Where did I misquote you? If I recall correctly, you DID say that the g5's wouldn't break 2.5ghz and you DID state that you worked for ibm? where's the misquote?
Where did I misquote you? If I recall correctly, you DID say that the g5's wouldn't break 2.5ghz and you DID state that you worked for ibm? where's the misquote?
i invite you to look at my original post regarding the 2.5GHz again.
It got brought back up because, whatever the merits of Nr9's prognostications, they nevertheless provoked some of the best discussions on this board.
His architectural claims in the original Powerbook thread (especially about cell) seem to have been on the button - even if no multicore laptop chip has yet appeared in an Apple product.
The main thing against him is his choice of name, based as it is on a John Lennon track: #9 Dream...
Correct, but that's his problem with the 970MP, it seems . By the way, not only Morpheus (the one who takes or makes certain forms) but his brothers too, Phobetor (the one who makes fear) and Phantasos (the one who makes unreal things, hence "fantasy") were responsible for the dreams of mortals. They are known as Oneiroi (dreamers or better dream makers), sons of Hypnos (sleep, god of sleep). If they can play some role in the rumor mill, that remains to be seen .
After the debut of Cell I'm not sure how anyone could argue about the processor described by NR9 as being a possibility. Frankly all it would take is to replace the <sorry> SPE's on Cell with multiple PPE's to give us what NR9 decribed. This wouldn't be 440 based of course but then details of the cores would likely be hard to come by.
I'm not a big believer in the idea that we can't continue to get frequency scaling. We simply need a procss break though that results in low leakage features on on the chips. Frankly I do not believe that it was even a goal at IBM to be leading edge process wise. They simply wnated to go to 90 nm first and did so without implementing a lot of new technology.
It is pretty much a given that the industry is going multiprocessor, that how ever is not completely tied to the difficuties in scaling. A number of realities ae converging right now to make multiprocessors and multi core hardware the development path in the future. One big issue is the reality that putting two core or even a large part of a whole system on a core is now possible and economical. Desktop OS's are now to the point where multiprocessing a parallel operations have matured enough that all hardware should be able to exploit or support the OS features. It doesn't hurt that tools are more or less here to. This boils down to one point, SMP hardware is now viable for use outside of the narrow ranges of implementaitons we saw a year or more ago.
As to Apple and where they are going, frankly I don't know at the moment. I would suggest though that SMP support in a PowerBook is likely to have priority at Apple. I'd also suggest that Apple is likely to be looking at several concepts to give them SMP support. They could very well stay 32 bit with the new Freescale core, more likely though they want to go 64 bit somehow. I would not be surprised oat all to see a core with two PPE's in a laptop when APPLE leaves tthe 32 bit world behind.
The clockspeed is not exactly leaving 2,5 when three out of four models are below that speed...
That's like saying that Intel isn't going to leave 3.6ghz even though they have 3.8ghz out. Leaving is a present tense word, hasn't happened yet but is in the process of it. The clock speed is "leaving" 2.5ghz to 2.7ghz and faster. So if apple hits 3.0ghz with next update this would make all this complete bs... no?
I could have told you that superscalar, sequentially driven CPUs are reaching their end, and that without dramatic enhancements in fabrication techniques clock speeds will be plateuing (i.e. not using photo-lithography).
Nr9 was right in assessing that "everyone knows this," because -- unless you're going to be precociously exacting -- everyone in the industry does know this, and we've seen clock speeds from all manufacturers level out. He was somewhat incorrect in his claim that the next powermac would be multicore, but it's entirely likely that the newly released powermac line is nothing more than a stopgap spurred by delays in developing the multicore G5.
I'd bet that the 2.7GHz G5's came from the same line as all of the 970fx chips, and are merely from a stockpile of lucky parts that has accumulated over the last year. In this sense, they only thing different about old 2.5GHz chips and the new 2.7GHz is their respective position on a probability density curve.
It could also be that in all of the improvements that IBM has been able to squeeze out of their 90nm process they've also improved their yields of faster parts. Who knows-- maybe they are actually getting yields of 3GHz parts!
...and placing each one of them in a solid gold trophy case.\
Comments
Originally posted by emig647
Where did I misquote you? If I recall correctly, you DID say that the g5's wouldn't break 2.5ghz and you DID state that you worked for ibm? where's the misquote?
i invite you to look at my original post regarding the 2.5GHz again.
So, what's next Nr9? Where are the 970MP and GX? What final form has the low-power laptop chip taken? Do you have a timeline?
Pip pip!
Originally posted by Nr9:
The latest news.
Power Mac will be stuck at rougly 2.5GHz for the next 3 years.
This is a known fact.
Every processor company is hitting a power wall and its not likely to change.
The next power mac are 2.5GHz multicore designs.
Let the readers decide.
Originally posted by New
I'd say 2,7 ghz is still roughly 2,5 ghz...
Perhaps, but we still have 2.5 years to go.
Edit:
Perhaps, but we still have 2.5 years to go.
The clockspeed is already leaving 2.5ghz and is 2.7ghz. I'm willing to bet will see 3ghz by years end.
I can't even believe this got brought back up.
Originally posted by emig647
Several things here to laugh at, it may get funnier if you can be more precise.
Originally posted by emig647
Edit:
The clockspeed is already leaving 2.5ghz and is 2.7ghz. I'm willing to bet will see 3ghz by years end.
I can't even believe this got brought back up.
we all wish right?
how about: no?
Originally posted by emig647
Edit:
The clockspeed is already leaving 2.5ghz and is 2.7ghz. I'm willing to bet will see 3ghz by years end.
I can't even believe this got brought back up.
The clockspeed is not exactly leaving 2,5 when three out of four models are below that speed...
Originally posted by emig647
I can't even believe this got brought back up.
It got brought back up because, whatever the merits of Nr9's prognostications, they nevertheless provoked some of the best discussions on this board.
His architectural claims in the original Powerbook thread (especially about cell) seem to have been on the button - even if no multicore laptop chip has yet appeared in an Apple product.
The main thing against him is his choice of name, based as it is on a John Lennon track: #9 Dream...
Morpheus, of course, is the god of dreams...
Pip pip!
Originally posted by Bucolic Old Sir Henry
Morpheus, of course, is the god of dreams...
Correct, but that's his problem with the 970MP, it seems
After the debut of Cell I'm not sure how anyone could argue about the processor described by NR9 as being a possibility. Frankly all it would take is to replace the <sorry> SPE's on Cell with multiple PPE's to give us what NR9 decribed. This wouldn't be 440 based of course but then details of the cores would likely be hard to come by.
I'm not a big believer in the idea that we can't continue to get frequency scaling. We simply need a procss break though that results in low leakage features on on the chips. Frankly I do not believe that it was even a goal at IBM to be leading edge process wise. They simply wnated to go to 90 nm first and did so without implementing a lot of new technology.
It is pretty much a given that the industry is going multiprocessor, that how ever is not completely tied to the difficuties in scaling. A number of realities ae converging right now to make multiprocessors and multi core hardware the development path in the future. One big issue is the reality that putting two core or even a large part of a whole system on a core is now possible and economical. Desktop OS's are now to the point where multiprocessing a parallel operations have matured enough that all hardware should be able to exploit or support the OS features. It doesn't hurt that tools are more or less here to. This boils down to one point, SMP hardware is now viable for use outside of the narrow ranges of implementaitons we saw a year or more ago.
As to Apple and where they are going, frankly I don't know at the moment. I would suggest though that SMP support in a PowerBook is likely to have priority at Apple. I'd also suggest that Apple is likely to be looking at several concepts to give them SMP support. They could very well stay 32 bit with the new Freescale core, more likely though they want to go 64 bit somehow. I would not be surprised oat all to see a core with two PPE's in a laptop when APPLE leaves tthe 32 bit world behind.
Dave
Originally posted by New
The clockspeed is not exactly leaving 2,5 when three out of four models are below that speed...
That's like saying that Intel isn't going to leave 3.6ghz even though they have 3.8ghz out. Leaving is a present tense word, hasn't happened yet but is in the process of it. The clock speed is "leaving" 2.5ghz to 2.7ghz and faster. So if apple hits 3.0ghz with next update this would make all this complete bs... no?
Nr9 was right in assessing that "everyone knows this," because -- unless you're going to be precociously exacting -- everyone in the industry does know this, and we've seen clock speeds from all manufacturers level out. He was somewhat incorrect in his claim that the next powermac would be multicore, but it's entirely likely that the newly released powermac line is nothing more than a stopgap spurred by delays in developing the multicore G5.
I'd bet that the 2.7GHz G5's came from the same line as all of the 970fx chips, and are merely from a stockpile of lucky parts that has accumulated over the last year. In this sense, they only thing different about old 2.5GHz chips and the new 2.7GHz is their respective position on a probability density curve.
...and placing each one of them in a solid gold trophy case.