OK, so Intel has launched their 8th gen. CPUs today. Unlike what some expected, the launch does not include the new “Coffee Lake” CPUs, but a Kaby Lake refresh. The new KL CPUs belong to the less power hungry “U” series, but with double the core count. This is basically the same KL architecture and CPU, with some tweaks and improvements in manufacturing. The improvements will result in higher clocks, if nothing else, of at least 0.2GHZ. All the new chips launched today have the same TDP as before of 15W (up top 25W cTDP) and have basically the same iGPU, maybe with a small clocks increase.
In fact, the 8th generation of Intel CPUs will include a mix of Kaby Lake refreshes, Coffee lake and even the new node “Cannon Lake” 10nm based CPUs. Intel said the additional CPUs from the the generation will come through the end of the year, but it’s unclear when exactly Coffee Lake and Cannon Lake will be released. Some slides show 10nm process only in 2018 and that means we’ll see the Cannon only in 2018. Just a reminder – “Coffee” is another 14nm refinement of the same process, while “Cannon” is a new microarchitecture (some details on the Cannon Lake desktop CPUs already leaked). For those who look for some of the new juice, I’d suggest waiting.
As far as performance hoes, Intel slides talk about 40% better performance compared to the 7th gen. U CPU. The performance bump is comprised of doubling the core count (+25% performance, per Intel claims) and manufacturing and and design improvements which add another 15%. The compared parts are the new Kaby Lake I7-8550U and the old, 7th gen, Kaby Lake I7-7500U. I guess the doubling of cache size could also help.
The interesting question is what will be the performance in practice. True, the new KLs are much more ample, but they are limited by the same 15W TDP. The 7th gen KL “U” 2c/4t CPU (like I5-7500U) could be limited by the TDP under very high load of GPU + CPU (link), but maintained high clocks under CPU load only (link), so a 4C/8T CPU will be considerably more limited if all of the CPU’s cores are utilized (or even 3 of them). Intel’s own numbers show that too – from doubling the cores you should get 100% performance bump if they are well utilized. We only get 25% according to Intel. Intel’s ark site shows a cTDP up frequency of 1.8GHZ for the I5-8250U in comparison to the 2.7GHZ in the I7-7500U (link) I don’t really understand why the extra HT cores are needed here and it almost seems to me like the marketing prevailed over other aspects once again.
But, I guess that these CPUs will shine in a more moderate situations, with shorter periods of very high loads, where system heat doesn’t get to build up. Such scenarios could be the common everyday use (like browsing), light video editing, compiling projects, and stuff like that.
I wonder what will the HQ series will bring – perhaps the same CPUs only with a 35-45W TDP instead of 15W? maybe 6 cores? We’ll have to wait and see, but I at this point, I would much rather have a more efficient, higher clocked 4C/4T /8T HQ CPU, then another bump in core count. I guess Intel is trying to block AMD core war from burning everything.
As of the gaming laptops aspect, I think these are no real news. The 15W limitation would be a problem with current systems and even for the lower end gaming laptops (for $400-$500), that’s probably not an interesting offer, because the GPU is fast enough to see benefits from more cores. For now, goodbye.