Who was that guy that predicted that CPUs wouldn't get any 'faster'
There was a guy on these boards a few months back who claimed that he had insider knowledge, and he predicted that the CPUs wouldn't get any faster - he claimed that all the manufacturers were looking to multiple cores as the future, and that they had given up trying to up the clock speeds of processors.
He was shouted down by almost everyone here, and shortly thereafter Intel annouced that it was scrapping it's faster processors and concentrating on multi-cores.
I can't remember the guys name, but from where I'm sitting, it looks as though he was spot on the money...
He was shouted down by almost everyone here, and shortly thereafter Intel annouced that it was scrapping it's faster processors and concentrating on multi-cores.
I can't remember the guys name, but from where I'm sitting, it looks as though he was spot on the money...
Comments
Originally posted by Messiah
There was a guy on these boards a few months back who claimed that he had insider knowledge, and he predicted that the CPUs wouldn't get any faster - he claimed that all the manufacturers were looking to multiple cores as the future, and that they had given up trying to up the clock speeds of processors.
He was shouted down by almost everyone here, and shortly thereafter Intel annouced that it was scrapping it's faster processors and concentrating on multi-cores.
Yup...that was Nr9 alright. But I remember him more for his bold prediction that Apple's so-called "Powerbook G5" wouldn't have a "G5" in it at all, but actually four separate, interconnected souped-up PPC 440-series SOC's w/Altivec...and then there was the big brouhaha over his lack of spelling and grammatical skills.
I can't remember the guys name, but from where I'm sitting, it looks as though he was spot on the money...
What about Cell? Aren't people saying it could scale up to 3-4 Ghz and beyond? I understand it's not necessarily a traditional "desktop" CPU, but still...
It's hard to be wrong all the time.
Originally posted by Messiah
There was a guy on these boards a few months back who claimed that he had insider knowledge, and he predicted that the CPUs wouldn't get any faster - he claimed that all the manufacturers were looking to multiple cores as the future, and that they had given up trying to up the clock speeds of processors.
He was shouted down by almost everyone here, and shortly thereafter Intel annouced that it was scrapping it's faster processors and concentrating on multi-cores.
I can't remember the guys name, but from where I'm sitting, it looks as though he was spot on the money...
He had no insider knowledge. It was already very well known at the time that every major manufacturer was switching to multiple cores.
As an aside they haven't given up on clock speeds they just now realise it isn't the most efficient means of increasing performance anymore. They now have transistors to play with and the best way to add performance is to just add separate cores rather than massively complex single cores.
Originally posted by Telomar
He had no insider knowledge. It was already very well known at the time that every major manufacturer was switching to multiple cores.
As an aside they haven't given up on clock speeds they just now realise it isn't the most efficient means of increasing performance anymore. They now have transistors to play with and the best way to add performance is to just add separate cores rather than massively complex single cores.
actually it is still possible under simulation to achieve performance through ultra high frequency cores, it is just leakage current makes the power dissipation ejaculate.
want to kno why both mot G5 and power 6 high frequency designs got scrapped?
I don't know an awful lot about CPU design, and I seem to remember that a lot of the stuff he was talking about made sense... and seemed to come true.
But I'll take your word for it that it was common knowledge at that point...
...man I hope that IBM can pull a multi-core G5 out of the hat!
Originally posted by Nr9
want to kno why both mot G5 and power 6 high frequency designs got scrapped?
Last I heard of POWER6 ultra high frequezy, multi core processor, it was still a Go! And considering what IBM has done with Cell, I really can't see what's the problem. They have stated all along that POWER6 will be a radical new design and Cell/PPE seems to fit the profile and time frame quite nicely.
Originally posted by Henriok
Last I heard of POWER6 ultra high frequezy, multi core processor, it was still a Go! And considering what IBM has done with Cell, I really can't see what's the problem. They have stated all along that POWER6 will be a radical new design and Cell/PPE seems to fit the profile and time frame quite nicely.
While at an IBM technology briefing/demonstration, they showed off lotsa technology, mostly POWER5, POWER5+ and spoke of POWER6 designs. So, yes, the POWER6 is still a go. Not a lot of details, but this was a public forum.
As far as Nr9's insight goes, bahhhh...I have at much insight.
Quad 440's in a PB. M-kay. Be gone.
Faster, faster, faster. It will always work that way. Nothing will stand in the way of faster processors. Stumbling blocks, yes. Impenetrable, get real.
Originally posted by Henriok
And considering what IBM has done with Cell, I really can't see what's the problem. They have stated all along that POWER6 will be a radical new design and Cell/PPE seems to fit the profile and time frame quite nicely.
Keep in mind that Cell achieves its clock rates with hand optimized circuits and a much simplified core (compared to the OoOE monster than is POWER4/5/970).
Nr9 was too black&white about his statement (i.e. the clock will never go one Hz higher no matter what), but the general sentiment that the rapid "easy" frequency scaling of the past decade or two has run into a wall at the 90 nm node. Going forward process and design optimization will be much harder, and further frequency scaling will hard won. The design choices of Cell support that, as does Intel backing away from NetBurst, IBM missing the 3 GHz 970 by at least 2 years, and AMD scaling slowly and choosing to optimize through different mechanisms.
Originally posted by Performa636CD
how does he write like he's 9? duz he rite lik dis all the tyme?
He may well be foreign. On different keyboards, typing can be a pain in the ass, and I've noticed that intentionally butchered english seems to come most from foreigners, particularly from India and the Far East.
Originally posted by Programmer
[BNr9 was too black&white about his statement (i.e. the clock will never go one Hz higher no matter what), but the general sentiment that the rapid "easy" frequency scaling of the past decade or two has run into a wall at the 90 nm node. Going forward process and design optimization will be much harder, and further frequency scaling will hard won. The design choices of Cell support that, as does Intel backing away from NetBurst, IBM missing the 3 GHz 970 by at least 2 years, and AMD scaling slowly and choosing to optimize through different mechanisms. [/B]
Yes, NR9 was too black and white about his statement, even if many other people he was right to see that the Mhz competition was slowing down dramatically.
The futur is multicore and multithreaded. Nightmarish days are waiting programmers ...
Originally posted by Henriok
Last I heard of POWER6 ultra high frequezy, multi core processor, it was still a Go!
The ultra high frequency part is incorrect
Originally posted by emig647
He claimed he was an insider from IBM though. He claimed that apple would NOT break the 2.5ghz barrier. We went rounds on this one, and I wish he was still around... I believe he made a little bet
I am still around, do you find it opportunity to misquote me just cuz you think imn not around