A recent report from GSMArena suggests Apple (AAPL 0.74%) will equip its next-generation iPads with an enhanced version of the A8 chip found inside the iPhone 6 and 6 Plus. The chip, unsurprisingly, would reportedly be called the A8X.
Let's dig a little bit into this rumor, shall we?
The motivation for an A8X
The A8, according to Apple, provided "up to" a 25% improvement in CPU-oriented tasks and "up to" a 50% improvement in graphics-oriented workloads. According to AnandTech's review of the iPhone 6, the A8 offers the best per-core CPU performance available in a smartphone today and is competitive with Qualcomm's (QCOM 2.55%) Adreno 420 graphics found inside the Snapdragon 805 .
However, given that a tablet can support a higher performance processor by virtue of the larger chassis and the increased battery size, it makes sense that Apple would want to include a higher-performance processor in its newest iPads.
Many have argued that the iPhone 6 Plus, which features a 5.5-inch display, might eliminate many customers' need for an iPad. Including a significantly faster processor would help Apple combat that notion.
What would an A8X look like?
My guess is that if Apple crafts an A8X for its next-generation iPads, it would improve upon the older A8 in the following ways:
- Upgrade the graphics block from the PowerVR GX6450 to the GX6650 (improving graphics performance by about 50%).
- Boost the clock speed of the CPU from 1.4GHz in the A8 to 1.5GHz. (Apple did this with last year's iPads; the A7 in the iPhone 5s clocked in at 1.3GHz, but the A7 in the iPad Air came in at 1.4GHz.)
This improvement wouldn't be Earth-shattering by any means, but such upgrades (particularly in the graphics subsystem) would be noticeable in intensive gaming applications. And remember: iPads are consumption devices, and gaming is probably the most intensive form of consumption.
Would this be economical?
As with almost every technical decision, the underlying economics are key. By including a larger graphics block, Apple would probably increase the die size of the graphics block on the system on a chip by about 50%. AnandTech measured the GPU inside of the A8 at 19.1 square millimeters, so this would add about 9.55 square millimeters to the die size.
Assuming all else is equal (disclaimer: it rarely is), this would imply an A8X with a die size of about 100 square millimeters. In an effort to quantify the financial impact of moving from an approximately 89 square millimeter die size to a 100 square millimeter die size, I turn to Silicon Edge's excellent die-per-wafer estimator tool.
According to Chipworks (via DailyTech), the A8 measures 8.5 millimeters by 10.5 millimeters. Assuming a constant aspect ratio for the new chip (this probably won't be the case, but is good enough for this purpose), the hypothetical chip would size in at about 9 millimeters by 11.12 millimeters.
This would imply about 673 die per wafer for the A8 and 598 die per wafer for the hypothetical A8X. Furthermore, if we assume that about 90% of the A8 chips on the wafer turn out good against 85% for the larger, more complex A8X, then assuming a wafer cost of $3,200 (per Handel Jones of IBS), we get the following die costs:
Calculated die cost
It seems it would cost about $1 extra per chip for the larger A8X. Now, keep in mind that there is a lot of guesswork going on here, but this is likely the right ballpark.
If a hypothetical, more expensive A8X could meaningfully add to the value proposition of the next-generation iPads, then it's probably worth the extra $1 or so per chip-
While it's unclear whether Apple will actually develop a separate chip for the next iPad, I can definitely see the case for such a move. Apple needs to offer customers a reason to buy iPads in addition to their iPhones. Though higher chip performance alone isn't likely to drive people to pick up iPads, higher chip performance coupled with higher quality iPad-specific apps that can use that extra performance could go a long way to doing so.