Intel "mesh" vs intel "Ring"

[jpiniero]

Well - I guess we'll know when benchmarks come out of SKL-X on games (since a high proportion of clowns use these exclusively as reflective of "performance") as to how important latency is when the software, compiler and scheduler are optimised to keep threads on the same cores where possible and prefetch data where possible.
Oh actually no. What will happen is the same clowns will pick out inappropriate results from inappropriate code and use it to justify their pre-conceived idiocy.

Sure doesn't seem this way now. Some games got patched to be much more aggressive on keeping threads on one CCX, or maybe even cutting the number of threads. But both Skylake-X and Ryzen are both suffering quite a bit due to the latency.

 

[jpiniero]

Speaking of the mesh, here is the relevant patent related to it:

http://pdfaiw.uspto.gov/.aiw?PageNu...f=G%26l=50%26s1=20160092396.PGNR.%26OS=%26RS=

I don't know if they are really going to do the Uncore die underneath stuff but everything else is pretty realistic.

 

[moinmoin]

Extensive article on this thread's topic by ServeTheHome: The New Intel Mesh Interconnect Architecture and Platform Implications

 

[bobhumplick]

im pretty sure i saw some numbers of TR having about 300ns latency when leaving the die. also intels mesh works totally different than amd's and memory speed has nothing to do with core to core latency. also low latency ram does not help core to core or ccx to ccx latency at all, only higher speeds because it raises the speed of IF as well. core to core communication does not occur through ram accesses in ANY cpu of the 2