Why is Kabylake-X a downgrade from Skylake-X?

Looks like Kabylake-X is a client core with a high power socket and possibly better cooling. The Skylake-X chips looks like server cores. What's the difference? Client cores have the cache subsystem etc tuned for client loads, mostly better latency. Inclusive caches etc. Server cores are made for throughput, usually have significantly higher latency thanks to that. This shows up in typical client loads like games, latency sensitive stuff.

in comparison, Ryzen is more of a server chip than a client one. Kabylake-X could be interesting if it over-clocks very well. Typical client loads like high FPS gaming will run like hell on it. There is a compelling use case. Add the other integrations like Optane etc.

Expensive as frakk, but probably the best overclocking client core on the market. Over-clockers like the LN2 guys will love it.

wow they have really degenerated from the x79/x99 era.

Haswell-DT (4790k and so on) are almost the same size as Ryzen 8 core is, and no they don't use TIM because the dies are too small, they use it because they're cheap.
AMD uses thermal paste too, but again that is for "small dies", small dies can handle solder but why bother with such cheap parts.

Essentially intel is selling cheap parts for 300+ dollars.

It's important to also point out that AMD only recently switched to using TIM on small-die chips. As late as 2015 they were selling Sargas series Semprons with a soldered IHS.

Linus shares my sentiments. Next time I see a leak that I think it's too stupid to be true, it probably is true.

5 Likes

The only answer you need is "Intel".

intel is a mess right now my friend. I say GO AMD good on you for fucking with them. they needed it

(this coming from a guy who actually loves intel and everyone here knows i run intel..)

Alternative answer: "Money."

1 Like