Tensor: Details of Google Pixel 6 series-bound custom silicon potentially revealed in new leak

[Redaktion]

Google's imminent flagship phones, the Pixel 6 and Pixel 6 Pro, will be powered by Tensor, a semi-custom chipset manufactured by Samsung. Details of that chipset may have now surfaced, with information on ground hinting heavily at performance to dominate the Snapdragon 888.

https://www.notebookcheck.net/Tensor-Details-of-Google-Pixel-6-series-bound-custom-silicon-potentially-revealed-in-new-leak.560784.0.html

[Anonymousgg]

Dual Cortex-X1, that's something new. Tensor (still a bad name) is now interesting.

[Drivel2787]

Just release the final specs and the date already!  I'm holding out on the fold 3 even with a 1k trade in from AT&T because I really want to see if this is finally the year Google makes a run at a true flagship phone!!!

[Deng Frosch]

Since the Qualcomm 888 has been out for over one year it is not a fair comparison.  Why not compare Tensor with the 898 or Apples latest processor?

[ChrisGX]

Rumours, Rumours.

This just appears to be bad conjecture on Digital Chat Station's part. Since the report from Axios last year the suggestion had always been that Whitechapel/Tensor would have three clock gated core sub-clusters - 2 x A78, 2 x A76 and 4 x A55. I have never believed the A76 reference in the Axios report - it could be right but it sounds a bit mad and just doesn't ring true. Still, there was an expectation for three sub-clusters and that is what the specs sheet pointed to by DCS seems to show. It is all supposition (and probably bad supposition) that the cores clocked at 2.8GHz are X1 cores. They are probably A78 cores as the Axios report suggested.

Also, there is a reason we don't have mobile phone SoCs with 2 x X1 cores in them - heat! When the X1 bursts to peak performance it gets very hot and in consequence it throttles down. This cycle of overheating and throttling happens successively. The peak performance of the Snapdragon 888 was higher than the Snapdragon 865 (not by a lot) but sustained performance barely improved on the performance of its predecessor (owing to the aforementioned need to keep a tight reign on the chip to avoid heat build up). Putting two X1 cores in a mobile SoC would be taking a bad problem and making it worse. This all seems exceedingly unlikely to me.