Upcoming Chromebook tablet ‘Ciri’ gets impressive Geekbench ML score

We’ve been waiting a good, long while on the next wave of Chromebook tablets to arrive. We first found evidence of a new MediaTek chip in a detachable form factor via the ‘Geralt’ baseboard and now we’re tracking ‘Ciri’ as the primary offshoot of that board that looks to be made by Lenovo and could be something like a Duet 3 successor.

Along the way, we’ve taken a look at some performance benchmarks for ‘Geralt’ and ‘Ciri’ and what we’ve seen has been encouraging. This processor isn’t as powerful as the Kompanio 1380 in the Acer Chromebook Spin 513, but it shares the same cores in a different configuration. Instead of 4 large cores and 4 smaller ones, the new MT8188 goes with 2 large cores and 6 small, meaning MediaTek is leaning in a bit more on battery life and efficiency with these new devices.

Big gains in ML benchmarks

One area of Geekbench I rarely look at, however, is their ML (machine learning) scores. With this new MT8188 processor getting some ML and AI treatment (both MediaTek Neuron and Google TensorFlow will be built into this chip), it looks like there will definitely be some gains in the AI department for this processor when it does land for ‘Ciri’ in the (hopefully) near future.

For reference, I’ve included Geekbench 6 ML scores for the MediaTek Kompanio 1380 (‘Cherry’), the Intel Core i3-N305 (‘Nissa’), and the 12th-gen Intel Core i5-1240P (‘Brya’) alongside the new MediaTek MT8188 (‘Geralt’). And ‘Geralt/Ciri’ handily beats them all in the ML category.

And by ‘beats’, I mean beats down. Though ‘Geralt’ isn’t as fast with standard CPU processing as what we see with ‘Brya’ or ‘Cherry’, it does hold its own with ‘Nissa’ (specifically ‘Pujjoteen’ – a.k.a. the Slim 3i Chromebook Plus) in that department and completely obliterates them all when it comes to AI and ML sorts of tasks.

How this will affect the day-to-day operations of ‘Ciri’ will only be seen once we eventually get hands on with this device. I’m hopeful this means that some of that constant ML processing will be easily unloaded on the cores meant for ML and AI tasks, leaving a bit more raw performance for the CPU for daily duties. For now, we just don’t know, but it seems MediaTek and Google’s work to add some additional AI hardware with ‘Ciri’ is paying off.