The Opteron A1100 will be the name born by AMD's first SoC, which we knew previously as Seattle and is the first chip which will contain ARM Cortex A57 architecture working in tandem with AMDs. It will be a full 64bit chip and will sport up to 4MB of shared L2 cache and 8MB of shared L3 cache and it will support of to four DIMMs of either DDR3 or DDR4 in dual channel with ECC. It will boot using UEFI into a Linux environment based on Fedora and will be optimized to handle web front ends and data centre tasks. As far as connectivity it will have 8 lanes of PCIe 3.0 and 8 SATA 3 ports. You can follow links from The Register to see the AMD Press Release.
"CHIP DESIGNER AMD is preparing to sample its 64-bit ARM based server processors codenamed Seattle, which will be the company's first stab at a system on chip (SoC) design for data centre products."
Here is some more Tech News from around the web:
- Heatsink Tester Shows Thermal Resistance Isn’t Futile @ Hack a Day
- Socle agrees to merge with Foxconn @ DigiTimes
- Teeny, tiny state machine could BREATHE NEW LIFE into Moore's Law @ The Register
- Toshiba twirls its HAMR, ponders whether to smash spinning rust @ The Register
- Good news! Today is Data Privacy Day … Stop sniggering at the back @ The Register
- Ubuntu Now Often Leads Windows 7 On Intel SNB Graphics Performance @ Phoronix
- Warning – Microsoft Opened A Security Hole In Internet Explorer 11 @ Tech ARP
amds first major soc is
amds first major soc is kabini/temas
This one is still based on
This one is still based on the ARM refrence design(A57) CPU, with some AMD IP integrated into to DIE, but it’s a start! AMD needs to get a custom ARM ISA based design out the door ASAP, beacuse Nvidia’s Denver is probably like Apple’s A7, a custom wide superscalar design with more execution pipelines, that is a more powerfull CPU than the ARM holdings refrence A57s. If AMD could get a wide superscalar design ARMv8 ISA based APU with all the HSA abilities of its x86 based Kaveri(GPU acceleration, unified memory), then they could be first to market for an ARM ISA powered CPU that has integrated GPU acceleration abilities for servers, Nvidia is not standing still with their Denver based designs.
See this AMD, better get to
See this AMD, better get to cloning some of your own custom ARMv8 ISA based kit, PDQ!
http://www.brightsideofnews.com/news/2012/9/20/nvidia-project-boulder-revealed-tegras-competitor-hides-in-gpu-group.aspx
These aren’t SoCs. These are
These aren’t SoCs. These are just normal processors like many others. Though, by moving SATA on chip, they’re moving in the SoC direction.
It’s not hard to integrate
It’s not hard to integrate many functional SOC blocks to any of ARM’s refrence designs like AMD is doing, and adding some of AMD’s functional SOC blocks onto the ARM refrence designs, along with other Third party SOC functional blocks, that most companies that manufacture both x86, and ARM do, even Intel has used third party IP(GPU IP) on some of its products! Nvidia has the same access to the same licensed IP from the providers that supply functional blocks of, for example, CPU on chip interconnect fabrics, and other specialized on chip SOC blocks, in fact Nvidia will probably be a supplier of GPU IP for others in the near future, as the licensed IP model, that ARM Holdings bulit its business around, takes hold of the server divices market, like it did with the Mobile devices market from day one, and the already existing internet of things markets is mostly built around that very same licensed IP model. ARM Holdings offers SOC blocks integration serverces to many midsize and smaller clients, who do not have the In House resources, of the Intels, and AMDs, and that includes working with other sellers of third party SOC IP, to assist Arm holding’s clients in getting their produvts to market.
There is a whole system integration services market that has over the years grown around offering system integration services to any company that decides to build devices around ARM Holdings, and others CPU licensed IP. These modular compoments will allow Nvidia, and others to get into the ARM based server market, much quicker than when the IP was not available in the past. For Nvidia, the hard part is already done, now that they have working Denver based CPUs.
Nice spam bot you have there.
Nice spam bot you have there. It almost sounds like a human.