The factors:
a) AMD's much more precise knowledge of Fermi's competition at the time TSMC killed the 32nm process.
b) AMD's not talking about Fusion's gpu power management features.
c) AMD's CES statements regarding Fusion/AIB integration.
d) The recent rumors the 6000 series = N.I. = all new architecture.
e) Nvidia's silence on their next generation.
At the end of 2009, when TSMC killed the 32nm process and AMD was planning their next move with N.I. they had a much more accurate picture of how bad Nvidia had stumbled on Fermi and the competitive picture for the next 18 months. This made transitioning N.I. relatively intact to TSMC's 40nm process and increased die areas feasible.
AMD has talked up the intensive power management improvements in the CPU portion of it's Fusion chips, but has said nothing about the power management improvements to the GPU part of the chips, which logically should be commensurate as that would be vital in their Bobcat chips.
AMD has indicated it's fusion chips will seamlessly integrate with it's AIBs, allowing the AIB to stay on standby until the gpu load needs it. In a highend Llano chip, and probably starting with N.I., this can be expected to mean your gaming board can in standby drawing a <watt until called upon for actual gaming. Having high end gaming capability on tap while drawing <100 watts total while NOT gaming is a compelling feature.
Since N.I. was designed with Fusion in mind, one might expect some extensive power management features to be built into the 6000 series, which is probably why AMD has kept mum on their Fusion gpu power management implementations.
All of which leaves Nvidia in a very ugly competitive picture for the next 12 months. One would expect they would leaking info on their next architecture by now to stay in the picture, unless there is no chance of their next generation coming before late into 2012, perhaps, roughly co-incident with AMD releasing S.I. at 28nm.
By then Nvidia will be arterially bleeding out on all of their OEM x86 markets, which used to provide the bulk of their profits.
In all probability Nvidia is going to end up a much smaller company than it is now and far less able to directly compete with AMD in the consumer graphics markets.
Repost. Please use the search feature to see if there already is a thread on this subject.
http://forums.anandtech.com/showthread.php?t=2104030
esquared
Anandtech Administrator
a) AMD's much more precise knowledge of Fermi's competition at the time TSMC killed the 32nm process.
b) AMD's not talking about Fusion's gpu power management features.
c) AMD's CES statements regarding Fusion/AIB integration.
d) The recent rumors the 6000 series = N.I. = all new architecture.
e) Nvidia's silence on their next generation.
At the end of 2009, when TSMC killed the 32nm process and AMD was planning their next move with N.I. they had a much more accurate picture of how bad Nvidia had stumbled on Fermi and the competitive picture for the next 18 months. This made transitioning N.I. relatively intact to TSMC's 40nm process and increased die areas feasible.
AMD has talked up the intensive power management improvements in the CPU portion of it's Fusion chips, but has said nothing about the power management improvements to the GPU part of the chips, which logically should be commensurate as that would be vital in their Bobcat chips.
AMD has indicated it's fusion chips will seamlessly integrate with it's AIBs, allowing the AIB to stay on standby until the gpu load needs it. In a highend Llano chip, and probably starting with N.I., this can be expected to mean your gaming board can in standby drawing a <watt until called upon for actual gaming. Having high end gaming capability on tap while drawing <100 watts total while NOT gaming is a compelling feature.
Since N.I. was designed with Fusion in mind, one might expect some extensive power management features to be built into the 6000 series, which is probably why AMD has kept mum on their Fusion gpu power management implementations.
All of which leaves Nvidia in a very ugly competitive picture for the next 12 months. One would expect they would leaking info on their next architecture by now to stay in the picture, unless there is no chance of their next generation coming before late into 2012, perhaps, roughly co-incident with AMD releasing S.I. at 28nm.
By then Nvidia will be arterially bleeding out on all of their OEM x86 markets, which used to provide the bulk of their profits.
In all probability Nvidia is going to end up a much smaller company than it is now and far less able to directly compete with AMD in the consumer graphics markets.
Repost. Please use the search feature to see if there already is a thread on this subject.
http://forums.anandtech.com/showthread.php?t=2104030
esquared
Anandtech Administrator
Last edited by a moderator:
Facebook
Twitter