Thanks for the link pax.
@Nunz
So I looked into Ghostwire Tokyo to try and get a better idea of how Intel's 12900K handles the game and here is a link to CPU usage on a 12900K for this game. https://www.youtube.com/watch?v=4LW1UU2Uzsg
It looks like the game uses all 8 cores and 16 threads generally between 10% and 60% load simultaneously. The 12900K in question is running at 5.1Ghz on 8 P cores.
Here is a link to Ghostwire Tokyo on a 5950X. Interestingly it's using 10 cores but it doesn't show thread load. https://www.youtube.com/watch?v=-91VcNf832k&t=641s
In this last interview Robert changed his wording from all 32 threads at 5.5Ghz to all threads used by the game at a solid 5.5Ghz. So I'm guessing the argument about weather it could run at 5.5Ghz all core with only 8 cores loaded up was a pointless argument because it appears that it WAS using PBO after all. However this does clear up that the game uses 8 to 10 cores and up to 16 threads which were all running at 5.5 Ghz on the 7950X during the demo.
Intel's full throttle all core mode is pretty insane regarding power and honestly I don't want anything like the 12900K when it comes to multithreading because I could run into prolonged CPU use at all core full load usage. If they improve the all core power usage to a more usable level then I wouldn't have a problem with the 13900K.
I will concede that Intel's game mode does have some wiggle room because the chips overall power usage is not going to be stressed when playing games. I don't doubt that people have overclocked these chips to run at 5.5Ghz on 8 P cores when gaming and it's possible that Intel could push 5.5Ghz on the P cores with 5.8Ghz single core out of the box as long as those cores clock down if they are fully loaded. In order for Intel to be competitive to me and my cash they would also need to achieve a heavy reduction of power via 10nm optimization. If the 13900K is anything like the 12900K regarding heat and power in full load multithreading then I won't be interested.
Regarding AMD's 15% or greater thread performance. If the minimum boost of 5Ghz occurs when in all core multithreading then minimum thread performance would also be in effect during all core multithreading. Which means the closer you get to single core performance the higher the clock boosts and the higher the thread performance. If AMD can manage a solid 5 Ghz in 16 core load and 5.5Ghz on 8 core load then I would assume they will hit 5.7Ghz on a single core. They are achieving at least 15% improvement per thread at 5Ghz boosts so if you add another 700Mhz in single core mode they could be close to 30% single core performance increase over the 5000 series. They gave an example of their mulit-threading performance running 40% faster than the 5000 series. Do you think this sounds sensible or do you think I'm out of my mind? Also of note, I'm aware that the example they gave of multithreading performance could be a best case scenario but we should also take into consideration that the chip will probably improve a bit before launch. Anyway, enjoy the memorial day weekend, I know I will.
@Nunz
So I looked into Ghostwire Tokyo to try and get a better idea of how Intel's 12900K handles the game and here is a link to CPU usage on a 12900K for this game. https://www.youtube.com/watch?v=4LW1UU2Uzsg
It looks like the game uses all 8 cores and 16 threads generally between 10% and 60% load simultaneously. The 12900K in question is running at 5.1Ghz on 8 P cores.
Here is a link to Ghostwire Tokyo on a 5950X. Interestingly it's using 10 cores but it doesn't show thread load. https://www.youtube.com/watch?v=-91VcNf832k&t=641s
In this last interview Robert changed his wording from all 32 threads at 5.5Ghz to all threads used by the game at a solid 5.5Ghz. So I'm guessing the argument about weather it could run at 5.5Ghz all core with only 8 cores loaded up was a pointless argument because it appears that it WAS using PBO after all. However this does clear up that the game uses 8 to 10 cores and up to 16 threads which were all running at 5.5 Ghz on the 7950X during the demo.
Intel's full throttle all core mode is pretty insane regarding power and honestly I don't want anything like the 12900K when it comes to multithreading because I could run into prolonged CPU use at all core full load usage. If they improve the all core power usage to a more usable level then I wouldn't have a problem with the 13900K.
I will concede that Intel's game mode does have some wiggle room because the chips overall power usage is not going to be stressed when playing games. I don't doubt that people have overclocked these chips to run at 5.5Ghz on 8 P cores when gaming and it's possible that Intel could push 5.5Ghz on the P cores with 5.8Ghz single core out of the box as long as those cores clock down if they are fully loaded. In order for Intel to be competitive to me and my cash they would also need to achieve a heavy reduction of power via 10nm optimization. If the 13900K is anything like the 12900K regarding heat and power in full load multithreading then I won't be interested.
Regarding AMD's 15% or greater thread performance. If the minimum boost of 5Ghz occurs when in all core multithreading then minimum thread performance would also be in effect during all core multithreading. Which means the closer you get to single core performance the higher the clock boosts and the higher the thread performance. If AMD can manage a solid 5 Ghz in 16 core load and 5.5Ghz on 8 core load then I would assume they will hit 5.7Ghz on a single core. They are achieving at least 15% improvement per thread at 5Ghz boosts so if you add another 700Mhz in single core mode they could be close to 30% single core performance increase over the 5000 series. They gave an example of their mulit-threading performance running 40% faster than the 5000 series. Do you think this sounds sensible or do you think I'm out of my mind? Also of note, I'm aware that the example they gave of multithreading performance could be a best case scenario but we should also take into consideration that the chip will probably improve a bit before launch. Anyway, enjoy the memorial day weekend, I know I will.
Comment