INTEL REACTS TO RYZEN
When AMD launched Ryzen, Intel to its credit never blinked. Instead giving AMD a long, cold and hard stare down, confident that its product line-up would withstand whatever AMD could throw at it. Not budging on prices once to lower itself to AMD’s undercutting ways. And to be honest, it has stood its ground well.
However, cracks are beginning to show in Intel’s product oerings where AMD can demonstrate clear performance superiority, all thanks to simply having more cores on oer at the same price.
Well it seems this great showdown has come to a head and Intel is finally reacting. Intel has o cially announced the first products in its new eighth generation Core series. As has been the method since Intel dropped its infamous tick-tock R&D cycle, we’re now starting the second generation of processors built using the PAO cycle (process, architecture, optimise).
The processors announced are at the start of this cycle (‘process’) and thus essentially take the previous 7th generation optimised cores (Kaby Lake) and simply re-create them again but on a newer process, in this case 14nm++ (Kaby Lake Refresh). As such, architecturally these aren’t expected to be exciting in anyway shape or form. The excitement comes in what an improved process node aords. Namely better power consumption (thus clock speed headroom) and better transistor density.
With all that in mind we have a very clear idea of what’s going on with these four new processors (see table 1).
INTEL 8TH-GEN CPUS
Aimed at ultra-portables, over the 7th gen equivalents these impressively gain two extra cores (and four threads) eectively doubling the amount of raw resources available, plus the peak turbo boost speeds are increasing some 500MHz! This comes at the cost of base clock speeds, which to fit into the tiny 15W power envelope, suer a massive 900MHz hit, now down to to sub-2GHz levels.
This is just the first bullet in Intel’s chamber though. Leaks are already beginning to uno cially confirm coming 8th-gen desktop parts specifications. Clearly showing a reactionary move to AMD’s Ryzen product lines where core counts help them beat Intel’s oerings, Intel is basically adding two extra cores to everything across its line-up (see table 2).
This is a big moment, as until now Intel has simply refused to budge beyond quad-core processors for its mainstream parts for a good six years now. AMD has facilitated the start of the CPU core wars - a war Intel intends to win if clock speeds are anything to go by - those peak turbo clocks are far beyond what AMD’s Ryzen can achieve currently, even when overclocked.
These 8th gen parts unfortunately will require yet another chipset (and thus motherboard) to work, being the 300 series. And thanks to AMD’s sneaky marketing ploy in claiming the B350 name first, Intel has had to alter course from its previous B150 and B250 chipset naming convention, and opt for calling its next mainstream chipset B360.
THREADRIPPER SUCKS ON TWO DUMMIES
Sitting in the TR4 socket – being the same socket which AMD’s own Epyc server processors also use (to save on R&D and assist in ecosystem uptake of the platform) – the fact that Threadripper is just half of an Epyc processor poses a slight problem. You see, Epyc is an MCM package made up of four dies (of eight cores each). Physically removing half (two) of them poses a problem for distributing the heat to and physical clamping forces from a TR4 heatsink.
Enthusiasts wondered if AMD simply used failed Epyc processors with over half of their cores disabled to create Threadrippers (expensive!). It wasn’t long before someone de-lidded a Threadripper and discovered the truth. AMD simply use two dummy cores, placed diagonally opposite each other in place of where Epyc’s two extra dies go, so that they help prop up the IHS (Integrated Heat Spreader) and distribute clamping loads evenly over the actual working cores. A simple yet eective solution!
Table 2: And the desktop parts
A die shot of Intel new 8th-gen quad core U-series processor October 2017 www.pcandtechauthority.com.au