Don't trust those 'leaked' Microsoft Surface benchmarks (or any benchmarks)

Surface Pro 6
Surface Pro 6 (Image credit: Daniel Rubino / Windows Central)

Ahead of every big device launch, there comes the inevitable speculation on what new hardware and features are on the horizon. Watching the pace of the evolving tech landscape is what makes covering it so much fun, and this annual cycle is very much a part of that.

However, there's one area of the annual refresh hype train in which it's healthy to practice some genuine skepticism: benchmark "leaks."

A tale of two benchmarks

This week, we saw two alleged benchmarks pop up for what are reported to be upcoming Microsoft Surface devices. One is alleged to be for a Surface Go that sports an Intel Core m3 processor, while the other is thought to be either a Surface Pro or Surface Laptop with one of Intel's new "Ice Lake" CPUs.

Benchmarks can be, and have been, faked.

Both are upgrades that would make sense for this refresh cycle. Plus, with this year's Surface event less than a month away, it's not surprising that we might see Geekbench scores starting to trickle out (these scores were dated April and June, respectively). Digging up benchmark scores from Geekbench, however, doesn't necessarily indicate anything concrete.

To put it simply, benchmarks can be, and have been, faked.

This is particularly true in the Android space, but it can be done with PC hardware, too. Even if Geekbench reads directly from a processor for an ID, the data eventually has to pass through a network and application to be reported, which means it's possible for someone who wants to expend the time and effort to manipulate it.

Fakes and mistakes

Surface logo

Surface logo (Image credit: Windows Central)

That might sound like a lot of work for little gain, but it's been done before. Someone in the Android modding community recently faked a benchmark for a Xiaomi "POCO F1 Lite" by running a Geekbench test on a phone that had some system files modified to report a different device ID. I'll save you the dirty details, but XDA Developers has a great rundown of the whole saga, along with some examples of just how easy it is to pull this off.

Benchmark leaks are an exceedingly unreliable indicator of what's coming down the pike.

There's also the complication of reports that might be legitimate but reflect a configuration that never makes it to market. While it's unclear whether this was precisely the case with the new Samsung Galaxy Book S, it was widely rumored that the laptop would ship with a Snapdragon 855 chip based on a "leaked" benchmark. Once the Galaxy Book S was announced, it was revealed to be running on the Snapdragon 8cx platform.

As for why anyone would want to fake a benchmark, it's most likely "for the lulz" and trolling tech blogs, as was the case in the Xiaomi POCO F1 Lite example. What matters is that benchmark leaks are an exceedingly unreliable indicator of what's coming down the pike, and it's worth exercising an abundance of caution when you see reports based on them.

Circling back to Surface, it's certainly possible that a Core m3-based Surface Go is in the works. Likewise, it would make sense for Microsoft to slot Intel's new Ice Lake chips in a new Surface Pro or Surface Laptop. But take any benchmarks you see with a massive grain of salt.

As for me, I'm eagerly awaiting a tease for the rumored dual-screen Surface "Centaurus" ... though I'll beware any leaked benchmarks

All the new Surface hardware we expect to see on October 2 in NYC

Dan Thorp-Lancaster

Dan Thorp-Lancaster is the former Editor-in-Chief of Windows Central. He began working with Windows Central, Android Central, and iMore as a news writer in 2014 and is obsessed with tech of all sorts. You can follow Dan on Twitter @DthorpL and Instagram @heyitsdtl

  • You shouldn't trust anything leaked, period.
  • Newer saw the leaks, you should perhaps include the alegied results, or at least say something about twm being better, worse or just as expected. Or if the point was to know the CPU used in each model.
  • Really appreciate the follow up on this by WindowsCentral. I saw it blasted all over the front page of other, less reputable MS news sites and was hopeful that someone could corroborate, verify, or refute. The only source to express any kind of skepticism or critical analysis was this one. Everyone else was parroting the original source of the Geekbench find, probably to garner more clicks. That said, I am hopeful that it is representative of what a Surface model can offer. 25w TDP seems to be a pretty significant step up compared to the 15w based on the limited benchmarks on Intel reference units.
  • The article writer referenced 1 confirmed faked and the other suposedly (Samsung Galaxy S with 8CX SoC versus leaked 855 SoC) faked. In the case of the Galaxy S laptop, I'll counter contend that it might be possible that Samsung modelled 2 chipsets and went with 8CX? In all, you did not show a wide spread proof of faked benchmark data or even 10% of it for that matter to justify wide spead unreliability argument of these benchmark results based on your one or couple of sited fake results.
  • This article is a little lacking in content no? No theories about what would be likely components in the new surfaces? I personally hope they add thunderbolt across the line this time, upgrade the surface book to processors on par with current 8 core MacBook pro, and improve the digitizer so that it's on par with the Apple pencil experience. Definite Insta buy for me then.
  • Want a surfacebook type device that is the dimensions of surface laptop
  • As the article says, those are both things that are likely to show up in the Surface event on October 2. Nothing is certain, but I'd be more surprised if they didn't arrive. The Surface Pro is ripe for a significant redesign, and given the timing it's likely that it will include Ice Lake. (I consider it more likely than Comet Lake because of Ice Lake's much better graphics performance, plus the early reports that suggest big improvements in battery life for Ice Lake systems.) I also expect to see USB-C make an appearance, hopefully with Thunderbolt 3 support, but I expect that Microsoft will also keep at least one USB-A port and the Surface Connect port; their business-oriented customer base would not appreciate an Apple-style abandonment of legacy ports. (The Mini DisplayPort will probably be history, but since you usually have to carry a dongle for that anyway it's not a major sacrifice to carry a USB-C dongle instead.) Putting the bezels on a diet will make it look more modern; I think Microsoft will go with a larger display in the same size package rather than shrinking the system. The Surface Go is a newer design so I don't expect major changes this year. The small form factor has been a success, and a version with more speed might convince more people to give it a go (sic). The Surface Go is a premium product in surprising ways -- a display with an excellent color gamut, super-fast WiFi, and build quality far better than usual at its price point -- so the platform can support a performance option that will cost a bit more. The m3 version of the Surface Pro is discontinued, so there will still be enough of a performance gap between an enhanced Surface Go (Surface Go Faster?) and the Surface Pro line for product differentiation.
  • I look forward to Centaurus as well. Even if Centaurus is not a Surface Phone, releasing a dual-screen device will be the first step, as well as being motivation to hopefully accelerate the creation of a SDK for it and increase interest (for both a Surface Phone as well as dual-screen devices in general).
  • Windows Central seemed convinced that the next Surface Go would be ARM-based, now everyone seems convinced it will be an m3-8100Y. But what about Lakefield? That launches in 2019 and sounds ideally suited to it. Intel would need a flagship product to show off its hybrid architecture, and a cheaper Microsoft tablet would be great for that too. There would be no issues with improved cooling needed for an m3 in a 10" form factor, and no issues with compatibility/performance concerns with a switch to ARM.