs twice as much and it gets half as much?

</kidding>
 
Re: Re: Re: Re: Re: clearly a fake

Originally posted by jobberwacky
Yep, you're right about Sweden not (yet) having joined Euroland. However, why single out Sweden? The UK of A (as the country was quite appropriately called years ago in "Spitting Image") hasen't joined either. (Nor has Denmark, but it's long been known that there's something rotten in that state ;-)

Andreas
Click to expand...

As I said, I was joking. Sweden because the guy was from Sweden (no good lashing out at the UK then, is it ?).
Denmark, yes, well... *no comment*.
As for the UK, I hardly consider them to be European anymore. If the UK takes the Euro or not, frankly, I couldn't care less. But Sweden, now, those used to be reasonable, nice ppl :)

BTW, "UK of A" that's superb (ich lach' mich rund und kugelig). :D :D :D
 
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