Tweet
Generally, AMD has not been a company that jumps onto every technology bandwagon that passes by. While Intel was promoting RDRAM memory as the wave of the future, AMD was sticking with SDRAM and looking to move to DDR. The same thing happened in the graphics arena, as Intel boarded the PCI Express train a bit early, while AMD waited to support the new bus until actual graphics cards were available to mainstream consumers.
Now the chipmaker looks to continue this trend with its migration to DDR-2, which is slated to occur in 2006 -- almost two years after Intel got the ball running with its LGA775-socket Pentium 4s and 915/925 chipsets. The transition will be an across-the-board move for AMD, involving all desktop, server, and mobile processors -- yes, even the Sempron will get a DDR-2 makeover, as will the Opteron, Athlon 64, and Turion 64 lines.
This is a big move for AMD, as the change from an integrated DDR to ditto DDR-2 memory controller will require a redesign of the processor core. But DDR-2 offers increased scalability for future AMD CPUs, and the arrival of DDR-2/1000 modules has effectively doubled the available memory bandwidth of even PC4200 DDR-1. There is still the question of higher latencies, but the ability to scale to 16GB/sec for dual-channel DDR-2/1000 will certainly give AMD a lot of headroom and may give Intel some headaches.
We've also been hearing rumblings that we've seen the last AMD (or Intel) high-end, single-core CPU, and the new AMD roadmap confirms this. The Athlon 64 FX-57 looks to be the last single-core speedster from AMD, with the Athlon 64 FX line adding a second core in 2006. Exactly how this will be done is still an open question, but the most logical path would be to split the current Athlon 64 X2 Toledo (2x1MB of Level 2 cache) and Manchester (2x512K) core revisions into Athlon 64 X2 and Athlon 64 FX lines. Meanwhile, AMD will likely leave the single-core door open just a crack for the entry-level Sempron to squeeze through.
Mobile Processors and Virtual Processors
The dual-core revolution will also extend to AMD's mobile processors, a move we've been anticipating since Intel first announced its "Yonah" dual-core CPU, scheduled to ship in early 2006. AMD's move will almost certainly come later, but while Yonah is an enhanced dual-core version of Intel's existing Pentium M technology, AMD will be moving its Turion 64 chips to dual-core, arguably creating a more current line of mobile processors. The 90-nanometer-process Turion 64 is a very impressive product, and by adding dual-core to the mix, AMD could seriously shake up the performance notebook market.
The move to dual-core technology is only part of AMD's mobile plan for 2006, with new memory formats and architectures debuting as well. Not only does the addition of DDR-2 support boost the mobile line's performance, but AMD also plans to add a dual-channel memory controller to all of its portable platforms. Currently, AMD mobile processors mimic their desktop counterparts in many ways, but even the 939-pin chips utilize only a single-channel memory interface. Combining dual-channel DDR-2 and a dual-core Turion 64 should have a profound impact on overall performance, and just might help AMD carve out a larger market share.
Virtualization is a buzzword we've been hearing for a long time, and the conversation has really heated up now that Intel's revealed that its latest core revision supports the technology. Virtualization basically spoofs the system into thinking there are multiple computers or virtual machines at work, thereby allowing multiple operating systems to run simultaneously. This involves both hardware and software partitioning into multiple environments, hence a high-end dual-core system can help make virtualization a reality without a negative performance impact.
Where this technology really comes in handy is in running multiple instances of the same OS, allowing a different usage pattern or security protocol, as well as ensuring that any unused portion of system resources is allocated and utilized. This can allow legacy applications to run on an older OS, provide seamless software or hardware upgrades, and help emulate an older, more compliant hardware configuration. There is also the obvious benefit of loading a different operating system alongside your standard desktop, allowing users an unprecedented selection of cutting-edge software and OS-specific applications.
AMD will be looking to add virtualization to its high-end and mainstream desktop, server, and mobile processor lines in 2006, which essentially translates into everything but the entry-level Sempron models receiving a virtualization upgrade.
Just in the Nick of Time
AMD's longtime strategy of adopting new technologies only when absolutely required seems to have paid significant dividends in 2005 and will likely continue into 2006. The desktop Athlon 64 processors have no problem outpacing their DDR-2-powered Intel counterparts, just as the Turion 64 remains competitive with the Pentium M, and the addition of dual-core, DDR-2, and dual-channel architectures and system virtualization will only make the AMD line even stronger. To be sure, Intel has a few tricks up its sleeve for 2006, but so far AMD's been playing its follow-the-leader strategy so smartly it may wind up being the leader. Credit: Hardware Central
Now the chipmaker looks to continue this trend with its migration to DDR-2, which is slated to occur in 2006 -- almost two years after Intel got the ball running with its LGA775-socket Pentium 4s and 915/925 chipsets. The transition will be an across-the-board move for AMD, involving all desktop, server, and mobile processors -- yes, even the Sempron will get a DDR-2 makeover, as will the Opteron, Athlon 64, and Turion 64 lines.
This is a big move for AMD, as the change from an integrated DDR to ditto DDR-2 memory controller will require a redesign of the processor core. But DDR-2 offers increased scalability for future AMD CPUs, and the arrival of DDR-2/1000 modules has effectively doubled the available memory bandwidth of even PC4200 DDR-1. There is still the question of higher latencies, but the ability to scale to 16GB/sec for dual-channel DDR-2/1000 will certainly give AMD a lot of headroom and may give Intel some headaches.
We've also been hearing rumblings that we've seen the last AMD (or Intel) high-end, single-core CPU, and the new AMD roadmap confirms this. The Athlon 64 FX-57 looks to be the last single-core speedster from AMD, with the Athlon 64 FX line adding a second core in 2006. Exactly how this will be done is still an open question, but the most logical path would be to split the current Athlon 64 X2 Toledo (2x1MB of Level 2 cache) and Manchester (2x512K) core revisions into Athlon 64 X2 and Athlon 64 FX lines. Meanwhile, AMD will likely leave the single-core door open just a crack for the entry-level Sempron to squeeze through.
Mobile Processors and Virtual Processors
The dual-core revolution will also extend to AMD's mobile processors, a move we've been anticipating since Intel first announced its "Yonah" dual-core CPU, scheduled to ship in early 2006. AMD's move will almost certainly come later, but while Yonah is an enhanced dual-core version of Intel's existing Pentium M technology, AMD will be moving its Turion 64 chips to dual-core, arguably creating a more current line of mobile processors. The 90-nanometer-process Turion 64 is a very impressive product, and by adding dual-core to the mix, AMD could seriously shake up the performance notebook market.
The move to dual-core technology is only part of AMD's mobile plan for 2006, with new memory formats and architectures debuting as well. Not only does the addition of DDR-2 support boost the mobile line's performance, but AMD also plans to add a dual-channel memory controller to all of its portable platforms. Currently, AMD mobile processors mimic their desktop counterparts in many ways, but even the 939-pin chips utilize only a single-channel memory interface. Combining dual-channel DDR-2 and a dual-core Turion 64 should have a profound impact on overall performance, and just might help AMD carve out a larger market share.
Virtualization is a buzzword we've been hearing for a long time, and the conversation has really heated up now that Intel's revealed that its latest core revision supports the technology. Virtualization basically spoofs the system into thinking there are multiple computers or virtual machines at work, thereby allowing multiple operating systems to run simultaneously. This involves both hardware and software partitioning into multiple environments, hence a high-end dual-core system can help make virtualization a reality without a negative performance impact.
Where this technology really comes in handy is in running multiple instances of the same OS, allowing a different usage pattern or security protocol, as well as ensuring that any unused portion of system resources is allocated and utilized. This can allow legacy applications to run on an older OS, provide seamless software or hardware upgrades, and help emulate an older, more compliant hardware configuration. There is also the obvious benefit of loading a different operating system alongside your standard desktop, allowing users an unprecedented selection of cutting-edge software and OS-specific applications.
AMD will be looking to add virtualization to its high-end and mainstream desktop, server, and mobile processor lines in 2006, which essentially translates into everything but the entry-level Sempron models receiving a virtualization upgrade.
Just in the Nick of Time
AMD's longtime strategy of adopting new technologies only when absolutely required seems to have paid significant dividends in 2005 and will likely continue into 2006. The desktop Athlon 64 processors have no problem outpacing their DDR-2-powered Intel counterparts, just as the Turion 64 remains competitive with the Pentium M, and the addition of dual-core, DDR-2, and dual-channel architectures and system virtualization will only make the AMD line even stronger. To be sure, Intel has a few tricks up its sleeve for 2006, but so far AMD's been playing its follow-the-leader strategy so smartly it may wind up being the leader. Credit: Hardware Central
Comment