Some questions for all the CPU oriented folk here...
After looking at the specs (proposed, I suppose) of the MPC8560, I started wondering why Apple couldn't use this chip. Anyone know?
I seem to recall that someone around here said the e500 was a slow core, but the listed Drystone puts it about 15% higher than the current G4, so what makes it so slow? Or is this just a higher bandwidth improvement, meaning the core isn't data starved?
How difficult would it be to tack in a second core? Would that make any real difference? How about another core entirely?
So the SIMD on the e500 isn't AltiVec, but what is it? How does it compare to AltiVec, especially with these 222 added new instructions?
The connection (RapidIO) lists at 500mhz at 8 bit. Is 8 bit standard for this type of connection? How does this compare to AMD/Intel's FSB? I also saw a listing for 1ghz RapidIO at 16 and 32 bit elsewhere from Moto. How far away is this?
Next set....
Someone had a link to Ars - specificly BadAndy's comments on the possibility of an Apple/Sun combination. Mostly what seems to have been discussed here has been the possibility of a POWER usage and/or AMD production. Would there be any benefit to going with SPARC? How about a MIPS offering? Someone else entirely (who else is doing RISC developement? PA-RISC?)?
I don't waste my time listing what's wrong with you....so what's up with the insults? I don't even know who the fu[lk you are, but there you go with a post totally devoid of content save a an unprovoked put-down.
Get a life, buddy. Step outside, have a walk, enjoy the sun, turn off the computer for a few hours a day. It's hard at first, but it will get easier.
Now go and try to think of something to add to this thread worth reading. After you do, maybe you should post it. I'd like to read it, at least!
Good luck!</strong><hr></blockquote>
JD, I am always entertained by what you write. I did not intend that post to be an insult. (I would like you to remember that you have 10x the post count that I do, however, and telling ME to get off my computer or post something meaningful is a little hypocritical now isn't it?)
I would like to officially apologize for any hard feelings my post may have caused, intended or not.
As for this thread, I guess I'll just say again what everyone else said,
An event is coming up to release new powermacs, that is assumed.
Eventually apple will have to release something more powerful than they have right now, especially if they want the high end video/3d market.
that is assumed.
This thread makes it seem that this more powerful computer (or the processor inside) has a timeline of whenever OS X is at 40% user base.
Apples choices are to: Stick with moto, move to IBM's modified POWER 4 or 5, Move to some new AMD technology, Move to IBM for a new technology (possibly with the Altivec bought from moto). Either that, or they will do something COMPLETLY unexpected.
Right now, I'm hoping for a new IBM tech, but you never know.
Yeah timing is everything. It's like, right now Apple is in the back of the limo with the Prom queen trying to get it up, and the prom queen is kind, loving, understanding, and fairly patient. But if Apple doesn't get it up soon, then the Prom Queen will be more than happy to give her virginity to the meathead quarterback jock asshole who smacks her ass everyday and calls her his little bitch.
It's looking more and more like Apple's not going to get it up in time....even 12 throbbing inches doesn't cut it if she's already left for the jock's backseat and he's happily driving all 5 inches of wilt into her sweaty ass while she screams for more, more, more!!! Too late, she coulda' had more, but Apple was too slow to get their 12" up and throbbing.</strong><hr></blockquote>
You forgot one important item, "it's not size that matters but how you use your tongue". OS X and design flare is a big enough tongue to entertain the prom queen while stroking to arouse the 12 inches. This can buy a lot of time and has in the past.
[quote] think the point isn't so much we will see a POWER 4 come to the mac as much as something that may pay a passing resemblance.
Really currently if you want to talk about IBM's PPC processors only 2 spring quickly to mind to most people here; G3, POWER series.
Certainly most people don't wish to compare the G5 to current G3 tech even if IBM has a decent roadmap ahead for it.
The POWER series, however, shows what IBM can do when they set their mind to it. I agree it's a server chip, and won't be seen in a desktop computer as it currently stands. The POWER 5 will be more heavily aimed at workstations and cheaper products but still even then it would need some alterations most likely.
This doesn't necessarily mean IBM can't aim to design something more desktop oriented using what they already have though. Of course they could also just build something near entirely from scratch. Taking the POWER 4/5 chip, which at it's core is a relatively good design, then working it into a next generation processor for Apple might be more cost effective.
<hr></blockquote>
Look for a POWER4-core tower that will be released by this time next year if not sooner. There is no G5, it is the POWER4 core that is our saviour tho Apple may call it the G5. And a little AMD somethin' somethin' as well as prototypes of both (IBM and AMD) are floating around. So faster G4s with same processor but with DDR and faster bus in a new case next month and around the same in January only faster. Wait till next summer for the splash.
<strong>Some questions for all the CPU oriented folk here...
After looking at the specs (proposed, I suppose) of the MPC8560, I started wondering why Apple couldn't use this chip. Anyone know?
I seem to recall that someone around here said the e500 was a slow core, but the listed Drystone puts it about 15% higher than the current G4, so what makes it so slow? Or is this just a higher bandwidth improvement, meaning the core isn't data starved?
How difficult would it be to tack in a second core? Would that make any real difference? How about another core entirely?
So the SIMD on the e500 isn't AltiVec, but what is it? How does it compare to AltiVec, especially</strong><hr></blockquote>
The e500 core in the 8560 is severely limited for desktop use. It can manage to keep its integer units fed and achieve a good Dhrystone score, but basically:
It has no FPU.
It is only two-way superscalar (no more than two instructions can be issued at a time).
The APUs (auxiliary processing units) such as the SIMD unit must share the integer registers and instruction issue. (This effectively rules out adding an effective FPU as well.
The SIMD unit has nowhere near the flexibility of AltiVec.
On a 0.13 micron process it is limited to 1GHz. (This because it is an extremely low power chip), the G4 can easily reach 1GHz on a 0.18 micron process.
I do, however, think it is a beautiful piece of engineering, and marvellous for its intended market.
It should be (relatively) straightforward to put two e500 cores on one chip, talking of other cores, well, you have to ask Motorola what they have available.
[quote]
<strong>
The connection (RapidIO) lists at 500mhz at 8 bit. Is 8 bit standard for this type of connection? How does this compare to AMD/Intel's FSB? I also saw a listing for 1ghz RapidIO at 16 and 32 bit elsewhere from Moto. How far away is this? </strong><hr></blockquote>
RapidIO comes in multiples of 8 bits, there is no standard for this type of connection, but note that 8 bit 500MHz RapidIO is equivalent to DDR62.5MHz(effective 125MHz) standard 64 bit bus in terms of bandwidth (although RapidIO is full duplex and can send data in both directions at once). I have no idea yet of the relative overhead of RapidIO compared to MPX for transferring data. No information on when RapidIO will get to 1GHz, but 16 bit is just around the corner.
[quote]
<strong>
Next set....
Someone had a link to Ars - specificly BadAndy's comments on the possibility of an Apple/Sun combination. Mostly what seems to have been discussed here has been the possibility of a POWER usage and/or AMD production. Would there be any benefit to going with SPARC? How about a MIPS offering? Someone else entirely (who else is doing RISC developement? PA-RISC?)?
Just thought we needed more to discuss..... </strong><hr></blockquote>
A few quick notes: the performance of SPARC is nowhere near good enough to want to shift to it, MIPS is probably on its last legs as SGI cannot afford to keep developing it, PA-RISC development has been sidelined by HP in favour of Itanium (co-developed by HP and Intel engineers). Apple will not even consider transferring to a manufacturer whose future is not rock solid, and personally, I believe they have far too much invested in PPC and AltiVec (whose instruction set is pretty much a straight Apple development) to move away in the foreseable future, but there are a lot of foundries around.
It seems to me that, regardless of who makes the new chips, Apple has to make sure they do exactly what Apple wants them to do. Some kind of short term contract will have to be signed by Apple. Apple can't sign anything until they are absolutely positive that the chip does what Apple wants.
The rumours of various chips in prototype machines may (must) be true, and, if so, there may have been problems between the chips and what Apple wants for it's view of the future. Moreover, Apples views may have changed over time putting more requirements on the chip manufacturer (especially given the long time period that Apple has been trying to get something better and faster going).
Therefore, it is both Apple and the chip manufactures that have caused delays. No doubt the problems have been on both sides at different times. A lot of these problems are probably technical in nature between what yopu would like, think that can be done and can actually be made at a reasonable cost.
Before I signed a contract for millions of dollars I would make sure I was getting EXACTLY what I want.
So, take it easy on Apple they are trying. SJ doesn't like to get up on the stage with one-foot in the grave, but he also doesn't want to dissappoint his customers, who, for the most part, have been loyal to Apple (even at t he 3% to 5% level of PC market).
Hopefully he has our best interest at heart (and his too)
Seriously get your head out of your pants and quit the swearing. Not all of us want to see it and you sound like a kid who has never been laid Grow up.</strong><hr></blockquote>
Well I've been laid, just not for a while. But I think you're right, I should go wash my mouth out with soap. I won't talk about the prom queen's sweaty ass any more. Or Apple's throbbing 12". Or the "tongue" of Aqua and OS X that Apple is using to drive the Prom Queen insane from orgasmic pleasure.
I still say that we don't need a radical shift in CPU architecture to solve the performance gap. Apple can stick with PPC and deliver on that throbbing, eh, sorry, on that screaming CPU.
I still say that we don't need a radical shift in CPU architecture to solve the performance gap. Apple can stick with PPC and deliver on that throbbing, eh, sorry, on that screaming CPU.</strong><hr></blockquote>
But in order to do that they're going to need to make a major change somewhere (buy MOT assets... switch to POWER architecture...etc) because the G4 can't be scaled and by the look of this rumored new heat sink, overclocked forever. Apple's gonna do something... Apple's gotta do something... IMHO (and proabably some other people and companies) it needs to do whatever it's gonna do sonner than later.
I suggest the "heat sink: zealots go down and get a basic book on heat flow and thermodyanamics. Cooling efficiiency is not dependent on the weight of the object. A simple example is the heat shield tiles on the space shuttle. These blocks of silicon are lightweight and a block 4" x 6" x 2" weighs less than a lb each.
A bunch of air holes all over the box will not enhance cooling, it will allow for air to blow through the box without removing the heat trapped in all the knooks and crannies.
I realize this is a rumour board but you have to have some physical scientific basis for statements, although most just like to make statements on what they want to believe and not what is realistic, which is not rumours but pure BS. Period.....
So, go on, make up some other stupid stories about this idiotic "TurboCooling"
One question though, not that this has any direct relationship to any cpu that would be used in a desktop computer.
According to the documentation the MPC 8560 does single precision floating point in the SIMD unit , if I read it correctly. How is this possible without a FP unit?
<strong>Thank you for the description of the MPC 8560.
One question though, not that this has any direct relationship to any cpu that would be used in a desktop computer.
According to the documentation the MPC 8560 does single precision floating point in the SIMD unit , if I read it correctly. How is this possible without a FP unit?</strong><hr></blockquote>
The AltiVec unit does single precision floating point in the vector units too, but it is unrelated to the FPU. Basically it means that the traditional PowerPC floating point instructions haven't been implemented, but some of the 222 "new SIMD" instructions include floating point operations which are implemented somewhere in the eCORE. My understanding is that the eCORE only has one register set (the "integer" registers) and all the new instructions operate on those. This simplifies the machine state and execution units somewhat, speeding up interrupts and context switching -- all intended for the embedded market.
If there is a Motorola 85xx G5 desktop processor, its core isn't the eCORE.
A few quick notes: the performance of SPARC is nowhere near good enough to want to shift to it, MIPS is probably on its last legs as SGI cannot afford to keep developing it, PA-RISC development has been sidelined by HP in favour of Itanium (co-developed by HP and Intel engineers). Apple will not even consider transferring to a manufacturer whose future is not rock solid, and personally, I believe they have far too much invested in PPC and AltiVec (whose instruction set is pretty much a straight Apple development) to move away in the foreseable future, but there are a lot of foundries around.
Michael</strong><hr></blockquote>
Many thanx for the wonderful answer(s).
Now the question is, how difficult is it to drop a different core into what is "essentially" this chip? ie All the built in benefits, just a different proccessing core.
Isn't BookE supposed to make all this stuff interchangeable?
If so, then what about a modified Power4 core? How difficult would this be?
[quote] I suggest the "heat sink: zealots go down and get a basic book on heat flow and thermodyanamics. Cooling efficiiency is not dependent on the weight of the object. A simple example is the heat shield tiles on the space shuttle. These blocks of silicon are lightweight and a block 4" x 6" x 2" weighs less than a lb each.
<hr></blockquote>
Then why is it that hot CPUs like the pentium 4 have big heatsinks?
I think people are taking the "7 lbs" too literally. It probably just means a huge heatsink, since I doubt the guy reporting on it actually weighed it.
A bigger heatsink will dissapate heat better than a smaller heatsink, if they are designed well. I don't see any reason to assume that Apple didn't design a good heat sink.
Oh, and the heat tiles on the space shuttle are designed to BLOCK heat, which requires a porus material. It makes sense that they don't weight much. A heatsink must dissapate heat, a much different task...since aluminum is an excellent heat conductor and is used in all consumer heat sinks, most heatsinks tend to weigh more as they get larger.
<strong>I suggest the "heat sink: zealots go down and get a basic book on heat flow and thermodyanamics. Cooling efficiiency is not dependent on the weight of the object. A simple example is the heat shield tiles on the space shuttle. These blocks of silicon are lightweight and a block 4" x 6" x 2" weighs less than a lb each.
A bunch of air holes all over the box will not enhance cooling, it will allow for air to blow through the box without removing the heat trapped in all the knooks and crannies.
I realize this is a rumour board but you have to have some physical scientific basis for statements, although most just like to make statements on what they want to believe and not what is realistic, which is not rumours but pure BS. Period.....
So, go on, make up some other stupid stories about this idiotic "TurboCooling"</strong><hr></blockquote>
Cooling capacity isn't dependent on weight but there is a simple case of if you design a larger heat exchanger more often than not you can also design it to cool your system more as well. I haven't designed one since uni but I still remember that much
The perforations are well enough explained in another thread by mmicist but in heat exchangers more coolant flowing = greater heat exchange. Improve the air flow and ultimately you will improve the case temperature.
Now the question is, how difficult is it to drop a different core into what is "essentially" this chip? ie All the built in benefits, just a different proccessing core.
Isn't BookE supposed to make all this stuff interchangeable?
If so, then what about a modified Power4 core? How difficult would this be?</strong><hr></blockquote>
Yes, Motorola have worked hard on making their system on chip (SOC) work using drop-in bits. You could drop in another core, but it would have to be designed to be BookE, and almost certainly Motorola specific as well, you couldn't easily take someone else's core and drop it in. The cores themselves are full custom designs, not any form of plug and play system.
However, there have been hints from Motorola (mention of 64 bit chips etc.) that there is another eCORE being developed which could be dropped in and use at least some of the goodies in the 8560, although no hard data yet.
[quote]Yeah, but the G5 may not have altivec on it<hr></blockquote>
Future PowerPC processors are looking quite modular (I remember somone mentioning OCEAN a while back).
The next 20 will come quickly because OS X is no longer "optional" for new Macs.
[quote]A Power4 type PPC processor could handle this new software very well.<hr></blockquote>
[quote]Look for a POWER4-core tower that will be released by this time next year if not sooner<hr></blockquote>
How fast would a reasonable desktop Power4 be? Remove 3 of the four cores, most of the L3 cache, halve the L2 cache to 768KB, no Altivec but does have a better bus than the G4. Actually looks fairly interesting, especially when the Power4 is manufactured at 0.18 microns and reaches 1.3GHz. If Altivec could be added cheaply to the Power4 (highly unlikely ) then this could work. Otherwise, too much of Apple's software benefits heavily from Altivec.
[quote]Gaming on the Mac lives because every Mac can run the latest games. <hr></blockquote>
Will my Performa 6400/G3 + Voodoo 3 run Warcraft 3? Don't know, but I'll find out . While I'm on the subject of WC3, thanks to Blizzard & their publishers for coming out with a similtaneous Windows/Mac release and for the CD being dual format. I haven't bought (or been able to buy) a Mac game from a high street store since Starcraft.
RS/6000 machines (Deep Blue is an RS/6000 cluster) can have a variety of processors. Shouldn't take much effort to work out what Deep Blue uses:
I know it's not much, but I'm in the "industry" and was at Moto Austin, and just happened to mention to an engineer the rumor I heard about Apple leaving moto. With out missing a beat he said, "yeah, they're moving to IBM".
Take that with a grain of salt but my guess is there will be no "new moto chips".
<strong>I know it's not much, but I'm in the "industry" and was at Moto Austin, and just happened to mention to an engineer the rumor I heard about Apple leaving moto. With out missing a beat he said, "yeah, they're moving to IBM".
Take that with a grain of salt but my guess is there will be no "new moto chips".</strong><hr></blockquote>
shhh, people around here don't like hearing from people who are actually in the industry, they'd rather hear random speculation from people with real insight such as graphic designers and web developers . And my friends at MOS13 and MOS12 have said the same thing.
If they'd said 'mongo heat sink' or something similar, we'd all nod our heads. But 7lbs is INSANE. Do a google search for fridge cooling units less than 7 lbs... and start wading through a LOT of matches. Now search for mongo heat sinks and check weights.
The other insane part of it is it was described (though there was no picture). Something like: 7" by 6" by 6.5". Any sort of heat fin that's 6" long off a measly 2x2" heat source is INSANE. Even four POWER4 CPUs ganged together aren't going to produce a high enough temperature flux or gradient in that sucker to make it worthwhile for a 6" fin. Long before you get to 6" on a fin you're going to slam another fan in there mounted right on the heat sink.
Now, if it's not a heat sink, but a fridge, or a proprietary 'black box' or a prototype anti-tampering camera or whatever, fine. But there's just no way there's a heat sink even half that size in there. Even a quarter that volume or weight would be pretty astonishing.
So, '7lbs' means 'mongo', it was meant to be 0.7lbs, it's not a heat sink, or it's complete fluff.
Comments
After looking at the specs (proposed, I suppose) of the MPC8560, I started wondering why Apple couldn't use this chip. Anyone know?
I seem to recall that someone around here said the e500 was a slow core, but the listed Drystone puts it about 15% higher than the current G4, so what makes it so slow? Or is this just a higher bandwidth improvement, meaning the core isn't data starved?
How difficult would it be to tack in a second core? Would that make any real difference? How about another core entirely?
So the SIMD on the e500 isn't AltiVec, but what is it? How does it compare to AltiVec, especially with these 222 added new instructions?
The connection (RapidIO) lists at 500mhz at 8 bit. Is 8 bit standard for this type of connection? How does this compare to AMD/Intel's FSB? I also saw a listing for 1ghz RapidIO at 16 and 32 bit elsewhere from Moto. How far away is this?
Next set....
Someone had a link to Ars - specificly BadAndy's comments on the possibility of an Apple/Sun combination. Mostly what seems to have been discussed here has been the possibility of a POWER usage and/or AMD production. Would there be any benefit to going with SPARC? How about a MIPS offering? Someone else entirely (who else is doing RISC developement? PA-RISC?)?
Just thought we needed more to discuss.....
<strong>
I don't waste my time listing what's wrong with you....so what's up with the insults? I don't even know who the fu[lk you are, but there you go with a post totally devoid of content save a an unprovoked put-down.
Get a life, buddy. Step outside, have a walk, enjoy the sun, turn off the computer for a few hours a day. It's hard at first, but it will get easier.
Now go and try to think of something to add to this thread worth reading. After you do, maybe you should post it. I'd like to read it, at least!
Good luck!</strong><hr></blockquote>
JD, I am always entertained by what you write. I did not intend that post to be an insult. (I would like you to remember that you have 10x the post count that I do, however, and telling ME to get off my computer or post something meaningful is a little hypocritical now isn't it?)
I would like to officially apologize for any hard feelings my post may have caused, intended or not.
As for this thread, I guess I'll just say again what everyone else said,
An event is coming up to release new powermacs, that is assumed.
Eventually apple will have to release something more powerful than they have right now, especially if they want the high end video/3d market.
that is assumed.
This thread makes it seem that this more powerful computer (or the processor inside) has a timeline of whenever OS X is at 40% user base.
Apples choices are to: Stick with moto, move to IBM's modified POWER 4 or 5, Move to some new AMD technology, Move to IBM for a new technology (possibly with the Altivec bought from moto). Either that, or they will do something COMPLETLY unexpected.
Right now, I'm hoping for a new IBM tech, but you never know.
<strong>
Yeah timing is everything. It's like, right now Apple is in the back of the limo with the Prom queen trying to get it up, and the prom queen is kind, loving, understanding, and fairly patient. But if Apple doesn't get it up soon, then the Prom Queen will be more than happy to give her virginity to the meathead quarterback jock asshole who smacks her ass everyday and calls her his little bitch.
It's looking more and more like Apple's not going to get it up in time....even 12 throbbing inches doesn't cut it if she's already left for the jock's backseat and he's happily driving all 5 inches of wilt into her sweaty ass while she screams for more, more, more!!! Too late, she coulda' had more, but Apple was too slow to get their 12" up and throbbing.</strong><hr></blockquote>
You forgot one important item, "it's not size that matters but how you use your tongue". OS X and design flare is a big enough tongue to entertain the prom queen while stroking to arouse the 12 inches. This can buy a lot of time and has in the past.
[quote] think the point isn't so much we will see a POWER 4 come to the mac as much as something that may pay a passing resemblance.
Really currently if you want to talk about IBM's PPC processors only 2 spring quickly to mind to most people here; G3, POWER series.
Certainly most people don't wish to compare the G5 to current G3 tech even if IBM has a decent roadmap ahead for it.
The POWER series, however, shows what IBM can do when they set their mind to it. I agree it's a server chip, and won't be seen in a desktop computer as it currently stands. The POWER 5 will be more heavily aimed at workstations and cheaper products but still even then it would need some alterations most likely.
This doesn't necessarily mean IBM can't aim to design something more desktop oriented using what they already have though. Of course they could also just build something near entirely from scratch. Taking the POWER 4/5 chip, which at it's core is a relatively good design, then working it into a next generation processor for Apple might be more cost effective.
<hr></blockquote>
Look for a POWER4-core tower that will be released by this time next year if not sooner. There is no G5, it is the POWER4 core that is our saviour tho Apple may call it the G5. And a little AMD somethin' somethin' as well as prototypes of both (IBM and AMD) are floating around. So faster G4s with same processor but with DDR and faster bus in a new case next month and around the same in January only faster. Wait till next summer for the splash.
<strong>Some questions for all the CPU oriented folk here...
After looking at the specs (proposed, I suppose) of the MPC8560, I started wondering why Apple couldn't use this chip. Anyone know?
I seem to recall that someone around here said the e500 was a slow core, but the listed Drystone puts it about 15% higher than the current G4, so what makes it so slow? Or is this just a higher bandwidth improvement, meaning the core isn't data starved?
How difficult would it be to tack in a second core? Would that make any real difference? How about another core entirely?
So the SIMD on the e500 isn't AltiVec, but what is it? How does it compare to AltiVec, especially</strong><hr></blockquote>
The e500 core in the 8560 is severely limited for desktop use. It can manage to keep its integer units fed and achieve a good Dhrystone score, but basically:
It has no FPU.
It is only two-way superscalar (no more than two instructions can be issued at a time).
The APUs (auxiliary processing units) such as the SIMD unit must share the integer registers and instruction issue. (This effectively rules out adding an effective FPU as well.
The SIMD unit has nowhere near the flexibility of AltiVec.
On a 0.13 micron process it is limited to 1GHz. (This because it is an extremely low power chip), the G4 can easily reach 1GHz on a 0.18 micron process.
I do, however, think it is a beautiful piece of engineering, and marvellous for its intended market.
It should be (relatively) straightforward to put two e500 cores on one chip, talking of other cores, well, you have to ask Motorola what they have available.
[quote]
<strong>
The connection (RapidIO) lists at 500mhz at 8 bit. Is 8 bit standard for this type of connection? How does this compare to AMD/Intel's FSB? I also saw a listing for 1ghz RapidIO at 16 and 32 bit elsewhere from Moto. How far away is this? </strong><hr></blockquote>
RapidIO comes in multiples of 8 bits, there is no standard for this type of connection, but note that 8 bit 500MHz RapidIO is equivalent to DDR62.5MHz(effective 125MHz) standard 64 bit bus in terms of bandwidth (although RapidIO is full duplex and can send data in both directions at once). I have no idea yet of the relative overhead of RapidIO compared to MPX for transferring data. No information on when RapidIO will get to 1GHz, but 16 bit is just around the corner.
[quote]
<strong>
Next set....
Someone had a link to Ars - specificly BadAndy's comments on the possibility of an Apple/Sun combination. Mostly what seems to have been discussed here has been the possibility of a POWER usage and/or AMD production. Would there be any benefit to going with SPARC? How about a MIPS offering? Someone else entirely (who else is doing RISC developement? PA-RISC?)?
Just thought we needed more to discuss.....
A few quick notes: the performance of SPARC is nowhere near good enough to want to shift to it, MIPS is probably on its last legs as SGI cannot afford to keep developing it, PA-RISC development has been sidelined by HP in favour of Itanium (co-developed by HP and Intel engineers). Apple will not even consider transferring to a manufacturer whose future is not rock solid, and personally, I believe they have far too much invested in PPC and AltiVec (whose instruction set is pretty much a straight Apple development) to move away in the foreseable future, but there are a lot of foundries around.
Michael
The rumours of various chips in prototype machines may (must) be true, and, if so, there may have been problems between the chips and what Apple wants for it's view of the future. Moreover, Apples views may have changed over time putting more requirements on the chip manufacturer (especially given the long time period that Apple has been trying to get something better and faster going).
Therefore, it is both Apple and the chip manufactures that have caused delays. No doubt the problems have been on both sides at different times. A lot of these problems are probably technical in nature between what yopu would like, think that can be done and can actually be made at a reasonable cost.
Before I signed a contract for millions of dollars I would make sure I was getting EXACTLY what I want.
So, take it easy on Apple they are trying. SJ doesn't like to get up on the stage with one-foot in the grave, but he also doesn't want to dissappoint his customers, who, for the most part, have been loyal to Apple (even at t he 3% to 5% level of PC market).
Hopefully he has our best interest at heart (and his too)
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
<strong>
Seriously get your head out of your pants and quit the swearing. Not all of us want to see it and you sound like a kid who has never been laid
Well I've been laid, just not for a while. But I think you're right, I should go wash my mouth out with soap. I won't talk about the prom queen's sweaty ass any more. Or Apple's throbbing 12". Or the "tongue" of Aqua and OS X that Apple is using to drive the Prom Queen insane from orgasmic pleasure.
I still say that we don't need a radical shift in CPU architecture to solve the performance gap. Apple can stick with PPC and deliver on that throbbing, eh, sorry, on that screaming CPU.
<strong>
I still say that we don't need a radical shift in CPU architecture to solve the performance gap. Apple can stick with PPC and deliver on that throbbing, eh, sorry, on that screaming CPU.</strong><hr></blockquote>
But in order to do that they're going to need to make a major change somewhere (buy MOT assets... switch to POWER architecture...etc) because the G4 can't be scaled and by the look of this rumored new heat sink, overclocked forever. Apple's gonna do something... Apple's gotta do something... IMHO (and proabably some other people and companies) it needs to do whatever it's gonna do sonner than later.
A bunch of air holes all over the box will not enhance cooling, it will allow for air to blow through the box without removing the heat trapped in all the knooks and crannies.
I realize this is a rumour board but you have to have some physical scientific basis for statements, although most just like to make statements on what they want to believe and not what is realistic, which is not rumours but pure BS. Period.....
So, go on, make up some other stupid stories about this idiotic "TurboCooling"
[ 07-28-2002: Message edited by: Bigc ]</p>
"It has no FPU."<hr></blockquote>
Thank you for the description of the MPC 8560.
One question though, not that this has any direct relationship to any cpu that would be used in a desktop computer.
According to the documentation the MPC 8560 does single precision floating point in the SIMD unit , if I read it correctly. How is this possible without a FP unit?
<strong>Thank you for the description of the MPC 8560.
One question though, not that this has any direct relationship to any cpu that would be used in a desktop computer.
According to the documentation the MPC 8560 does single precision floating point in the SIMD unit , if I read it correctly. How is this possible without a FP unit?</strong><hr></blockquote>
The AltiVec unit does single precision floating point in the vector units too, but it is unrelated to the FPU. Basically it means that the traditional PowerPC floating point instructions haven't been implemented, but some of the 222 "new SIMD" instructions include floating point operations which are implemented somewhere in the eCORE. My understanding is that the eCORE only has one register set (the "integer" registers) and all the new instructions operate on those. This simplifies the machine state and execution units somewhat, speeding up interrupts and context switching -- all intended for the embedded market.
If there is a Motorola 85xx G5 desktop processor, its core isn't the eCORE.
<strong>
A few quick notes: the performance of SPARC is nowhere near good enough to want to shift to it, MIPS is probably on its last legs as SGI cannot afford to keep developing it, PA-RISC development has been sidelined by HP in favour of Itanium (co-developed by HP and Intel engineers). Apple will not even consider transferring to a manufacturer whose future is not rock solid, and personally, I believe they have far too much invested in PPC and AltiVec (whose instruction set is pretty much a straight Apple development) to move away in the foreseable future, but there are a lot of foundries around.
Michael</strong><hr></blockquote>
Many thanx for the wonderful answer(s).
Now the question is, how difficult is it to drop a different core into what is "essentially" this chip? ie All the built in benefits, just a different proccessing core.
Isn't BookE supposed to make all this stuff interchangeable?
If so, then what about a modified Power4 core? How difficult would this be?
<hr></blockquote>
Then why is it that hot CPUs like the pentium 4 have big heatsinks?
I think people are taking the "7 lbs" too literally. It probably just means a huge heatsink, since I doubt the guy reporting on it actually weighed it.
A bigger heatsink will dissapate heat better than a smaller heatsink, if they are designed well. I don't see any reason to assume that Apple didn't design a good heat sink.
Oh, and the heat tiles on the space shuttle are designed to BLOCK heat, which requires a porus material. It makes sense that they don't weight much. A heatsink must dissapate heat, a much different task...since aluminum is an excellent heat conductor and is used in all consumer heat sinks, most heatsinks tend to weigh more as they get larger.
<strong>I suggest the "heat sink: zealots go down and get a basic book on heat flow and thermodyanamics. Cooling efficiiency is not dependent on the weight of the object. A simple example is the heat shield tiles on the space shuttle. These blocks of silicon are lightweight and a block 4" x 6" x 2" weighs less than a lb each.
A bunch of air holes all over the box will not enhance cooling, it will allow for air to blow through the box without removing the heat trapped in all the knooks and crannies.
I realize this is a rumour board but you have to have some physical scientific basis for statements, although most just like to make statements on what they want to believe and not what is realistic, which is not rumours but pure BS. Period.....
So, go on, make up some other stupid stories about this idiotic "TurboCooling"</strong><hr></blockquote>
Cooling capacity isn't dependent on weight but there is a simple case of if you design a larger heat exchanger more often than not you can also design it to cool your system more as well. I haven't designed one since uni but I still remember that much
The perforations are well enough explained in another thread by mmicist but in heat exchangers more coolant flowing = greater heat exchange. Improve the air flow and ultimately you will improve the case temperature.
<strong>
Many thanx for the wonderful answer(s).
Now the question is, how difficult is it to drop a different core into what is "essentially" this chip? ie All the built in benefits, just a different proccessing core.
Isn't BookE supposed to make all this stuff interchangeable?
If so, then what about a modified Power4 core? How difficult would this be?</strong><hr></blockquote>
Yes, Motorola have worked hard on making their system on chip (SOC) work using drop-in bits. You could drop in another core, but it would have to be designed to be BookE, and almost certainly Motorola specific as well, you couldn't easily take someone else's core and drop it in. The cores themselves are full custom designs, not any form of plug and play system.
However, there have been hints from Motorola (mention of 64 bit chips etc.) that there is another eCORE being developed which could be dropped in and use at least some of the goodies in the 8560, although no hard data yet.
Michael
Future PowerPC processors are looking quite modular (I remember somone mentioning OCEAN a while back).
The next 20 will come quickly because OS X is no longer "optional" for new Macs.
[quote]A Power4 type PPC processor could handle this new software very well.<hr></blockquote>
[quote]Look for a POWER4-core tower that will be released by this time next year if not sooner<hr></blockquote>
How fast would a reasonable desktop Power4 be? Remove 3 of the four cores, most of the L3 cache, halve the L2 cache to 768KB, no Altivec but does have a better bus than the G4. Actually looks fairly interesting, especially when the Power4 is manufactured at 0.18 microns and reaches 1.3GHz. If Altivec could be added cheaply to the Power4 (highly unlikely
[quote]Gaming on the Mac lives because every Mac can run the latest games. <hr></blockquote>
Will my Performa 6400/G3 + Voodoo 3 run Warcraft 3? Don't know, but I'll find out
RS/6000 machines (Deep Blue is an RS/6000 cluster) can have a variety of processors. Shouldn't take much effort to work out what Deep Blue uses:
32-bit 604e,
64-bit POWER3-II,
64-bit RS64 IV
64-bit POWER4 64-bit RS64-III,
64-bit RS64 IV,
64-bit POWER4 64-bit RS64-III,
64-bit RS64 IV,
64-bit POWER4
Take that with a grain of salt but my guess is there will be no "new moto chips".
<strong>I know it's not much, but I'm in the "industry" and was at Moto Austin, and just happened to mention to an engineer the rumor I heard about Apple leaving moto. With out missing a beat he said, "yeah, they're moving to IBM".
Take that with a grain of salt but my guess is there will be no "new moto chips".</strong><hr></blockquote>
shhh, people around here don't like hearing from people who are actually in the industry, they'd rather hear random speculation from people with real insight such as graphic designers and web developers
If they'd said 'mongo heat sink' or something similar, we'd all nod our heads. But 7lbs is INSANE. Do a google search for fridge cooling units less than 7 lbs... and start wading through a LOT of matches. Now search for mongo heat sinks and check weights.
<a href="http://www.coolinnovations.com/products/heatSinksFor_cpu.asp"; target="_blank">Heat Sinks</a> Note the longest pins are 1.7".
The other insane part of it is it was described (though there was no picture). Something like: 7" by 6" by 6.5". Any sort of heat fin that's 6" long off a measly 2x2" heat source is INSANE. Even four POWER4 CPUs ganged together aren't going to produce a high enough temperature flux or gradient in that sucker to make it worthwhile for a 6" fin. Long before you get to 6" on a fin you're going to slam another fan in there mounted right on the heat sink.
Now, if it's not a heat sink, but a fridge, or a proprietary 'black box' or a prototype anti-tampering camera or whatever, fine. But there's just no way there's a heat sink even half that size in there. Even a quarter that volume or weight would be pretty astonishing.
So, '7lbs' means 'mongo', it was meant to be 0.7lbs, it's not a heat sink, or it's complete fluff.