Notice there are a few outliers in games and also a few apps that skew the average . Once the core parking stuff and schedulers are sorted out, Ryzen will be a fine gaming CPU.
It is already nice gaming CPU and i believe 1600x stuff will bring even more epic value, but Intel is ahead in gaming performance.
I don't want to turn this into another "Ryzen" thread, but Ryzen has inherent disadvantage in memory hierarchy and that is going to hurt in games and certain apps like archivers. I believe they are going to hurt in some server workloads like JVM app servers as well, what AMD has shown us is basically a load of super parallel work - each thread works on it's own matrix memory chunk and there is little (if any?) inter core traffic. Now imagine instead JVM garbage collecting memory region that was dirtied by some CPU in other CCX and is in L2 and L3 of that chip. Broadwell and Skylake have some serious platform advantages here for these types of workloads.
You can't fix L3 cache over the night, when it is eviction cache and you can't do hw prefect to it. Same with L3 cache being split in CCX domains. Some of AMD performance is definitely coming from 512KB L2 cache, it fits more working sets and is very performant and probably does help in hiding weaker L3 and memory controller performance.
Intel has been lazy with L2 sizes, cause they were able to get away with it, server Skylake cores are finally fixing it and if they can get clocks 4+Ghz that CPU is going to perform even greater with those larger working sets.[/QUOTE]