Peak performance isn't everything, especially for a system running on a ~40 W-h battery.
Spurred on by a recent “10% Ebay Bucks Back” promotion, I decided to upgrade beyond the first-generation Microsoft Surface Pro I’d been regularly using (albeit not heavily … I’m a mostly-Mac guy) since early January 2014. Both it and its Arm-based Surface RT sibling were first announced in June 2012 and began shipping the following February. Specifically, I went with a factory-refurbished Surface Pro 4 for $299, containing 4 GBytes of RAM and a 128 GB SSD (identical memory and storage sizes to those of the first-generation Surface Pro predecessor).
Why’d I do the deal? Let’s start off with photos. Here’s the first-generation Surface Pro:
And here’s its Surface Pro 4 successor, also with the optional Type Cover keyboard attached:
As you can probably tell right away, the displays are quite different, as are the overall device dimensions and other physical specifications. The original Surface Pro’s 10.6″ LCD was 1920×1080 pixels in resolution, i.e. a 16:9 aspect ratio. Microsoft kept these same specs for the Surface Pro 2, introduced in October 2013; for the mid-2013 Surface Pro 3 the company switched to a 12″ 2160×1440 pixel (3:2 aspect ratio) LCD. And beginning with my Surface Pro 4, Microsoft kept the size and aspect ratio pretty much the same (a thinner bezel led to a 12.3″ diagonal display) but upped the resolution to 2736×1824 pixels.
Commensurate with the first-generation Surface Pro’s 16:9 aspect ratio, the device had dimensions of 10.81″ (wide) by 6.81″ (high) by 0.53″ (thick), along with a weight of 2 lbs. The fourth-generation Surface Pro noticeably shrunk that latter dimension by around a third, to 0.33″; not quite as thin as my 9.7″ iPad Pro, but close. The Surface Pro 4’s width and height are 11.5″ and 7.93″, respectively, and its weight is 1.7 lbs.
Speaking of Apple’s tablet competition, there’s at least one other notable differentiator I’ve discovered so far. The first-generation Surface Pro contains an Intel Core i5-3317U “Ivy Bridge”-generation processor with an integrated HD Graphics 4000 subsystem. The particular variant of the Surface Pro 4 that I selected for purchase conversely is based on a Core m3-6Y30 “Skylake”-generation CPU with an HD Graphics 515 subsystem. Both processors contain dual physical processor cores along with HyperThreading support for two more “virtual” cores. The Core i5-3317U includes 3 GBytes of L3 “Smart Cache” and runs up to 2.6 GHz, while the Core m3-6Y30 embeds 4 GBytes of L3 cache but only runs up to 2.2 GHz.
Based only on the specs I’ve given you so far, you might think that the processors would perform similarly, and you’d be right, as a comparison of various peak-performance benchmark results for the Core i5-3317U and Core m3-6Y30 makes clear. But peak performance isn’t everything, especially for a system running on a 42 W-h (first-generation Surface Pro) or 38.2 W-h battery. The comparative clock speeds between the two CPUs that I’ve already mentioned are in their highest Turbo Boost modes. Conversely, the base frequencies for the two processors are 1.7 GHz (for the Core i5-3317U) and 0.9 GHz (i.e. 900 MHz, for the Core m3-6Y30). This disparity, coupled with the fact that the Core i5-3317U was fabricated on a 22 nm process whereas the Core m3-6Y30 was built on a 14 nm lithography, means that the older processor is classified as having a 17W thermal design power (TDP) specification, whereas the newer CPU is a 4.5W TDP design.
What’s this all translate into? Much longer battery life for a given-sized battery with the newer design, for one thing. Plus, thanks to the “hybrid liquid cooling system,” which appeared for the first time in the Surface Pro 4 line, the Core m3-6Y30 variant I own is completely fanless (versus its first-generation predecessor, whose fan fired up frequently). That all said, my first stab at a newer Surface Pro system ended up having at least one notable shortcoming, which I discovered shortly after I began using it. Stay tuned for more info on that (and how I addressed it) in another post to come shortly. For now, I welcome your thoughts in the comments!
—Brian Dipert is Editor-in-Chief of the Embedded Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.