- Učlanjen(a)
- 12.07.2000
- Poruke
- 6,934
- Poena
- 800
...megahercnu trku!
najbolje se da videti iz sledeceg clanka (prevod sa Japanskog):
- Throwing Merom to the both of the desktop and Mobile
Intel throws the desktop CPU of new architecture and Mobile CPU in 2006. Mobile CPU "Merom (???)", and the desktop CPU becomes the same serial core as Merom. In other words, Mobile CPU enters to the desktop CPU area in 2006.
The desktop CPU changes dramatically with this. It rises, on the one hand the electric power consumption which is goes down previous life generation (Tejas/Cedarmill) from for the first time. And depending upon multiple core conversion, the performance improvement of CPU reaches the point where improvement of frequency compared to it owes to multithread operation efficiency improvement. Because of that, it does not keep changing you do not obtain either the technique of software development mid-term.
As for the Mobile CPU of Intel, present 0.13 µ m edition Pentium M (Banias: ????) after, 90nm edition Pentium M (Dothan: ???), it is presumed 65nm "Yonah (??)" to with it continues. Dothan this year 2nd quarter, Yonah 2005 latter half, and, as the succession of Yonah Merom appears in 2006 first half. The after of Yonah becomes the multiple core, Yonah loads cash of 2MB, Merom loads cash of 4MB.
"Napa (??)" it prepares Intel, as a platform for the Yonah - Merom generation. This, this fall appears "Sonoma (the Soviet Union NOMA)" the platform (the Alviso chip set base) with succession, PCI Express chip set "Calistoga of the 2nd generation (the chalice Tsuga)" makes the base. In the Napa generation, as for CPU TDP it is seen that 45W is planned. In other words, TDP of Merom means that 45W is supposed.
One after another the generation where Intel is planned desktop CPU becomes the derivative item of Merom. As for Mobile edition there is also a possibility the code name which differs having been attached, (reporting, Conroe is), does not do verification yet. As for largest difference of desktop edition and Mobile edition of Merom core TDP. Desktop edition is called approximately 90W vis-a-vis TDP of Mobile edition starting from 45W. Somehow, there is the difference of TDP of 2 times, it seems.
- The Merom architecture which is supposed
So, Merom it probably is CPU which has what kind of architecture. Several points are clear.
0 65nm process
0 multiple cores
0 Mobile 45W, TDP framework of desktop 90W
Bulk cash of 0 4MB or more
0 64bit Extension Technology (alias IA-32e)
0 LaGrande Technology
0 Vanderpool Technology
0 Hyper-Threading?
There are 0 plural configurations?
Bus architecture of 0 Pentium 4/M types?
First from the time, 2006 first half, as for Merom it is found that it is CPU of 65nm process. With Mobile CPU Yonah 65nm, in the desktop CPU the next generation CPU "Tejas (???)" "CederMill (?????)" of shrinking edition becomes empty 65nm. In other words, Merom with the desktop and, becomes the 2nd generation CPU of 65nm with Mobile.
From the fact that Yonah is the dual core, it is secure also for Merom to be multiple core constitution of dual or more. Intel September last year Intel Developer Forum (IDF) with, has made that dual core CPU is thrown to the desktop several years later clear. It is presumed that announcement of this time points to Merom.
The desktop Merom ?????? (the CPU core 4) can be also a case, constitution as a possibility. Because in other words, TDP of desktop edition has reached 2 times that of Mobile edition. The core increasing simply to 2 times, it can also think that TDP reached time. By the way, in the 65nm generation, as for the area of the Pentium M CPU core of Intel, inserting extensions in consideration, it is presumed that they are 40 square mm or less. Because of that, either ?????? die/di size (the area of the semiconductor itself) is not impossible from the surface.
Though, as for Intel, as for the desktop CPU because dual core it has explained, in the generation of Merom the possibility of being dual is high. When it becomes ??????, FSB (the front side bus) also expansion and the like probably will become necessary. But, if this pace in the near future it is thought (such as 45nm process) that it moves to ??????.
Presently, concerning Yonah operation of the CPU core of dual, at the time of AC and being changed at the time of the battery it is understood. In other words, at the time of AC the dual core, it becomes the single core at the time of battery drive, it seems. As for the following step of approach of this direction it probably is to adjust the number of cores which perhaps, operate dynamically according to CPU load. Actually, as for a certain information source, you say that in the Merom generation it has become theme to offer performance with on demand. Of course, in addition there there is a technical hurdle.
- Finally TDP of the desktop CPU goes down
The desktop Merom is said that it becomes TDP approximately of 90W. TDP of the desktop CPU goes down, it is good news. Because as for that, as for present CPU, process technology refining, electric power consumption /TDP does not go down. Because of that, 90nm edition Pentium 4 (Prescott: The press cot) later TDP rises, it has become on the one hand.
For example, as for the TDP framework of Tejas of desktop CPU 2005 125W. This guideline "Performance of the motherboard design for Tejas FMB (Flexible Motherboard) is clear from the fact that TDP of 2005" is 125W.
Intel with Microprocessor Forum of the time before, 140W has explained the upper limit of CPU TDP which can be loaded onto the desktop PC. In other words, in the Tejas generation as for TDP it reaches the level which is approached to the upper limit rather. As for the thermal design you suffer hardship rather in even the µ PGA478 edition Prescott of 103W. When that becomes 125W, it is close to the limit which can be designed at proper cost. Because of that, Intel has become unavoidable to propose the form factor standard like BTX which did thermal design simulation accurately.
But when it becomes Merom, TDP goes down to Northwood level again. Because of that, the design of the desktop PC is presumed that considerably it becomes easy.
As for doubt, as for the difference of TDP of 2 times that both Merom of the desktop and Mobile it has occurred from somewhere, is. When we assume that Mobile Merom the dual core, the desktop Merom is ??????, as for story being simple, the difference of TDP means difference of the number of cores. But, if you inquire about the explanation of present Intel, the possibility is thin so.
When it does, the difference of the TDP framework of the desktop and Mobile, is supposed that it has come from the difference of frequency & drive voltage. When we assume, if so is, as for Merom as for present Pentium M compared to, it is presumed that it has the scalability whose high frequency conversion is possible. About NetBurst is not deep, but it probably is to become the pipeline whose is deeper than Banias/Dothan.
- Merom which loads cash of 4MB
Intel CPU has increased rapidly the on-board cash quantity. It is said that 1MB of Prescott/Banias, with Tejas/Dothan becomes 2MB, with Merom becomes 4MB. There are two meanings in this. Size of the SRAM cell having reduced (1) with refining the process, (2) the die/di (the semiconductor itself) heat above spread and electrical density are lowered by the fact that one it measures loads SRAM whose electric power consumption is low.
As for the CPU core of Banias of 0.13 µ m process when with presumption from the die/di photograph approximately 50 squares mm, it becomes Dothan of 90nm, approximately 37 squares mm. When process refines 1 generation, in principle as for the area of the same functional block it is the expectation which becomes half, but with Banias and Dothan the quantities of 70 % to only the die/di has not reduced. This, is seen that there is various reasons such as extensions and optimization to process.
When you suppose, that from Dothan with the movement to Yonah, the present die/di reduces in 60 - 70%, the CPU core with 65nm in regard to calculation becomes approximately 22 - 27 squares mm. Became 50% increase assuming that it compared and that expansion such as 64bit conversion joins there, as for core size 33 - 40 squares mm. If this, loading L2 cash of 4MB to a dual core, die/di size becomes the calculation which is settled 150 square mm or less. This is general size as a die/di of the main stream CPU of Intel. When it is ??????, 200 squares mm super it becomes, but even then it is Willamette class.
The bus architecture of Merom is not found yet. But, from the fact that it is seen that Yonah has adopted the same bus architecture as Dothan, also Merom at least as for the first generation as for the possibility of being the Pentium 4/M bus is high. Because in other words, as for Yonah and Merom it corresponds with the same chip set/the platform. Either a certain industry authorized personnel, "of the for the present has not heard the story new bus with Mobile CPU", that you say.
However, when of expansion of the core is thought, it is seen that there is a direction which in the future keeps expanding the bus in the long term. Following to the expansion of the core, because also necessary peak bus zone keeps becoming high.
2006 AMD ce imati K9 koji ce se proizvodtiti u FAb36 i bice izradjivan u 65nm na 300 mm podlogama.
Akos e u obzir uzmu sadasnji trendovi i predvidjanja i Merom i K9 ce raditi na otprilike istim frekvencijama, a mozda ce K9 biti i nesto brzi u cistim megahercima (90nm K9 bi trebalo da ide vema blizu 4 GHz) Merom ce sa svojih 4MB cachea imati duplo veci sekundarni kes od K9, a oba procesora ce omoguciti multycore konfiguracije...
Te 2006 ce biti relevantni jedino 64bitni procesori i 64bitini OS (Longorn)
najbolje se da videti iz sledeceg clanka (prevod sa Japanskog):
- Throwing Merom to the both of the desktop and Mobile
Intel throws the desktop CPU of new architecture and Mobile CPU in 2006. Mobile CPU "Merom (???)", and the desktop CPU becomes the same serial core as Merom. In other words, Mobile CPU enters to the desktop CPU area in 2006.
The desktop CPU changes dramatically with this. It rises, on the one hand the electric power consumption which is goes down previous life generation (Tejas/Cedarmill) from for the first time. And depending upon multiple core conversion, the performance improvement of CPU reaches the point where improvement of frequency compared to it owes to multithread operation efficiency improvement. Because of that, it does not keep changing you do not obtain either the technique of software development mid-term.
As for the Mobile CPU of Intel, present 0.13 µ m edition Pentium M (Banias: ????) after, 90nm edition Pentium M (Dothan: ???), it is presumed 65nm "Yonah (??)" to with it continues. Dothan this year 2nd quarter, Yonah 2005 latter half, and, as the succession of Yonah Merom appears in 2006 first half. The after of Yonah becomes the multiple core, Yonah loads cash of 2MB, Merom loads cash of 4MB.
"Napa (??)" it prepares Intel, as a platform for the Yonah - Merom generation. This, this fall appears "Sonoma (the Soviet Union NOMA)" the platform (the Alviso chip set base) with succession, PCI Express chip set "Calistoga of the 2nd generation (the chalice Tsuga)" makes the base. In the Napa generation, as for CPU TDP it is seen that 45W is planned. In other words, TDP of Merom means that 45W is supposed.
One after another the generation where Intel is planned desktop CPU becomes the derivative item of Merom. As for Mobile edition there is also a possibility the code name which differs having been attached, (reporting, Conroe is), does not do verification yet. As for largest difference of desktop edition and Mobile edition of Merom core TDP. Desktop edition is called approximately 90W vis-a-vis TDP of Mobile edition starting from 45W. Somehow, there is the difference of TDP of 2 times, it seems.
- The Merom architecture which is supposed
So, Merom it probably is CPU which has what kind of architecture. Several points are clear.
0 65nm process
0 multiple cores
0 Mobile 45W, TDP framework of desktop 90W
Bulk cash of 0 4MB or more
0 64bit Extension Technology (alias IA-32e)
0 LaGrande Technology
0 Vanderpool Technology
0 Hyper-Threading?
There are 0 plural configurations?
Bus architecture of 0 Pentium 4/M types?
First from the time, 2006 first half, as for Merom it is found that it is CPU of 65nm process. With Mobile CPU Yonah 65nm, in the desktop CPU the next generation CPU "Tejas (???)" "CederMill (?????)" of shrinking edition becomes empty 65nm. In other words, Merom with the desktop and, becomes the 2nd generation CPU of 65nm with Mobile.
From the fact that Yonah is the dual core, it is secure also for Merom to be multiple core constitution of dual or more. Intel September last year Intel Developer Forum (IDF) with, has made that dual core CPU is thrown to the desktop several years later clear. It is presumed that announcement of this time points to Merom.
The desktop Merom ?????? (the CPU core 4) can be also a case, constitution as a possibility. Because in other words, TDP of desktop edition has reached 2 times that of Mobile edition. The core increasing simply to 2 times, it can also think that TDP reached time. By the way, in the 65nm generation, as for the area of the Pentium M CPU core of Intel, inserting extensions in consideration, it is presumed that they are 40 square mm or less. Because of that, either ?????? die/di size (the area of the semiconductor itself) is not impossible from the surface.
Though, as for Intel, as for the desktop CPU because dual core it has explained, in the generation of Merom the possibility of being dual is high. When it becomes ??????, FSB (the front side bus) also expansion and the like probably will become necessary. But, if this pace in the near future it is thought (such as 45nm process) that it moves to ??????.
Presently, concerning Yonah operation of the CPU core of dual, at the time of AC and being changed at the time of the battery it is understood. In other words, at the time of AC the dual core, it becomes the single core at the time of battery drive, it seems. As for the following step of approach of this direction it probably is to adjust the number of cores which perhaps, operate dynamically according to CPU load. Actually, as for a certain information source, you say that in the Merom generation it has become theme to offer performance with on demand. Of course, in addition there there is a technical hurdle.
- Finally TDP of the desktop CPU goes down
The desktop Merom is said that it becomes TDP approximately of 90W. TDP of the desktop CPU goes down, it is good news. Because as for that, as for present CPU, process technology refining, electric power consumption /TDP does not go down. Because of that, 90nm edition Pentium 4 (Prescott: The press cot) later TDP rises, it has become on the one hand.
For example, as for the TDP framework of Tejas of desktop CPU 2005 125W. This guideline "Performance of the motherboard design for Tejas FMB (Flexible Motherboard) is clear from the fact that TDP of 2005" is 125W.
Intel with Microprocessor Forum of the time before, 140W has explained the upper limit of CPU TDP which can be loaded onto the desktop PC. In other words, in the Tejas generation as for TDP it reaches the level which is approached to the upper limit rather. As for the thermal design you suffer hardship rather in even the µ PGA478 edition Prescott of 103W. When that becomes 125W, it is close to the limit which can be designed at proper cost. Because of that, Intel has become unavoidable to propose the form factor standard like BTX which did thermal design simulation accurately.
But when it becomes Merom, TDP goes down to Northwood level again. Because of that, the design of the desktop PC is presumed that considerably it becomes easy.
As for doubt, as for the difference of TDP of 2 times that both Merom of the desktop and Mobile it has occurred from somewhere, is. When we assume that Mobile Merom the dual core, the desktop Merom is ??????, as for story being simple, the difference of TDP means difference of the number of cores. But, if you inquire about the explanation of present Intel, the possibility is thin so.
When it does, the difference of the TDP framework of the desktop and Mobile, is supposed that it has come from the difference of frequency & drive voltage. When we assume, if so is, as for Merom as for present Pentium M compared to, it is presumed that it has the scalability whose high frequency conversion is possible. About NetBurst is not deep, but it probably is to become the pipeline whose is deeper than Banias/Dothan.
- Merom which loads cash of 4MB
Intel CPU has increased rapidly the on-board cash quantity. It is said that 1MB of Prescott/Banias, with Tejas/Dothan becomes 2MB, with Merom becomes 4MB. There are two meanings in this. Size of the SRAM cell having reduced (1) with refining the process, (2) the die/di (the semiconductor itself) heat above spread and electrical density are lowered by the fact that one it measures loads SRAM whose electric power consumption is low.
As for the CPU core of Banias of 0.13 µ m process when with presumption from the die/di photograph approximately 50 squares mm, it becomes Dothan of 90nm, approximately 37 squares mm. When process refines 1 generation, in principle as for the area of the same functional block it is the expectation which becomes half, but with Banias and Dothan the quantities of 70 % to only the die/di has not reduced. This, is seen that there is various reasons such as extensions and optimization to process.
When you suppose, that from Dothan with the movement to Yonah, the present die/di reduces in 60 - 70%, the CPU core with 65nm in regard to calculation becomes approximately 22 - 27 squares mm. Became 50% increase assuming that it compared and that expansion such as 64bit conversion joins there, as for core size 33 - 40 squares mm. If this, loading L2 cash of 4MB to a dual core, die/di size becomes the calculation which is settled 150 square mm or less. This is general size as a die/di of the main stream CPU of Intel. When it is ??????, 200 squares mm super it becomes, but even then it is Willamette class.
The bus architecture of Merom is not found yet. But, from the fact that it is seen that Yonah has adopted the same bus architecture as Dothan, also Merom at least as for the first generation as for the possibility of being the Pentium 4/M bus is high. Because in other words, as for Yonah and Merom it corresponds with the same chip set/the platform. Either a certain industry authorized personnel, "of the for the present has not heard the story new bus with Mobile CPU", that you say.
However, when of expansion of the core is thought, it is seen that there is a direction which in the future keeps expanding the bus in the long term. Following to the expansion of the core, because also necessary peak bus zone keeps becoming high.
2006 AMD ce imati K9 koji ce se proizvodtiti u FAb36 i bice izradjivan u 65nm na 300 mm podlogama.
Akos e u obzir uzmu sadasnji trendovi i predvidjanja i Merom i K9 ce raditi na otprilike istim frekvencijama, a mozda ce K9 biti i nesto brzi u cistim megahercima (90nm K9 bi trebalo da ide vema blizu 4 GHz) Merom ce sa svojih 4MB cachea imati duplo veci sekundarni kes od K9, a oba procesora ce omoguciti multycore konfiguracije...
Te 2006 ce biti relevantni jedino 64bitni procesori i 64bitini OS (Longorn)