If Apple doesn’t tell us how many solder bumps the die has, then it’s false advertising.
I have spoken.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
New iPad Mini Has Downclocked A15 Chip Compared to iPhone 13
- Thread starter MacRumors
- Start date
- Sort by reaction score
If Apple doesn’t tell us how many solder bumps the die has, then it’s false advertising.
I have spoken.
I'm baffled about the fact that out of all the people around here, you are the one not getting it. As a veteran CPU designer you should understand this better than anybody else: same CPU at the same clock equals same performance.
If you were buying a computer from me and I gave you two options telling you both machines had the same processor in it, wouldn't you feel cheated if after buying you discovered that the one you got has a lower clock speed?
And if you were vigilant and wanted to make sure the processor is actually the same, wouldn't you ask me if the actual clock speed on both machines is the same?
is something to be concerned over or cancel my mini order? my sons 3 year old mini is very fast as it is!
LOL No. In practice you’ll never tell the difference, and it will be way faster than any existing mini.
Even if it it were weaker, why would that be odd? But anyway, it has a GPU core over the iPhone 13 which is much more important than a small difference in clock.
I'm talking about the specs sheet.
Here's the one from the iPad mini 6:
Here's the one from the iPhone 13 Pro:
They are detailed as the same thing, but they are not. One is faster than the other.
Apple specifically said in the presentation, and on their website, that the A15 features new performance and efficiency core designs (supposedly called Avalanche and Blizzard, respectively). Seems logical that the redesigned cores allow for the A15 to be clocked higher in the smaller thermal envelope and power constraints of the iPhone. This in turn might mean that the future M2 and its variants based off of these cores could be clocked higher.
Still doesn’t explain why the Mini is clocked slower but I’ll be curious how additional benchmarks will average out.
Hardly. Cache architecture and size, memory subsystem bandwidth and latency, interprocessor and IO bottlenecks, etc. can have a significant effect on total system performance, even given identical CPU cores.
Perhaps they slowed down the CPU clocks as a trade-off against higher utilization of the 5 core GPU to maximize total system thruput per Watt on some typical apps. Or perhaps to better match a different DRAM configuration and speed grade.
What does "binning" mean, please? Is it related to what @cmaier is talking about with utilizing some older chips in a lower-volume product?
Don't pick a specific thing and take it out of context. We're talking of SOC. System on a chip. All that stuff you mentioned is part of the same package denominated A15.
If both products have the same package, performance should be the same. If not, then it should be detailed in the specs.
Honestly, it seems like a pretty minor difference so idgaf. The only thing apple did with the mini that actually deserves criticism is not upgrading or adding more storage options. That’s a lot more relevant to the buyers than the fact that the ipad mini is 7.48362% slower than the pro iphone that costs 2x its price.
More than one package is denominated A14. Almost certainly, more than one package is denominated A15.
See: https://en.wikipedia.org/wiki/Bean_machine When manufacturing a large number of "identical" chips, various parameters (max clock rate, etc.) have a statistical distribution with non-zero variance, which, after testing, can be dropped into different bins, just like a Bean machine or pachinko machine.
Yeah, not older. When I design a chip and send it to the fab, say they make 10,000 chips. They will come back so that 5 percent run really slow, 10 percent run slow, 20 percent run pretty slow, 30 percent run at the speed I was aiming at, 20 percent run a little faster, 10 percent run really fast, and 5 percent run super fast. Each of those is a bin. Like “bin 1” equals 2700 MHz, bin 2 equals 2800, etc.
Same thing happens for other properties besides speed. For example some burn a little more power than others even at the same speed. Some have broken ram cells, etc.
You can bin on whatever properties are interesting to you and sell or use the chips accordingly, giving them different model numbers if you want, even though they are nominally identical.
there are hundreds of "chips" or "die" on a wafer, every ingle one is tested for performance and such, and not every single one turns out to be performing exactly the same, some do better, some might plain fail, others perform "normal", so the die are then "binned", the best performing in Bin 1 then Bin 2 and on, this is done electronically, not physical bins.
You mean iPhone 13 not the iPhone 13 Pro. Pro has 2GB more RAM, better cameras, pro motion, faster speed A15. I think they did this so the upcoming iPad Air will have the same speed as the iPhone 13.
Someone may have already countered this...
Isn't processor speed just one of the things that can be seen in tech specs? Are you trying to say that all consumers buy the products for practical use or the physical or what is visible to them? Are tech savvy users not considered consumers? Why do some people purchase computers with Pentium processors rather than Celeron processors? Why do people go for quad-core processors rather than dual-core processors? Why do people go for Snapdragon powered phones than Exynos powered ones? I could go on and on...
They didn't do that last year. They announced the iPad Air before the iPhone 12 and talked about the A14.
This is interesting and something I didn't know. Makes a lot of sense though. Thanks!
I don’t care how tech savvy you are. You can’t tell me anything about the capabilities of a device, how fit it is for any purpose, or whether it is better or worse than anything just by knowing the clock speed of the a15 inside it. I can’t tell, and I friggen designed CPUs for a dozen years (including for AMD). All it is good for is to compare it to another a15.
However, Apple doesn't offer choices of processors for the same iPhone or iPad. M1 Macs are all the same except for the number of GPU cores. That said, binning processors improves yields and reduces costs. If some A15 chips are not quite good enough to run at full speed but have 5 good GPUs, they go into the iPad mini. Those with a defective GPU can go into the iPhone 13 (non-Pro).
This seems like a basic efficiency practice in mass production. I would imagine that every industry mass producing products has some form of this.