- Oct 14, 2003
- 8,686
- 3,787
- 136
*Warning: Reading this piece MAY result in mild depression and sadness* 
Anyone feel that every "miracle" tech has yet to have proven themselves?
-Sun's Rock with runahead threads
-AMD's original K8: http://chip-architect.com/news/2001_10_02_Hammer_microarchitecture.html
-Intel Larrabee
-IA64
Original K8: If it ever released in the original timeframe, against the K7's and the Prescott's it might have been beyond impressive. For whatever reasons though, it never materialized. Bulldozer, which some people say there are similarities to the original K8 design would be facing much more potent competitors from Intel.
Larrabee: http://www.semiaccurate.com/2010/05/17/larrabee-3-now-larrabee-2/
Let's assume Charlie's infos are true. If it released in early 2009 as per the original roadmap again it would have been way beyond a "game changer". But the reality is that by the time whatever iteration is released(Larrabee 3?), the competitors in graphics would have closed lot of the GPGPU related gap. Even if Intel executes brilliantly, it would not be impressive as their original promises.
IA64: Shown by Intel as end-all be-all architecture that would eventually replace x86, almost 10 years later its laughingstock of the industry and even the most faithful are losing hopes of it ever succeeding. The premise was that thanks to Intel's manufacturing prowess and their financial strength it would be able to overcome previous attempts to overthrow x86. Again, if Poulson(the next Itanium) executes well, it would have lost much of the hype and original steam.
Lastly, Sun's Rock with runahead threads and out of order retirement. Who knows if this will ever surface again?
I think the lesson here is that "one ring rules them all" type of approach is no longer feasible. As the devices and technologies become more complex, "innovation" is merely a tool to keep everything from stagnating.
Anyone feel that every "miracle" tech has yet to have proven themselves?
-Sun's Rock with runahead threads
-AMD's original K8: http://chip-architect.com/news/2001_10_02_Hammer_microarchitecture.html
-Intel Larrabee
-IA64
Original K8: If it ever released in the original timeframe, against the K7's and the Prescott's it might have been beyond impressive. For whatever reasons though, it never materialized. Bulldozer, which some people say there are similarities to the original K8 design would be facing much more potent competitors from Intel.
Larrabee: http://www.semiaccurate.com/2010/05/17/larrabee-3-now-larrabee-2/
Let's assume Charlie's infos are true. If it released in early 2009 as per the original roadmap again it would have been way beyond a "game changer". But the reality is that by the time whatever iteration is released(Larrabee 3?), the competitors in graphics would have closed lot of the GPGPU related gap. Even if Intel executes brilliantly, it would not be impressive as their original promises.
IA64: Shown by Intel as end-all be-all architecture that would eventually replace x86, almost 10 years later its laughingstock of the industry and even the most faithful are losing hopes of it ever succeeding. The premise was that thanks to Intel's manufacturing prowess and their financial strength it would be able to overcome previous attempts to overthrow x86. Again, if Poulson(the next Itanium) executes well, it would have lost much of the hype and original steam.
Lastly, Sun's Rock with runahead threads and out of order retirement. Who knows if this will ever surface again?
I think the lesson here is that "one ring rules them all" type of approach is no longer feasible. As the devices and technologies become more complex, "innovation" is merely a tool to keep everything from stagnating.
Last edited:
Facebook
Twitter