It is. Skylake-derived Intel CPUs are fairly mediocre at this point.
You'd have to look at Tiger Lake to see some competitive numbers again.
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.
First A14 Benchmarks from iPad Air 4 Appear Online
- Thread starter MacRumors
- Start date
- Sort by reaction score
It is. Skylake-derived Intel CPUs are fairly mediocre at this point.
You'd have to look at Tiger Lake to see some competitive numbers again.
Holy heck, these new chips when put in an MBP are going to scream....
I still wouldn’t buy first gen, but it would certainly be tempting.
Yikes, whose rumor was that. The previous model was 3GB, why on earth would it jump to 6GB? There was nothing added to the Air 2020 that would demand twice the RAM. Nonsensical rumor.
What exactly does "normalized" mean in this case? Not using the features available exclusively in Intel CPUs? AVX 512 would be one example.
It may not be true for OS's, but it is often true for applications.
The contemporaneous versions of Word and Excel on my then-new mid-range 2008 based MBP were snappier than the contemporaneous versions of Word and Excel on my then-new top-of-the-line BTO 2104 MBP. Plus spinning beachballs are much more common with the latter than the former.
So in some ways the chips (at least for applications requiring single-core performance) are not keeping up with the added overhead that accompanies the newer apps.
Here's a comparison I did using the same machine (the 2014 SSD MBP listed in my profile below) with three different versions of Word, along with TextEdit and Pages. This is not disk I/O limited. After the clipboard contents are copied into Word (and I assume Pages as well), the programs have to "process" it for formatting. So nearly all of this is due to application overhead.
Note the stunning difference in overhead between Word 2008 and Word 2011/2016. My 2014 MBP would need to have a single-core speed 10x faster than that of my 2008 MBP, to have Word 2011/2016 on the former be as snappy as Word 2008 on the latter. And since single-core processing speeds didn't increase 10-fold between 2008 and 2014, we're actually going backwards here in real-world user waiting times.
Time to copy contents of 3 MB text file (114K lines) from the clipboard into:
TextEdit (Yosemite): < 1 sec
Pages (Yosemite): 10 sec
Word 2008: 14 sec
Word 2011: 142 sec
Word 2016 (Preview): 121 sec
And don't get me started on Excel. I often need to switch quickly between tabs, and the delay in doing so (as well as the delay in renaming tabs) drives me crazy.
Part of this is that Office was at the time (and I find still is) poorly optimized for the Mac. Here I performed the same test, at the same time, on the same machine, in Bootcamp under Windows 7:
NotePad (text editor, similar to TextEdit): < 1 sec
Word 2016 (Preview): 18 sec (8 x faster than under OS X)
Contributing to the problem is that most of the increased processing power offered by newer chips over the past decade has come from more cores, rather than increased single-core speed. There are reasons for this but, regardless, this means most of the advancement in chip performance doesn't benefit Office programs, which (with the exception of certain functions in Excel) are purely single-threaded (yes, an open Word application will have many threads, but only one of these can be running at a time).
Last edited:
I agree with the other poster — the Pro is clearly a better value, and it will still be relevant for many years to come.
For reference, I still have an iPad Air 2 from 2014 running an A8X that performs well with zero problems.
Watching Excel take perceivable time to move between cells and PowerPoint beachball on my 2017 5K SSD-only iMac earlier today made me wonder if these apps will ever feel snappy enough.
the whole RISC vs CISC thing is an anachronism. Nobody takes it seriously any more.
intel’s complex instructions are split in to simpler micro-ops internally, until the result looks very RISC-like. At the same time, ARM has introduced all kinds of extensions such as pointer authentication (not to mention other instructions introduced by licensees), which looks very CISC-like.
ultimately it’s all about chip designers, and where they feel the appropriate balance is for each chip they design.
Until Microsoft also realizes that using reference designs won’t cut it. I wouldn’t be surprised to find out they’re working on their own 64 bit only ARM version of a system that will run Microsoft Office and Xbox Game Pass.
Priced right, it would sell tons.There don’t need to be many switchers for Apple to maintain their current macOS marketshare... around half of all folks buying a Mac every year are new to Mac, most likely having never owned a computer before. Plus, remember, the big future goal for Apple is to move everyone over to iOS. And, when you consider that there’s more iPads sold than all laptops, they’re well on the way to that end.
except the whole point of the benchmark is to make it possible to compare benchmarks across architectures...
Not sure what's speculation and hoping when it comes to
- single core speed
- mulit core speed
- clock rate
- TDP
Each of these are hard facts & numbers, nothing about it is speculation.
About putting 9 A14s in one machine, admittedly, that's wishful thinking and won't happen -it's just not as simple as that. But Apple making an A15x with 12 or 24 cores is definitely future expansion possibility
I think they'll hit the 3Ghz wall also by the way, so they will need to scale on cores and multi core scores. Even if the 3Ghz wall has now been expanded to 5Ghz under certain right conditions and circumstances, for short boost times, etc. The fact is around 3Ghz it becomes increasingly more difficult to up clock rates as certain physical limits are encountered.
agreed, never before devs have had such a tech stack API that largely handles under hood: lazy evaluation, performant memory management, even multithreaded and asynchronous operation with minimal boiler plate code...
In fact, I think making a very badly performing application with today’s APIs and cpu power boils down to not knowing properly some key basics.
Last edited:
That is what I am looking forward to see the next generation iPad Pro models. The iPad Air 4 is a good deal, but its not a iPad Pro replacement.
Last edited:
Why do the newer iPads bend; is it the square, hollow aluminum chassis? My iPad 6 is a workhorse. It can handle a lot. Although it did incur the bright, white spot and the home button. I hope Apple solved this as well. Super annoying.
There isn’t a single Mac, period.
But the Mac Pro’s strength is in many-core applications, anyway.
Apperantly the middle part of the ipad is the weakest, where the microphone is placed on the side. Not sure why but apple says bending is normal but wont cover it as a defect.
Bending is normal. Amazing. They say the same about the bright white spot, until trade-in time.
We (at least many of us) used to have to turn off automatic recalculation to be able to make spreadsheets usable, esp if you were using certain exponential(?) functions. I’m probably thinking of Lotus 1-2-3, but iirc F9 was recalc.
You’d add some rows, copy/fill, etc. then hit recalc. Maybe a second, maybe much longer. You’d literally see the changes ripple through on-screen as the cells recalculated 😡 Good times 🤣
“Bending is normal” is not something Apple has ever said. Nor have they said the same about any “bright white spot.”
[automerge]1601765524[/automerge]
i forgot about that!
Used to also be an issue where you could create calculation loops, and freeze lotus. And yes, I believe it was f9.
Wow, a 2.99GHz chip almost twice as fast as a 1.8GHz chip, imagine that. The skilz Apple has.
You need to look up "premature optimisation".
Now imagine a z series of the A14 on upcoming MacBooks with thermal solution. Gonna be awesome!