4 ms·
Multithreaded applications tend to share a lot of data between threads, and we already know Intel's mesh fabric core interconnect[1] seems to be inferior to AM
by throwaway2048 7y ago
Multithreaded applications tend to share a lot of data between threads, and we already know Intel's mesh fabric core interconnect[1] seems to be inferior to AMD's chiplet+io die in terms of performance and access latency uniformity.
[1] used in all 10core+ Intel CPUs.
- wmf 7y agoThe 10900 uses a ring and IMO the mesh is superior to Infinity Fabric.
- lliamander 7y agoAMD's 4000 APUs are monolithic and don't use separate chiplets and IO die. The chiplet + infinity fabric has higher latency than Intel, but that's not a problem for the APUs.