Snapdragon 8 Gen 3: Leak reveals clock speed and core configuration info of Qualcomm's next-gen flagship SoC

[Redaktion]

The Snapdragon 8 Gen 3 has been tipped to debut much later in the year, around October. The next-gen flagship chipset also looks set to edge Apple's A16 Bionic in the CPU department, and a new leak has now revealed more details of the SoC.

https://www.notebookcheck.net/Snapdragon-8-Gen-3-Leak-reveals-clock-speed-and-core-configuration-info-of-Qualcomm-s-next-gen-flagship-SoC.698970.0.html

[Randy chan]

Regarding the upcoming Qualcomm snapdragon 8 gen 3 chip will definitely be a high performance screamer as even the current 8 gen 2 chip is awesomly powerful. Unfortunately for the 8 gen 3 chip, it's performance will take a back seat to Apple's upcoming A17 bionic chip which got preferential treatment for the 3nm process which means the 8 gen 3 will remain at 4nm. Despite that, customers that purchase a 8 gen 3 powered android phone will still have a great phone they can be proud.

[Aliyu]

It's funny how when Companies catch up or surpass Apple, then Apple are given special preferential treatment. Look at the 3nm process and the new Geekbench 6.

[ikjadoon]

It's funny how when Companies catch up or surpass Apple, then Apple are given special preferential treatment. Look at the 3nm process and the new Geekbench 6.

Arm Ltd. designs 99.99% of all Android devices' CPU uArches and Arm Ltd. relies on Geekbench. So does NUVIA. You all are arguing against the very architects of major non-Apple SoCs.

https://fuse.wikichip.org/wp-content/uploads/2022/06/arm-cortex-x3-ipc.png

Prove this inane claim that Geekbench 6 has a hidden, non-performance-based benefit to Apple SoCs. We'll wait. "But, see it's different now!" As it should be. The multi-core changes are much more consumer representative.

//

Apple literally pays billions for a node advantage. Anybody can set-up a multi-year agreement with TSMC, with their billions in tow, and then take the advantage from Apple.

Bringing a cutting-edge node online + fab expansion costs more than the GDP of small countries ($33 billion). Maybe AMD or Intel or Qualcomm can go rob a bank? Or sell more CPUs at higher prices, I guess.

[Mr Majestyk]

Apple paying a ultra premium for TSMC's 3nm node is in no way a smart move. Let them have it, their iSheep buyers can pay through the nose for a bleeding edge and frankly disappointing node update. Far smarter would be to use the highly refined N4P node that costs a lot less and is not far behind in density or power efficiency. Many experts in the field are questioning the immense amounts of money TSMC is investing in the 3nm node for some of the worst ever generational scaling improvements. Even worse SRAM has totally stagnated.

Qualcomm hopefully will use N4P for Snadragon Gen 3 and maybe next year can pick up N3E for a far better price for Gen 4.

[Anonymousgg]

Apple paying a ultra premium for TSMC's 3nm node is in no way a smart move. Let them have it, their iSheep buyers can pay through the nose for a bleeding edge and frankly disappointing node update. Far smarter would be to use the highly refined N4P node that costs a lot less and is not far behind in density or power efficiency. Many experts in the field are questioning the immense amounts of money TSMC is investing in the 3nm node for some of the worst ever generational scaling improvements. Even worse SRAM has totally stagnated.

Qualcomm hopefully will use N4P for Snadragon Gen 3 and maybe next year can pick up N3E for a far better price for Gen 4.

TSMC has made incremental node improvements for Apple for years. If N3B (original N3) is even slightly better than the N4 nodes, all the better for Apple. The effect on the BOM cost will be minimal and Apple is clearly doing well in smartphone sales. Qualcomm and MediaTek can stay one step behind and still deliver SoCs with more performance than almost anybody needs in a smartphone.

The nodes to get excited about are the TSMC N2 nodes using GAAFETs for the first time. The first-generation N2 will deliver bad scaling but a decent efficiency increase and slight performance over N3E. Then they will probably increase density on future N2/N1 nodes.