Alleged Qualcomm Snapdragon 898 Geekbench score showcases remarkable improvement over the Snapdragon 888

[Redaktion]

The Qualcomm Snapdragon 898 has shown up on Geekbench. It fares a lot better this time around, with single and multi-core scores of 1,300 and 3,900, respectively. While this is better than the Snapdragon 888's score, it still falls flat against Apple's A15 Bionic.

https://www.notebookcheck.net/Alleged-Qualcomm-Snapdragon-898-Geekbench-score-showcases-remarkable-improvement-over-the-Snapdragon-888.577502.0.html

[Lazar]

I love articles like this...   new snapdragon...  one core 1900  all cores 400000000   and no info about old one... news about compersation and i need to check somewhere else how many points had old one...

[ariliquin]

Now Qualcomm has acquired ex Apple CPU architects its likely we will see many improvements. Apple should have invested more in securing their talent, even if it meant acquisition, not enough forethought.

[David Sullivan]

Remarkable in the title and then use decent is two totally different words.  You lured me in with a lie. Hope you get fired

[ChrisGX]

First, the IceUniverse leak I have seen says 1200/3900. Second, the X1 core on Samsung 5LPE was a failure. We didn't see performance in line with ARM's projections. A score of 1200 (GB5 single threaded) from X2 (probably on 4nm) while it indicates tangible performance growth (something we didn't see this year) it also shows that performance is still constrained by a) an inadequate silicon  process and b) (most likely) an X series core that is still failing to live up to expectations. If these reports are true, we can now expect to belatedly see performance gains in 2022 that we should already have seen this year.

That said and without suggesting that SD898 performance will be in the league of the Apple A16 - it won't even be close - it is the case that Apple's A series SoCs show a greater deviation between peak performance (that GB5 rates) and sustained performance than SoCs based on licensed ARM tech. If 2022 turns out to be a good year for ARM and it licensees with technical objectives largely being realised (and the deviation between peak and sustained performance reduced) performance of the SD898 could be a bit closer to the A16 in practice than the GB5 score for the SD898 would indicate. That is merely a possibility and the best case scenario for the SD898.

[Leo Kuan]

Regardless of scores, would it generate as much heat as 888 is the key question to ask of using Samsung process.