Interesting article on next gen threadripper architecture

[Dann Corbit]

https://www.overclock3d.net/news/cpu_ma ... adripper/1

Very positive spin. If the assertions are correct, the new threadripper will be a monsterpiece.

[mwyoung]

https://www.overclock3d.net/news/cpu_ma ... adripper/1

Very positive spin. If the assertions are correct, the new threadripper will be a monsterpiece.

It is already a monster. What the next gen will offer is fixing the memory controller issues in the 24, and 32 core versions that hurt performance in some application. Along with shrinking the die, better clock speed, and IPC, and I am sure more cache . The question is with the smaller die will we see a 64 core Threadripper? That would be a monster killer.

[mwyoung]

https://www.overclock3d.net/news/cpu_ma ... adripper/1

Very positive spin. If the assertions are correct, the new threadripper will be a monsterpiece.

It is already a monster. What the next gen will offer is fixing the memory controller issues in the 24, and 32 core versions that hurt performance in some application. Along with shrinking the die, better clock speed, and IPC, and I am sure more cache . The question is with the smaller die will we see a 64 core Threadripper? That would be a monster killer.

Here are the best C/ R 20 scores right now. The 24 core and 32 core already dominate. And the scores hold true as long as the your application does not have suffer from the memory controller issues of the 24, and 32 cores versions. Chess programs don't seem to suffer on Threadripper gen 2.

With the issues being fixed with the gen 3 Threadripper. The Threadripper scores should go up dramatically looking at how AM4 ryzen 3rd gen progressed. What will be Intel's answer?

Cinebench best R20.jpg