I don't think Apple swapped to Intel to take on the X86 architecture. Time will tell.
Sure they did.
The developer docs are all about converting to 32bit X86, optimising for SSE2 or SSE3, endian issues. They aren't about some mythical new non-X86 architecture.
The developer docs are all about converting to 32bit X86, optimising for SSE2 or SSE3, endian issues. They aren't about some mythical new non-X86 architecture.
...and time will tell if that's always the case.
Thanks for that. Fair enough. By the way, is it 'mythical' or 'new'. 'New' I like!
I don't quite agree with you that a Dothan has 2x the performance/watt over a G5 but I'll agree with you that Intel are increasing the performance/watt ratio and quicker than IBM. It's near 1.5x now and Yonah would be about 3x which makes Merom at 6x provided the G5 stands still with no improvements.
How much quicker Intel gets there seems to be the whole debate though for Job's credibility. If Intel's Dev conference gives out that Merom/Conroe is just Yonah+ then I reckon the cred is shot already as I can't see IBM standing still and taking it. Their blade servers at least need low power 970MPs.
The 7448 is a good chip despite what the FSB speed weenies say - it makes up for it in other aspects, but yes, 9 months later than was needed for Apple to compete with Pentium M.
Something many people overlook is that the performance per watt stats were for integer performance, which x86 beats macs handily at already.
The Pentium 4 beats the PPC970 as far as max scalar integer performance, but I'm not sure if it beats it in as far as performance per watt. That would be interesting to see.
Also, there are plenty of integer tests where the 970 will mop the floor with the P4 in max performance and performance per watt. . . They just need to be tests that do four integer operations at the same time. (And I do think it's fair to use Altivec in integer tests, since applications compiled with IBM's compiler will indeed make decent use of it.) You don't build a car with a turbo and then release performance numbers with the turbo off.
Anyway, where there's smoke there's fire, and to me it's obvious that Intel is up to something; afterall, it is Intel's job to stay ahead of the curve. The more relevant questions regard their plans for releasing it, and if it will debut on the Intel Powermac or whatnot.
So don't expect anything earth-shaking. More like apologies clothed as strategies.
No offence, but the Register has led me down the path too many times. I can never figure out if they know something or if they just think they know something that is shocking.
Otellini showed the first public demonstration of Merom, Conroe and Woodcrest - processors for notebook, desktop and server platforms designed on Intel's advanced 65-nanometer technology manufacturing process. He also said Intel has more than 10 processor projects that contain four (quad-core) or more processor cores per chip.
Otellini also announced that forthcoming lower-power products will lead to a new category of ultra energy-efficient "Handtop PC" devices that provide a converged communication and PC-like experience but require less than a watt of processing power and weigh under a pound.
This is from this morning Intel's press release. Nothing new really.
As far as we think Apple is concerned, Yonah's already out for iBooks, mac minis and possibly PowerBooks for spring 2006, and the 64-bit trio mentioned above will probably go into the next-gen xBooks/minis (Merom), PowerMacs and iMacs (Conroe), and XServes (Woodcrest).
Don't know about availability yet...
Curious about what Apple would do with an "ultra energy-efficient "Handtop PC" devices that provide a converged communication and PC-like experience but require less than a watt of processing power and weigh under a pound"...
Well, Merom is a dual core mobile processor expected to ship H2 2006. That we knew.
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.
Well, Merom is a dual core mobile processor expected to ship H2 2006. That we knew.
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.
What I found astonishing is that they had working demonstrations of everything up to and including Woodcrest. I also found some timing issues, like maybe the shipping date keeps sliding closer, am I wrong here? I thought that some of these were for 2007, it appears that they are intended for 2006, or maybe I'm reading something wrong.
Well, Merom is a dual core mobile processor expected to ship H2 2006. That we knew...
Apparently, all 3 new CPUs will ship before the end of 2006.
I'm still pretty convinced that Apple will start with Yonahs as soon as possible (feb-march 2006 on iBooks and minis), PowerBooks (dualcore) at the 2006 WWDC, then if available, Conroe PowerMacs for the fall (Apple Expo Paris), iMacs before XMas, then Woodcrest XServes.
At the beginning of 2007 (MWSF): Merom PowerBooks, then Merom iBooks and minis before june.
At the 2007 WWDC, all Macs should be Intel-based and Leopard will be "ready".
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
The 5 Watts is likely "typical" or "normal" usage for Merom. TDP per the Inquirer is 35 Watts in the low 2 GHz range.
Quote:
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.[/url]
Again per the Inquirer, Conroe and Woodcrest at 65 Watts TDP and 80 Watts TDP respectively in the 2.5 to 3 GHz range.
From the little details we have, it looks like Merom will be 30% faster per clock than Dothan through a variety of means: better cache, wider issue, more execution units, more micro-ops fusion, and some macro-ops fusion. So 2 GHz Merom ~ 2.5 Yonah. With macro-ops fusion, it may be possible to have pipelined multiply-add functionality. Yes, it is a very nice platform for Apple.
Also note that Conroe is dual core but does not support dual processor. For multiple processors, you need the Woodcrest (Xeon-like) server processor, which would allow a dual-dual setup.
Because we're expecting future ProMacs to be Conroe-based, they will be dual-core but not dual-processor, and we shouldn't expect Apple to release a dual-dual G5; this because of their "pecking-order" thinking.
The 5 Watts is likely "typical" or "normal" usage for Merom. TDP per the Inquirer is 35 Watts in the low 2 GHz range.
Now that I think about it some more, I think this is incorrect. The 0.5 and 5 Watts figures are likely the TDP for the ultra-low voltage variant of the processors, and at lower frequencies as well.
Not too sure I would believe a 2+ GHz Merom running at 5 Watts TDP yet.
Also note that Conroe is dual core but does not support dual processor. For multiple processors, you need the Woodcrest (Xeon-like) server processor, which would allow a dual-dual setup.
Because we're expecting future ProMacs to be Conroe-based, they will be dual-core but not dual-processor, and we shouldn't expect Apple to release a dual-dual G5; this because of their "pecking-order" thinking.
Hmm... what we are expecting isn't necessarily correct. If Apple wants to have dual-dual setup, they simply would use Woodcrest:
bah. 8PM EST USA and still no keynotes. i'm here in a hotel in Malaysistan and its some good 512kbps up/down from here all the way to a US server. FAST !!
was hoping to catch the keynotes before i go in to this macromedia studio8 launch thingy
1 quick thought:
low-power for handheld thing: certainly something more promising to jobs than IBM trying to put a Power6-derivatve or a g5 in an iPod
1a.
goodbye, IBM, good riddance.
2. second quick thought:
80 watts average for conroe dual-core desktop is impressive. all amd athlon64 single-cores come in at 85W ++ IIRC (somebody feel free to spank me and correct this... better yet. can someone make a nice graph with the power consumptions "range bars"??? pretty please???)
3. and yes, OMFG they actually DEMONSTRATED REAL PRODUCTS NOT FUCKING VAPORWARE. good on Intel.
3a.
goodbye, IBM, good riddance.
4. OMFG they actually are hitting they're target dates and as someone pointed out, they might be ahead of schedule ???
Now that I think about it some more, I think this is incorrect. The 0.5 and 5 Watts figures are likely the TDP for the ultra-low voltage variant of the processors, and at lower frequencies as well.
Not too sure I would believe a 2+ GHz Merom running at 5 Watts TDP yet.
Comments
Originally posted by IQatEdo
I don't think Apple swapped to Intel to take on the X86 architecture. Time will tell.
Sure they did.
The developer docs are all about converting to 32bit X86, optimising for SSE2 or SSE3, endian issues. They aren't about some mythical new non-X86 architecture.
...and time will tell if that's always the case.
Originally posted by aegisdesign
Sure they did.
The developer docs are all about converting to 32bit X86, optimising for SSE2 or SSE3, endian issues. They aren't about some mythical new non-X86 architecture.
...and time will tell if that's always the case.
Thanks for that. Fair enough. By the way, is it 'mythical' or 'new'. 'New' I like!
Originally posted by THT
Jobs was exaggerating obviously.
You'd have thunk he's have learned by now. ;-)
I don't quite agree with you that a Dothan has 2x the performance/watt over a G5 but I'll agree with you that Intel are increasing the performance/watt ratio and quicker than IBM. It's near 1.5x now and Yonah would be about 3x which makes Merom at 6x provided the G5 stands still with no improvements.
How much quicker Intel gets there seems to be the whole debate though for Job's credibility. If Intel's Dev conference gives out that Merom/Conroe is just Yonah+ then I reckon the cred is shot already as I can't see IBM standing still and taking it. Their blade servers at least need low power 970MPs.
The 7448 is a good chip despite what the FSB speed weenies say - it makes up for it in other aspects, but yes, 9 months later than was needed for Apple to compete with Pentium M.
Originally posted by 1337_5L4Xx0R
Something many people overlook is that the performance per watt stats were for integer performance, which x86 beats macs handily at already.
The Pentium 4 beats the PPC970 as far as max scalar integer performance, but I'm not sure if it beats it in as far as performance per watt. That would be interesting to see.
Also, there are plenty of integer tests where the 970 will mop the floor with the P4 in max performance and performance per watt. . . They just need to be tests that do four integer operations at the same time. (And I do think it's fair to use Altivec in integer tests, since applications compiled with IBM's compiler will indeed make decent use of it.) You don't build a car with a turbo and then release performance numbers with the turbo off.
Anyway, where there's smoke there's fire, and to me it's obvious that Intel is up to something; afterall, it is Intel's job to stay ahead of the curve. The more relevant questions regard their plans for releasing it, and if it will debut on the Intel Powermac or whatnot.
Here's a direct link to the webcast page:
http://www.intel.com/idf/us/fall2005/webcast.htm
Ladies and gentlemen, fasten your seatbelts.
Originally posted by sunilraman
Intel Developer Forum: less than 24hours to go. should we start a new thread for goodies that come out? probably Kasper will...
Here's a direct link to the webcast page:
http://www.intel.com/idf/us/fall2005/webcast.htm
Ladies and gentlemen, fasten your seatbelts.
Thanks for the link. Anyone know time??
http://www.theregister.co.uk/2005/08...tel_long_slip/
So don't expect anything earth-shaking. More like apologies clothed as strategies.
Originally posted by cubist
Here is the Register's take on the Intel Developer Forum...
http://www.theregister.co.uk/2005/08...tel_long_slip/
So don't expect anything earth-shaking. More like apologies clothed as strategies.
No offence, but the Register has led me down the path too many times. I can never figure out if they know something or if they just think they know something that is shocking.
Otellini showed the first public demonstration of Merom, Conroe and Woodcrest - processors for notebook, desktop and server platforms designed on Intel's advanced 65-nanometer technology manufacturing process. He also said Intel has more than 10 processor projects that contain four (quad-core) or more processor cores per chip.
Otellini also announced that forthcoming lower-power products will lead to a new category of ultra energy-efficient "Handtop PC" devices that provide a converged communication and PC-like experience but require less than a watt of processing power and weigh under a pound.
This is from this morning Intel's press release. Nothing new really.
As far as we think Apple is concerned, Yonah's already out for iBooks, mac minis and possibly PowerBooks for spring 2006, and the 64-bit trio mentioned above will probably go into the next-gen xBooks/minis (Merom), PowerMacs and iMacs (Conroe), and XServes (Woodcrest).
Don't know about availability yet...
Curious about what Apple would do with an "ultra energy-efficient "Handtop PC" devices that provide a converged communication and PC-like experience but require less than a watt of processing power and weigh under a pound"...
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.
More at http://www.anandtech.com/tradeshows/showdoc.aspx?i=2503
Originally posted by aegisdesign
Well, Merom is a dual core mobile processor expected to ship H2 2006. That we knew.
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.
More at http://www.anandtech.com/tradeshows/showdoc.aspx?i=2503
What I found astonishing is that they had working demonstrations of everything up to and including Woodcrest. I also found some timing issues, like maybe the shipping date keeps sliding closer, am I wrong here? I thought that some of these were for 2007, it appears that they are intended for 2006, or maybe I'm reading something wrong.
Originally posted by aegisdesign
Well, Merom is a dual core mobile processor expected to ship H2 2006. That we knew...
Apparently, all 3 new CPUs will ship before the end of 2006.
I'm still pretty convinced that Apple will start with Yonahs as soon as possible (feb-march 2006 on iBooks and minis), PowerBooks (dualcore) at the 2006 WWDC, then if available, Conroe PowerMacs for the fall (Apple Expo Paris), iMacs before XMas, then Woodcrest XServes.
At the beginning of 2007 (MWSF): Merom PowerBooks, then Merom iBooks and minis before june.
At the 2007 WWDC, all Macs should be Intel-based and Leopard will be "ready".
Originally posted by aegisdesign
But not that they were targeting 5 Watts TDP. That is astonishing if they can do it. and 0.5W by 2010 apparently but 10 times more powerful. Perhaps Jobs wasn't exaggerating.
The 5 Watts is likely "typical" or "normal" usage for Merom. TDP per the Inquirer is 35 Watts in the low 2 GHz range.
Conroe is targeting 65W for the desktop. That's not so astonishing although if it's the same performance/watt as a Merom, that's one kick ass desktop performane wise.[/url]
Again per the Inquirer, Conroe and Woodcrest at 65 Watts TDP and 80 Watts TDP respectively in the 2.5 to 3 GHz range.
From the little details we have, it looks like Merom will be 30% faster per clock than Dothan through a variety of means: better cache, wider issue, more execution units, more micro-ops fusion, and some macro-ops fusion. So 2 GHz Merom ~ 2.5 Yonah. With macro-ops fusion, it may be possible to have pipelined multiply-add functionality. Yes, it is a very nice platform for Apple.
Because we're expecting future ProMacs to be Conroe-based, they will be dual-core but not dual-processor, and we shouldn't expect Apple to release a dual-dual G5; this because of their "pecking-order" thinking.
Originally posted by THT
The 5 Watts is likely "typical" or "normal" usage for Merom. TDP per the Inquirer is 35 Watts in the low 2 GHz range.
Now that I think about it some more, I think this is incorrect. The 0.5 and 5 Watts figures are likely the TDP for the ultra-low voltage variant of the processors, and at lower frequencies as well.
Not too sure I would believe a 2+ GHz Merom running at 5 Watts TDP yet.
Originally posted by cubist
Also note that Conroe is dual core but does not support dual processor. For multiple processors, you need the Woodcrest (Xeon-like) server processor, which would allow a dual-dual setup.
Because we're expecting future ProMacs to be Conroe-based, they will be dual-core but not dual-processor, and we shouldn't expect Apple to release a dual-dual G5; this because of their "pecking-order" thinking.
Hmm... what we are expecting isn't necessarily correct. If Apple wants to have dual-dual setup, they simply would use Woodcrest:
was hoping to catch the keynotes before i go in to this macromedia studio8 launch thingy
1 quick thought:
low-power for handheld thing: certainly something more promising to jobs than IBM trying to put a Power6-derivatve or a g5 in an iPod
1a.
goodbye, IBM, good riddance.
2. second quick thought:
80 watts average for conroe dual-core desktop is impressive. all amd athlon64 single-cores come in at 85W ++ IIRC (somebody feel free to spank me and correct this... better yet. can someone make a nice graph with the power consumptions "range bars"??? pretty please???)
3. and yes, OMFG they actually DEMONSTRATED REAL PRODUCTS NOT FUCKING VAPORWARE. good on Intel.
3a.
goodbye, IBM, good riddance.
4. OMFG they actually are hitting they're target dates and as someone pointed out, they might be ahead of schedule ???
4a.
goodbye, IBM, good riddance.
man, the darkside is strong in me this morning.
Originally posted by THT
Now that I think about it some more, I think this is incorrect. The 0.5 and 5 Watts figures are likely the TDP for the ultra-low voltage variant of the processors, and at lower frequencies as well.
Not too sure I would believe a 2+ GHz Merom running at 5 Watts TDP yet.
TDP == Total Dissipated Power??
Originally posted by THT
Hmm... what we are expecting isn't necessarily correct. If Apple wants to have dual-dual setup, they simply would use Woodcrest:
Well we will see where Apple positions the PowerMac, if they use woodcrest they are positioning it in the workstation market.