This site may earn affiliate commissions from the links on this page. Terms of utilize.

An Avoton (Silvermont) wafer

For the by few months, details on the Goldmont architecture (that's the Atom microarchitecture refresh that replaced Silvermont this twelvemonth) have been extraordinarily scarce and difficult to come past. After doing full deep dives on past architectures, including the Silvermont update that powered Bay Trail and Cherry Trail devices, Intel refused to talk almost Goldmont in any significant degree at all. While it's true that these chips are being relegated to the lower-cost Pentium and Celeron lines as part of the Apollo Lake platform,  Goldmont hardware volition still bulldoze millions of devices over the next few years — and it's been interesting to sentry the evolution of Intel's small-cadre chips in relation to their big-core counterparts.

Now, thank you to Intel's updated x86-64 programming guides, we've gotten a look at what the new chip tin do and how it differs from Silvermont, which starting time debuted in 2013. Goldmont incorporates a number of improvements over Silvermont, though some of the diagrams are a bit bare-basic compared with what Intel typically creates for consumer publications:

Goldmont-Diagram

Where Silvermont (and AMD's Kabini / Jaguar / Puma) were all dual-issue decoders, Goldmont has three decoder units, and a maximum of 20 bytes decoded per cycle. The fetch and instruction cache pipelines are no longer coupled, large page back up accept both been added, and there's a minor L2 "precode" cache (16K) that didn't exist on prior Cantlet processors. Goldmont'due south triple-wide decoder is matched by its ability to retire up to three instructions per cycle, and the scrap is capable of executing one load and store per clock wheel (Silvermont could only perform one load or store per clock cycle). Three simple integer operations can be executed per cycle and accost generation is at present out-of-order in Goldmont (Silvermont generated and scheduled memory addresses in-order, but could complete them out-of-order.)

Goldmont-Silvermont

Goldmont also has generally improved instruction latencies (how much depends on the instructions in question, but some of the gains are considerable) and can decode a maximum of two branches per bicycle (Silvermont was limited to i). Overall, Goldmont is a much smaller gain over Silvermont than Silvermont is compared with the original Cantlet cadre, Bonnell — but it's simply fair to note that significantly less fourth dimension has passed between the debut of Bonnell (2008) and Bay Trail (2013) equally compared to Silvermont (2013) and Goldmont (2016). Intel also hasn't gone through near as many process node transitions. Bonnell was a 45nm core compared with Silvermont'southward 22nm, whereas Goldmont is a 14nm chip.

Reading the tea leaves

The Goldmont-Silvermont shift is conceptually similar to the upgrades AMD made to its 40nm "Bobcat" CPU when information technology built the 28nm Jaguar follow-up. While the ii companies fabricated unlike changes to their underlying architectures, in both cases, Intel and AMD chose to heighten and upgrade the basic designs they'd previously deployed rather than making a huge prepare of changes or taking a dramatic jump forward.

Before this year, I speculated that Intel may have been forced to pull Goldmont CPU clocks down compared with Silvermont, considering the CPU performed more piece of work per bike and had more difficulty hitting loftier frequencies within previous TDP ranges as a outcome. This could notwithstanding be true — generally speaking, nosotros'd expect Goldmont to exist at least modestly faster than Silvermont on a clock-for-clock footing, and higher efficiency CPU architectures often use more than ability at the same clock speeds than lower-efficiency cores.

How much any of these issues specifically impacted Goldmont, or whether the new compages's functioning influenced Intel's determination to impale Atom's smartphone and tablet hardware divisions is still open to discussion. Checking Intel'southward various Cantlet pages, it'due south clear that "Atom" is being phased out as a separate make — new fries haven't been launched in the desktop or conventional mobile markets for years, and there are only a scattering of rebranded 28nm Rockchip designs on the Smartphone and Tablet Ark page. That said, Intel is offer a 1.2GHz 28nm chip every bit part of its Rockchip understanding, where it previously topped out at ane.1GHz. Presumably it created the higher-stop SKU for a reason, but whether or not the company is shipping any volume on these parts at all is an open up question. It probably isn't.

Every bit for whether the Goldmont architecture has a future past 14nm or non, Intel really hasn't said. On the one paw, having a low-stop Pentium and Celeron-class cadre bachelor lets Intel position those systems every bit meaningfully dissimilar (and much less expensive) than Core i3/i5/i7, and it may continue to design Goldmont-derived CPUs for that reason alone. On the other, however, both Intel and AMD have moved abroad from multi-CPU architectures as they've abandoned the tablet and smartphone market. It may be that in the futurity, Intel will simply address this space with derivatives from Core chips, simply as it used to practise before Cantlet came on the scene dorsum in 2008.