Apple A13 SoC

The Twitter user Longhorn (@never_released) has information about the A13.
@never_released said:
Apple A13 is T8030. Its CPU codename is Lightning.
@never_released said:
Apple Lightning (next-gen CPU) has hardware tracing.
@never_released said:
Apple A13's codename as a whole is Cebu. It has Thunder and Lightning cores.
Links: [1] [2] [3]

There is discussion on the Real World Tech forums about these details including the potential implications of the core codenames. Neither codename is about wind so maybe they indicate larger architectural changes than the past five generations?
 
Apple is in big need of ew tentpole features to drive device upgrades. Unless hardware tracing helps deliver some whiz bang new feature, may not be the best use of R&D.

Apple needs to find way to deliver products at lower price points. iPhone XS and XS Max sales may show they can't charge whatever they want.

An Apple SOC in a Mac laptop may deliver better performance than low end Intel chips but they're going to need to lower costs so they can lower prices.
 
Apple needs to find way to deliver products at lower price points.

How about a phone based on the 4S shell with upgraded screen and internals? :p Something with actual grip and thickness for battery and lenses too :rolleyes: The new screen types would probably put the 4S chassis at the size close to the viewable area of the 5/6. *cough*

sigh duck.

Anyways.
 
I'm just saying the days of "cost no object" in SOC design and production may be coming to an end for Apple.
 
From Slashleaks: "IPHONE XI REDESIGNED LOGIC BOARD LEAKS OUT."

iphone-xi-logic-board-leaks-out-34.jpg


Which part of the logic board corresponds to the SoC? Is it the large 2:1 aspect ratio rectangle in the middle of the board?
 
Wonder what “high bandwidth cache”, “advanced silicon packaging” and “high bandwidth unified memory” mean from the feature mosaic.
 
Wonder what “high bandwidth cache”, “advanced silicon packaging” and “high bandwidth unified memory” mean from the feature mosaic.
High Bandwidth Cache is probably just a larger L3, and High Bandwidth Unified Memory is probably just LPDDR4X 4266MT/s.
Had they gone with LPDDR5 I think they'd boost up the GPU performance a bit more than 20% and they'd need less L3 for the GPU.
 
Geekbench 5 result.
I really miss the memory benchmark numbers of GB4 (and also feel that they belong in a single FigureOfMerit number if such a number is provided, and it is.)
If GB reads the cache sizes correctly (GB4 above) there is an interesting story to be told there.
I wonder how the new floating point capabilities of the CPU will be accessed.
Also, since they haven’t done the switch to LPDDR5 yet, there is some extremely low hanging performance fruit hanging around for future models.
 
Back
Top