If you look at the distribution, it’s multimodal. A lot of tests seem to be done under suboptimal conditions and produce less then ideal performance. For comparing the actual potential of the chip it makes sence to isolate the normal component with highest scores. For A17 Pro this appears to be 2925 with a standard deviation of around 10. For A16 it’s somewhere around 1630.
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.
M3 Chip Generation - Discussion Megathread
- Thread starter scottrichardson
- Start date
- Sort by reaction score
If you look at the distribution, it’s multimodal. A lot of tests seem to be done under suboptimal conditions and produce less then ideal performance. For comparing the actual potential of the chip it makes sence to isolate the normal component with highest scores. For A17 Pro this appears to be 2925 with a standard deviation of around 10. For A16 it’s somewhere around 1630.
Interesting points. On the other hand, we have geekerwan admitting they rushed the review and that it probably had errors. I think we need to wait for more reliable testing.
Hmmm I’m not sure we can take one result as being representative over the official figures from geekbench.
I mean, geekerwan admitted they phoned it in so...
Might be worth remembering that Tom’s got an official review unit. I’d be surprised if geekerwan did. So how did they get their unit and is it a pre-production unit, is it running the latest os etc. Very unreliable.
There is no "official" score. The number you shared is an average that changes daily. It includes submissions based on low power scores, low battery scores, throttled scores, cryo cooled scores.
The typical GB6 ST score for A16 running iOS 17 is about 2,600.
The typical GB6 ST score for A17 Pro running iOS 17 is about 2,850.
What...?
Geekerwan is one of the most popular reviewers in the scene. They were literally at the launch event on September 14. It's very known Chinese reviewers have better media access to Apple due to the size of the Chinese market. That's why you see these one-on-one interviews with Tim Cook you never see with western reviewers.
It’s as close to an official score as we have given it’s: from geekbench; and an average. It doesn’t change often after this many tests.
It’s too early to say what a typical test is for the A17 chip. We know that many of the tests were conducted soon after the phone has been turned on, within the first couple of days. This means the chances that many background processes were still taking place is very high which could skew the average towards the lower end. We will need to see after many hundreds, or even thousands of tests have taken place to judge the “typical“ a17 scores.
A16 scores changed because iOS 17 bumped up the score. It often changes with new iOS versions.
The average has not changed yet. Perhaps it will in the future as more people use iOS 17, but we can only judge it based on the figures we currently have.
Do you think the reviewers were born yesterday? All of them who submitted GB scores? Or that Apple doesn't include such instructions and notes in the package they send to reviewers?
I mean some of them aren’t particularly technical. All of them are rushed (Tom’s, geekerwan as they admitted) and benchmarking is full of nuance.
That doesn’t change the fact that given the rush you admitted the reviewers face, the benchmarks displayed lower scores early on than later. The effects of background set up processes on benchmarks are well known.
Bad news for all.
A17 Pro doesn't seem much better than A16 and yet, Qualcomm is now performing better for both CPU and GPU especially since A17 Pro consume way more power. I have to say since M1 released, so many developers left Apple and I think that might be the cause.
Don't expect too much from M3 series.
Better to wait for a review that doesn’t admit they rushed it and probably had errors. Especially when it seemingly contradicts other reviews.
Last edited:
Lorem ipsum dolor sit amet, consectetur adipiscing elit.
Last edited:
Perhaps you want to look up what Jeremiad means? Hint - spelling it correctly will help.
All reviewers make mistakes from time to time. So I wonder why Geekerwan's review is criticized so much. Is it because they're not Anandtench? Or because their review doesn't favor Apple? Or because they're Chinese?
Who could be?
We’re getting PS4 Pro-level of performance in a device that eats at most 8-10W, I don’t understand why people are complaining so much 🤷🏻♂️
Lorem ipsum dolor sit amet, consectetur adipiscing elit.
Last edited:
“B-b-but they lost one of hundreds of chip engineers a year ago, thing X that bugs me is clearly because of this” <- a common line of thinking her at MR.
For me, there are too many conflicts with previously established baselines. The power consumption for A14/A15 reported by Geekerwan is dramatically different from what Anandtech reported for similar tests (by almost 80%)! E.g. Andrei measured over 4Watts for these chips running SPEC (and close to 5W in FP tests), Geekerwan only reports under 3 watts (7:18 in the video). These numbers do paint a daunting picture for Apple's 3N product, showing A17 pro being 20% less efficient than A16.
Now, at around 8:20 he reports a different set of numbers (I assume total device power draw?), which are much closer to Anandtech measurements. According to them, in SPECint A17 Pro is 10% less efficient than A16, 5% less efficient than A15, and about as efficient as A14. In SPECfp, A17 is as efficient or more efficient than there predecessors.
If we go from this second set of numbers (which IMO make more sense), it would appear that Apple is now targeting peak performance while trying to roughly maintain efficiency to previous generations. Maybe this is what the new CPU u-arch are about? More performance? Apple M-series have been criticised for their relatively low performance ceiling given the chassis capability, especially on desktop. Maybe the new core is designed to address those limitations. For the phone, slightly higher peak power consumption won't matter. For Mac, it could mean a nice performance boost.
P.S. Speaking of performance. Those SPEC2017 numbers quoted by Geekerwan are absolutely insane. That's comparable to an Intel i9-13900K, a 5.8Ghz enthusiast desktop GPU that draws over 30 watts for the same tests. I really don't understand how one can be disappointed by this level of performance at 5x lower power draw. People shoulder really adjust their expectations. Apple engineers seem to be held to such a high standard that it's considered a disappointment if their smartphone does not outperform an overclocked x86 workstation while consuming less power than an ARM mid-performance core. Come on.
Last edited: