By: ? (0xe2.0x9a.0x9b.delete@this.gmail.com), January 17, 2011 2:50 am
Room: Moderated Discussions
anon (anon@anon.com) on 1/16/11 wrote:
---------------------------
>? (0xe2.0x9a.0x9b@gmail.com) on 1/16/11 wrote:
>---------------------------
>>Brett (ggtgp@yahoo.com) on 1/15/11 wrote:
>>---------------------------
>>>rwessel (robertwessel@yahoo.com) on 1/15/11 wrote:
>>>---------------------------
>>>>Brett (ggtgp@yahoo.com) on 1/14/11 wrote:
>>>>---------------------------
>>>>>AMD's Bulldozer Microprocessors Expected to Offer 50% Higher Performance than Core i7, Phenom II Chips.
>>>>>http://www.xbitlabs.com/news/cpu/display/20110114134306_AMD_s_Bulldozer_Microprocessors_Expected_to_Offer_50_Higher_Performance_than_Core_i7_Phenom_II_Chips.html
>>>>
>>>>
>>>>An eight core Bulldozer is expected to offer 50% higher performance than a quad-core
>>>>i7? Which side are you arguing here?
>>>
>>>A eight core Bulldozer is the same die size as a quad-core i7.
>>>
>>>Adding Hypethreading to i7 add ~5% to die size.
>>>Adding a second integer unit to each Bulldozer compute engine adds ~17%.
>>>
>>>A eight core Bulldozer can be looked at as 4 cores with 8 threads.
>>>
>>>Intel will try and use a die shrink 8 core to compare to AMD, but the thermal limits
>>>may make it no faster than Bulldozer.
>>>
>>>AMD will dominate compute per die area, compute per thermals, compute per cost.
>>>
>>>AMD has an overwhelming advantage for the first time, AMD needs to grab market
>>>share while they can. In two to four years Intel will copy this and be back to
>>>having a volume advantage, and be able to squeeze AMD again, unless AMD has grabbed
>>>enough more market share to negate that advantage.
>>>
>>>The next two years decides AMDs survival.
>>>And whether Intel is the next International Harvester.
>>>(A dominate monopoly replaced by John Deere)
>>>http://en.wikipedia.org/wiki/File:JDTractorDUBIA7800.jpg
>>
>>>
>>
>>I think your predictions are mostly wrong here. However, I agree it is probable
>>that Intel will copy the AMD's 2-integer-cores-per-1-FPU-core idea and similar ideas
>>that result in higher performance per watt.
>
>Err, the one that AMD copied from Niagara? (and I'm sure it wasn't new then).
>
>Look, armchair architects making disparaging comments about one company "copying"
>another just look like fools and have no idea about the history of design and expertise
>of the engineers and scientists involved.
What is your point? By "copying the AMD's idea" I meant "copying the AMD's idea to *use* such a feature in their x86 CPU that will be competing with an Intel's x86 CPU not having such a feature". That does not mean AMD invented it.
>All the idiots ranting about Intel "copying"
>AMD's ODMC obviously didn't know that ODMCs were known and used for a long long
>time, and even implemented at Intel long before Opteron.
You cannot deny that Intel copied some of AMD's *decisions* to use certain technologies in mainstream x86 CPUs (e.g: integrated memory controller). That does not mean AMD was 1st in the world to invent the idea - they were the 1st to implement it in mainstream x86 CPUs.
That is what I am talking about. You have a problem with that?
>Rant out of the way, I really don't know if Intel will go down that road. Both
>Intel and IBM have done SMT, which is actaully a huge advantage in one place --
>it allows single threaded execution resources to be boosted way past the knee of
>the perf/watt curve, which gives a more versatile performing chip.
>
>My prediction is that all the predictions about Intel being at death's door at the hands of BD are a little premature.
Who wrote such a thing and where?
---------------------------
>? (0xe2.0x9a.0x9b@gmail.com) on 1/16/11 wrote:
>---------------------------
>>Brett (ggtgp@yahoo.com) on 1/15/11 wrote:
>>---------------------------
>>>rwessel (robertwessel@yahoo.com) on 1/15/11 wrote:
>>>---------------------------
>>>>Brett (ggtgp@yahoo.com) on 1/14/11 wrote:
>>>>---------------------------
>>>>>AMD's Bulldozer Microprocessors Expected to Offer 50% Higher Performance than Core i7, Phenom II Chips.
>>>>>http://www.xbitlabs.com/news/cpu/display/20110114134306_AMD_s_Bulldozer_Microprocessors_Expected_to_Offer_50_Higher_Performance_than_Core_i7_Phenom_II_Chips.html
>>>>
>>>>
>>>>An eight core Bulldozer is expected to offer 50% higher performance than a quad-core
>>>>i7? Which side are you arguing here?
>>>
>>>A eight core Bulldozer is the same die size as a quad-core i7.
>>>
>>>Adding Hypethreading to i7 add ~5% to die size.
>>>Adding a second integer unit to each Bulldozer compute engine adds ~17%.
>>>
>>>A eight core Bulldozer can be looked at as 4 cores with 8 threads.
>>>
>>>Intel will try and use a die shrink 8 core to compare to AMD, but the thermal limits
>>>may make it no faster than Bulldozer.
>>>
>>>AMD will dominate compute per die area, compute per thermals, compute per cost.
>>>
>>>AMD has an overwhelming advantage for the first time, AMD needs to grab market
>>>share while they can. In two to four years Intel will copy this and be back to
>>>having a volume advantage, and be able to squeeze AMD again, unless AMD has grabbed
>>>enough more market share to negate that advantage.
>>>
>>>The next two years decides AMDs survival.
>>>And whether Intel is the next International Harvester.
>>>(A dominate monopoly replaced by John Deere)
>>>http://en.wikipedia.org/wiki/File:JDTractorDUBIA7800.jpg
>>
>>>
>>
>>I think your predictions are mostly wrong here. However, I agree it is probable
>>that Intel will copy the AMD's 2-integer-cores-per-1-FPU-core idea and similar ideas
>>that result in higher performance per watt.
>
>Err, the one that AMD copied from Niagara? (and I'm sure it wasn't new then).
>
>Look, armchair architects making disparaging comments about one company "copying"
>another just look like fools and have no idea about the history of design and expertise
>of the engineers and scientists involved.
What is your point? By "copying the AMD's idea" I meant "copying the AMD's idea to *use* such a feature in their x86 CPU that will be competing with an Intel's x86 CPU not having such a feature". That does not mean AMD invented it.
>All the idiots ranting about Intel "copying"
>AMD's ODMC obviously didn't know that ODMCs were known and used for a long long
>time, and even implemented at Intel long before Opteron.
You cannot deny that Intel copied some of AMD's *decisions* to use certain technologies in mainstream x86 CPUs (e.g: integrated memory controller). That does not mean AMD was 1st in the world to invent the idea - they were the 1st to implement it in mainstream x86 CPUs.
That is what I am talking about. You have a problem with that?
>Rant out of the way, I really don't know if Intel will go down that road. Both
>Intel and IBM have done SMT, which is actaully a huge advantage in one place --
>it allows single threaded execution resources to be boosted way past the knee of
>the perf/watt curve, which gives a more versatile performing chip.
>
>My prediction is that all the predictions about Intel being at death's door at the hands of BD are a little premature.
Who wrote such a thing and where?
