Intel’s past few weeks have seen the company enduring the ire of a large portion of the tech community, perhaps undeservedly in some instances -- certainly deservedly in others. We criticized the company for its initial marketing of the 7900X – but then, we criticize nearly everyone for marketing claims that borderline on silly. “Extreme Mega-Tasking,” for instance, was Intel’s new invention.
But it’d be folly to assume that Skylake-X won’t perform. It’s just a matter of how Intel positions itself with pricing, particularly considering the imminent arrival of Threadripper. Skylake-X is built on known and documented architecture and is accompanied by the usual platform roll-out, with some anomalies in the form of Kaby Lake X's accompaniment on that same platform.
Today, we're reviewing the Intel Core i9-7900X Skylake X CPU, benchmarking it in game streaming (Twitch, YouTube) vs. Ryzen, in Blender & Premiere rendering, VR gaming, and standard gaming.
Intel i9-7900X Specs
About That Skylake X RFID Chip
Let’s start by revisiting our earlier 7900X discussions. First, the RFID chip on the corner of the 7900X is actually disconnected – the CPU can’t read from or write to the chip, and it stores no data. We asked Intel about this at a recent event, and the company told us that the RFID chip is present because the HEDT and server CPUs use the same package, and the RFID module is used for the enterprise division. The SMBUS actually isn’t connected on the HEDT CPUs, meaning it’s a dead chip that’s only there because Intel is just making a lot of the same package for both enterprise and server. The enterprise tracking chip happens to be on both – but the SMBUS only connects on the server CPUs.
This could be something like RMA tagging or service life tags, among myriad other maintenance utilities that enterprise groups would make use of.
Intel’s Stacked Substrate for Skylake X & i9-7900X (Delidded) Die Size
We also think that this ties to the next point of interest, which is Intel’s odd substrate and package layout. When we delidded the i9-7900X, we showed that the CPU almost resembles an LGA115X substrate atop a 2066 substrate. Intel couldn’t answer why the design is like this, citing “manufacturing reasons,” which is the same reason Intel cited for usage of thermal paste rather than solder. Our present hypothesis, having not seen the server CPUs, is that the server CPUs might use more of the lower substrate and ditch the upper substrate, but Intel is keeping the same package for everything. We’re really not sure if there’s a legitimate, technical reason for this approach at this time, as Intel won’t tell us.
As for the CPUs themselves, our current understanding is that the 14C, 16C, and 18C CPUs are derived from 20-core silicon with some cores disabled, but we’re not sure about the lower SKUs – like the 7900X. We’ll learn more with time. We do have some rough die measurements, though, and the 7900X measures in at around 334mm^2. For reference, the previous generation 10C to 24C parts ranged from 246mm^2 to 456mm^2, respectively.
But anyway, the substrate design is genuinely interesting, and we have no idea why Intel took this approach. We do know that they’re still using TIM, though, and that’s something we’ll talk about in a separate video dedicated to the thermals of this CPU. The article will also contain more data on thermals.