Fast Intel Clover Trail+ Performance Seem to Have Substance - or was the Benchmark Rigged

Richard Fieldhouse (Updated - 19 July 2013)

[Update - 19/7/13 - Things have taken a pretty sticky turn since the news of this impressive benchmark performance first appeared. It seems all may not have been quite as it first appeared. While the ARM competitors in the tests were using a standard compiler, Intel had supplied a special compiler for the tests running on their chip - and it was bypassing some of the steps..! Whoops..!

Well, it sounds bad doesn't it? It seems that some of the transfers of data back and forth to memory weren't actually happening when Intel's proprietary compiler was in use. Maybe, Intel had noticed a way to save time for any program - that would be a good thing... Or maybe, they had found a way to identify the AnTuTu benchmark code as it compiled and then they chose not to compile some of the steps. That would be very bad - probably illegal. Definitely, the word rigged would then be appropriate. We are most unlikely ever to discover what was going on, but the upshot is that the new Intel chip seems now to be around 20% slower than it seemed before. It is still good, especially compared with earlier Intel mobile chips, but it isn't as good as we thought before. ]

It takes a while for a new chip like this one from Intel to make it from launch to the point where it's powering a device you can actually buy and test. While the Clover Trail big hitters - particularly the Samsung Galaxy Tab3 series - aren't here yet, this little known phablet has put in a pretty serious benchmark score. It's really targeted at the Asian market for some reason, but they are available in the UK, albeit in rather limited numbers. (While it's faster than the reference phablet - the Samsung Note 2 - this Lenovo K900 does lack the proper stylus interface that can sometimes be very handy in these smaller screens.)

A similar positive view of the Clover Trail's performance has been been published here. This analysis also shows that it promises to use around 40% less power than comparable ARM phones and tablets. With things like the screen also being pretty power-hungry, this more efficient processor won't translate to a 40% drop in the power demand for the whole device, but it is a step in the right direction for mobile computing adding an hour or two to battery life, or allowing the device to be lighter, or to support a larger or brighter screen.

The first proper competition to ARM chip designs in the mobile space could well also lead to a further price drop for the chips and the devices they power. That would certainly seem to be the prevailing view in the financial markets, where the ARMH stock price has not fared so well over the last few weeks.

Not all the benchmarks were quite so good, mind. While the scores on AnTuTu and SunSpider were very good, the device did not do so well on Quadrant. See the results here.

And here is a wider comparison of ARM and Intel benchmarks

Compare Phablet Specifications

(© Richard Fieldhouse, 2013)