Simple. The iPad Pro is the top model of its range while the Macbook Air is the bottom of the Macbook range.
Doesn't look good to cripple the top model!
As mentioned above, my guess is binning. Apple expects the entry level MBA to sell in the most quantity, so it gets the chip that can best be mass-produced (ie: the 7-core M1 chip). The upgraded MBA models and the iPad get whatever is left - the 8-core M1 chip variant.Is there a rationale behind doing this?
Some chips do not pass inspection on all cores. These are “binned” and set aside for lower cost systems. It avoids waste and helps keeps costs down.Is there a rationale behind doing this?
As others have said, its to reduce the entry level price to use chips that don't have all the cores working, that would otherwise be thrown in the bin, decreasing the overall manufacturing yield and therefore the increasing the cost for everyone if they were not used.
Or to put it another way, the binned SoCs fail verification with all cores at max clock but pass with a GPU core disabled.It is very unlikely that the core on the 7-core model is defective. These cores occupy such a small part of the chip that only very few SoC will have a GPU defect.
A more likely explanation is that some chips draw more power than Apple wants them to, so they disable one core to bring the power consumption down to the spec.
The only thing that is saving the 13“ MBA and 13“ MBP is Apple allowing it to run OS X, which Apple forbids the iPad Pro to do).
Or to put it another way, the binned SoCs fail verification with all cores at max clock but pass with a GPU core disabled.
Sounds like another 8GB = 16GB moment. More likely disabled core for differentiation than having that many not pass.
Apple would be crazy to design a high volume product like $999 MacBook Air by relying on chips with a “failed” GPU core. It doesn’t make any sense.
Part binning is standard practice in the computer industry.
Yes, for processor offerings in between the low-end and high-end. Nobody designs a high volume product like $999 MacBook Air by praying for enough failed chips to meet demand. You design the chip and set a clock speed that results in a high yield to match the product with the highest volume.
And it simply doesn't make sense if you look at the area of the GPU relative to the size of M1. The logic area is far larger than the GPU.
Some people are applying a general concept in semi design without thinking about the circumstances of the situation.
You don’t need to pray for enough “failed” chips to meet demand. You simply disable enough cores in the good chips to make them on par with the failed ones in order to hit your quota.
Imagine you project to sell 100 MBA and 10 iPad Pro. Your production yield is 80 7-core M1 chips and 30 8-core chips. You then disable a core in 20 of the 8-core chips to get the 100 7-core processors you need for your MBA. As for the iPad Pro, it doesn’t matter what chip version it gets; they just all need to be the same.
If production yields of the M1 chip improves in future, we may see Apple offer a 8-core MBA as the entry level model, but users are not going to see a noticeable improvement in performance.