I looked @ some of the posts here and I felt like a ?self predicting roadmap? about AMD, its only out of fun so don?t get all stressed , itsjust for fun, please no flame war, had one the other day (in person). Someone tried to argue a 2.8C is about 20% faster then a Newcastle 3200+ and that?s the reason he?s buying Intel, Mhz still matters in some peoples eyes.
A 4000+ ( 2.6Ghz), 4200+ ( 2.8Ghz, by this point I predict the gap between the P4 and A64 will be even more, considering how well a 200 Mhz boost in clock speed for a K8 does more then a 200 Mhz speed bump for any Netburst architecture based CPU) You have to remember back in the K7 days, one of the reasons the performance crown was lost was due to the lack of ability to scale, the .13 micron process was fine, but you noticed the athlon xp only went up in 66 Mhz jumps or around that for another 100+ points, which is true and deserving with the higher IPC, however when the thorton core hit 2Ghz ( 2500+ and later the 2800 was it ? 2.25 Ghz ?) AMD knew the K7 was running out of headroom for clock speed. Enter barton core, slap more cache on and drop some clock speed for reserve and take smaller jumps, and maybe a few more Mhz for the FSB. If the throton had carried on, a 3200+ would have been more like 2.5ghz, maybe this could have been achieved, however with the poor test run of yields for the K8 they needed to lower the bar, making the clawhammer/hammer look like it had some muscle ( which it sure does) on its debut. AMD is slowly getting faster speeds on .13 micron process and once 90 nm is matured there?s the hope 3.2 Ghz will be achieved, anyway back to the point !
90 nm improvments include:
Full SSE3 implementation
* Improved hardware data prefetch mechanism
* Increased number of writing combine buffers (D0 stepping A64's can now combine up to four non-cacheable streams compared to 2 on the C0 and CG stepping A64's)
* Improved on-die memory controller with more advanced open page policy
* On-die thermal throttling
* Black Diamond Low-K technology (slower less power hungry transistors in less used sections and faster and more power hungry transistors in frequently used sections of the cpu)
A 3Ghz should acquire the 4400 + tag which by now is well deserving, this is attainable, some early samples have hit 3Ghz from the 0.9 process, so come q2/q3 yields should be sufficient, and a 4600 (3.2 GHz) would be quite an achievement for AMD, but with FAB 36 not online till late 06, they have to stretch the 0.9 micron process to its limits, there are no plans to convert FAB 30 over to 65 nanometre tech. However you can see AMD doing what they did with the barton core , adding more cache ( brining back an modified clawhammer core and keeping speed reserve in the tank but keeping the numbers going up). I should think the 4400+ would be out in q4.
Dual core, I have no idea what the plans are, June/ July is pinned as the date, but software has to be there to make it mainstream to take the advantages/reap the benefits from two cores. Server market will adjust quickly, but as for desktop, fill me in someone ?
Anyone want to write up a Intel road map for 05/early 06 ?
A 4000+ ( 2.6Ghz), 4200+ ( 2.8Ghz, by this point I predict the gap between the P4 and A64 will be even more, considering how well a 200 Mhz boost in clock speed for a K8 does more then a 200 Mhz speed bump for any Netburst architecture based CPU) You have to remember back in the K7 days, one of the reasons the performance crown was lost was due to the lack of ability to scale, the .13 micron process was fine, but you noticed the athlon xp only went up in 66 Mhz jumps or around that for another 100+ points, which is true and deserving with the higher IPC, however when the thorton core hit 2Ghz ( 2500+ and later the 2800 was it ? 2.25 Ghz ?) AMD knew the K7 was running out of headroom for clock speed. Enter barton core, slap more cache on and drop some clock speed for reserve and take smaller jumps, and maybe a few more Mhz for the FSB. If the throton had carried on, a 3200+ would have been more like 2.5ghz, maybe this could have been achieved, however with the poor test run of yields for the K8 they needed to lower the bar, making the clawhammer/hammer look like it had some muscle ( which it sure does) on its debut. AMD is slowly getting faster speeds on .13 micron process and once 90 nm is matured there?s the hope 3.2 Ghz will be achieved, anyway back to the point !
90 nm improvments include:
Full SSE3 implementation
* Improved hardware data prefetch mechanism
* Increased number of writing combine buffers (D0 stepping A64's can now combine up to four non-cacheable streams compared to 2 on the C0 and CG stepping A64's)
* Improved on-die memory controller with more advanced open page policy
* On-die thermal throttling
* Black Diamond Low-K technology (slower less power hungry transistors in less used sections and faster and more power hungry transistors in frequently used sections of the cpu)
A 3Ghz should acquire the 4400 + tag which by now is well deserving, this is attainable, some early samples have hit 3Ghz from the 0.9 process, so come q2/q3 yields should be sufficient, and a 4600 (3.2 GHz) would be quite an achievement for AMD, but with FAB 36 not online till late 06, they have to stretch the 0.9 micron process to its limits, there are no plans to convert FAB 30 over to 65 nanometre tech. However you can see AMD doing what they did with the barton core , adding more cache ( brining back an modified clawhammer core and keeping speed reserve in the tank but keeping the numbers going up). I should think the 4400+ would be out in q4.
Dual core, I have no idea what the plans are, June/ July is pinned as the date, but software has to be there to make it mainstream to take the advantages/reap the benefits from two cores. Server market will adjust quickly, but as for desktop, fill me in someone ?
Anyone want to write up a Intel road map for 05/early 06 ?
Facebook
Twitter