07-03-2016 12:43
Con toda la rumorología que hay, esto es de cosecha propia. Viendo que Nintendo se está implicando muy mucho en la creación de Vulkan, me imagino que es posible que traten de usar hardware de móvil en la consola portátil (como hizo Sony con PSVita) y como AMD parece implicado en el desarrollo de la consola he visto que han presentado en Enero sus primeros procesadores con diseño ARM, lo que podría cuadrar con la "NX portátil":
![[Imagen: AMD-Opteron1-640x341.png]](http://www.extremetech.com/wp-content/uploads/2016/01/AMD-Opteron1-640x341.png)
AMDÔÇÖs first ARM-based processor, the Opteron A1100, is finally here
Today AMD is formally launching its first ARM processor core, the Opteron A1100. AMD first announced its plans to enter the ARM market in 2013, with the chip expected to ship by mid-2014. The company apparently began early sampling around that time frame, but is only now launching the processor.
The new Opteron A1100 is pretty much what AMD promised in its early previews of the device. It packs eight Cortex-A57 CPU cores, with each pair of cores sharing a 1MB L2 (512K effectively allocated to each chip). An 8MB L3 cache backs the entire CPU cluster, and the CPU supports both DDR3 and DDR4. ECC support is also provided.
The new A1100 supports up to 64GB per channel (128GB total) if registered DDR4 DIMMs are used, and 64GB of total memory when using standard DDR4. The board also includes dual 10GbE ports, 14 SATA3 controllers, and one PCIe 3.0 slot with x8 support.
The exact SKUs are shown below:
![[Imagen: Opteron-SKU-640x252.jpg]](http://www.extremetech.com/wp-content/uploads/2016/01/Opteron-SKU-640x252.jpg)
The quad-core A1120 is a 25W chip at 1.7GHz, the A1150 is a 32W core at 2GHz, and the A1170 is a 32W chip at 2GHz. The last two flavors are both eight-core processors, but all three SKUs share the same memory capabilities and L3 cache. If AMD follows its past history on evaluating TDP, the 25-32W figures will represent a worst-case scenario for the chip, rather than its performance in a general workload. AMD historically rates TDP in the first fashion, while Intel uses the second ÔÇö and thatÔÇÖs why you canÔÇÖt usually directly compare TDPs between the two companies.
Will it sell?
AMD has already lined up launch partners, with support from SoftIron ÔÇö though the banner on the companyÔÇÖs website states that systems are only available in ÔÇ£limited quantities.ÔÇØ
There are some major questions regarding AMDÔÇÖs decision to push into ARM, however. For one, the company told journalists that while it expected its new A1100 to compete well against IntelÔÇÖs Avoton line of Atom processors that first debuted in 2013. If AMD had kept its initial launch schedule and shipped the Cortex-A57 in 2014, it may well have had a leg up in the nascent dense server market.
Between the 2014 expected launch and the actual shipments today, Intel had time to bring another line of server products to market ÔÇö the 14nm ÔÇ£Xeon-D.ÔÇØ That chip has a very strong value argument ÔÇö Anandtech labeled it ÔÇ£the Xeon D is probably the most awesome product Intel has delivered in years.ÔÇØ ItÔÇÖs not particularly suitable for HPC or large-memory requirements, but itÔÇÖs great for everything else.
AMD has stated that it expects A1100 chips to start around $150, which would price them competitively against IntelÔÇÖs Atom-based hardware, but doesnÔÇÖt exactly offer the company a lot of profit margin.
AMDÔÇÖs delay may have limited the A1100ÔÇÖs sales market, but it may turn out to be a smart move in the long run. ItÔÇÖs hard to say that AMD is late when no other ARM server vendor has managed to field an Intel-compatible product. The ecosystem and support network that ARM servers need to be competitive with their x86 counterparts is still in its infancy, and it ultimately doesnÔÇÖt matter how good your hardware is if the software stack isnÔÇÖt there to support it. Rory Read, AMDÔÇÖs previous CEO, justified buying SeaMicro and investing in ARM as a long-term plan because he claimed the server market would be at least 15% ARM by 2018. This was always a dubious argument, and itÔÇÖs now clearly false ÔÇö ARM will not shoot from 0% to 15% of a market in less than 24 months.
From this viewpoint, the A1100 is a commercial proof-of-concept product that may not offer competitive performance across the board, but gives both AMD and its vendor partners an opportunity to earn a small ROI while ramping up second-generation designs more quickly. In the past, AMD claimed that the A1100 would be its first ARM product, with K12 following quickly behind. The company hasnÔÇÖt talked about K12 at all since Jim Keller left the company, but itÔÇÖs supposedly on track for a 2017 deployment.
I tend to think that AMD did the right thing by easing Seattle out the door rather than drop-kicking it in 2014. Server customers care just as much about stability and robust software solutions than they do about hardware performance. I wouldnÔÇÖt look to the A1100 to do much to reverse AMDÔÇÖs lost server market share or overall earnings, however ÔÇö the company has yet to announce any major purchasing wins, and it hasnÔÇÖt shown performance data.
At best, the Opteron A1100 will wedge a foot in the door of the ARM server market, with a few small wins and enough competitive strength to fuel customer interest in future markets. If AMD or Qualcomm want to bust this market open, theyÔÇÖll probably need custom designs and better hardware to do it.
Fuente
Vamos, pajas mentales mías...
AMDÔÇÖs first ARM-based processor, the Opteron A1100, is finally here
Today AMD is formally launching its first ARM processor core, the Opteron A1100. AMD first announced its plans to enter the ARM market in 2013, with the chip expected to ship by mid-2014. The company apparently began early sampling around that time frame, but is only now launching the processor.
The new Opteron A1100 is pretty much what AMD promised in its early previews of the device. It packs eight Cortex-A57 CPU cores, with each pair of cores sharing a 1MB L2 (512K effectively allocated to each chip). An 8MB L3 cache backs the entire CPU cluster, and the CPU supports both DDR3 and DDR4. ECC support is also provided.
The new A1100 supports up to 64GB per channel (128GB total) if registered DDR4 DIMMs are used, and 64GB of total memory when using standard DDR4. The board also includes dual 10GbE ports, 14 SATA3 controllers, and one PCIe 3.0 slot with x8 support.
The exact SKUs are shown below:
The quad-core A1120 is a 25W chip at 1.7GHz, the A1150 is a 32W core at 2GHz, and the A1170 is a 32W chip at 2GHz. The last two flavors are both eight-core processors, but all three SKUs share the same memory capabilities and L3 cache. If AMD follows its past history on evaluating TDP, the 25-32W figures will represent a worst-case scenario for the chip, rather than its performance in a general workload. AMD historically rates TDP in the first fashion, while Intel uses the second ÔÇö and thatÔÇÖs why you canÔÇÖt usually directly compare TDPs between the two companies.
Will it sell?
AMD has already lined up launch partners, with support from SoftIron ÔÇö though the banner on the companyÔÇÖs website states that systems are only available in ÔÇ£limited quantities.ÔÇØ
There are some major questions regarding AMDÔÇÖs decision to push into ARM, however. For one, the company told journalists that while it expected its new A1100 to compete well against IntelÔÇÖs Avoton line of Atom processors that first debuted in 2013. If AMD had kept its initial launch schedule and shipped the Cortex-A57 in 2014, it may well have had a leg up in the nascent dense server market.
Between the 2014 expected launch and the actual shipments today, Intel had time to bring another line of server products to market ÔÇö the 14nm ÔÇ£Xeon-D.ÔÇØ That chip has a very strong value argument ÔÇö Anandtech labeled it ÔÇ£the Xeon D is probably the most awesome product Intel has delivered in years.ÔÇØ ItÔÇÖs not particularly suitable for HPC or large-memory requirements, but itÔÇÖs great for everything else.
AMD has stated that it expects A1100 chips to start around $150, which would price them competitively against IntelÔÇÖs Atom-based hardware, but doesnÔÇÖt exactly offer the company a lot of profit margin.
AMDÔÇÖs delay may have limited the A1100ÔÇÖs sales market, but it may turn out to be a smart move in the long run. ItÔÇÖs hard to say that AMD is late when no other ARM server vendor has managed to field an Intel-compatible product. The ecosystem and support network that ARM servers need to be competitive with their x86 counterparts is still in its infancy, and it ultimately doesnÔÇÖt matter how good your hardware is if the software stack isnÔÇÖt there to support it. Rory Read, AMDÔÇÖs previous CEO, justified buying SeaMicro and investing in ARM as a long-term plan because he claimed the server market would be at least 15% ARM by 2018. This was always a dubious argument, and itÔÇÖs now clearly false ÔÇö ARM will not shoot from 0% to 15% of a market in less than 24 months.
From this viewpoint, the A1100 is a commercial proof-of-concept product that may not offer competitive performance across the board, but gives both AMD and its vendor partners an opportunity to earn a small ROI while ramping up second-generation designs more quickly. In the past, AMD claimed that the A1100 would be its first ARM product, with K12 following quickly behind. The company hasnÔÇÖt talked about K12 at all since Jim Keller left the company, but itÔÇÖs supposedly on track for a 2017 deployment.
I tend to think that AMD did the right thing by easing Seattle out the door rather than drop-kicking it in 2014. Server customers care just as much about stability and robust software solutions than they do about hardware performance. I wouldnÔÇÖt look to the A1100 to do much to reverse AMDÔÇÖs lost server market share or overall earnings, however ÔÇö the company has yet to announce any major purchasing wins, and it hasnÔÇÖt shown performance data.
At best, the Opteron A1100 will wedge a foot in the door of the ARM server market, with a few small wins and enough competitive strength to fuel customer interest in future markets. If AMD or Qualcomm want to bust this market open, theyÔÇÖll probably need custom designs and better hardware to do it.
Fuente
Vamos, pajas mentales mías...
