My beef with A10X

[macduke]

First of all, I want to say that I think the A10X is an impressive chip—probably even the fastest ARM chip in the world right now. I'm sure it will probably handle my workload fine. What I'm getting at here has to do with longevity.

My beef with the A10X is that, in 2016, we had the iPad Pro 9.7" debut with an A9X, and about six months later the iPhone 7 came out with a roughly 10% faster A10 chip. This iPad Pro was understandably marketed to Pro users, with a premium price over the older model, and yet was bested by a phone within six months. Will the same happen with the A10X and the A11, but only after three months? I have to wonder.

Does anyone else feel like Apple held the iPad Pro 2 update for WWDC so they could get more people to buy them after showing off iOS 11 features? I wonder if they had this A10X ready a lot earlier. This cycle was longer than normal and seems weird. I get the sense that Apple probably had a good portion of this new iPad Pro multitasking and drag and drop stuff ready last summer, but it wasn't quite ready, so they held it. People who have worked inside Apple say they do this all the time, so it's not unusual, and explains why the early betas seem to work so well. So they spent all this effort on the iPad software, and in the meantime sales just kept tanking. When it came time for new hardware, they just modified the A10 to save money. If Apple seriously wants to go after Pro users, well—there's a lot of stuff they should do that I won't go into here and that was partially addressed with iOS 11—BUT what they should't do is let this alleged professional hardware get usurped by a phone within three months. That's all I'm saying.

It's a really fast device, and Apple makes the best chips on the market, but given the context of what the device is supposed to be, it's just really weird how it gets owned by a phone so quickly. Especially since traditionally the iPad was faster than the iPhone, and now the Pro model is slower after six months, and possibly 3 months? Hopefully it at least ships with a decent amount of RAM for working with bigger files and projects. If it still has 2GB I'm sending it back, lol, as even the iPhone Plus has 3GB now.

Alternatively, I wonder if there will be much in the way of speed improvements in the A11. They might have put a lot into power efficiency, as there is that new rumored (was it U series?) machine learning chip that might eat into the power budget of the A11. And probably that chip is integrated into the A11 so they didn't want to out it early with the iPad. Honestly, I wish they had just held the iPad Pro 2 until October if that were the case. It's just a very weird strategy, but I ordered one because I passed on the 9.7" with 2GB RAM and really want to use the Apple Pencil.

Edit: Wrong chip number, thanks Relentless!

 

[Appleaker]

It'll be disappointing if A11 is better than the A10X. And with the 10th anniversary, they might be looking for impressive improvements over the A10. And based on their previous trends, this is the year that gets the performance increase. I feel like it may be around/ just over A10X performance.

 

[44267547]

First of all, I want to say that I think the A10X is an impressive chip—probably even the fastest ARM chip in the world right now. I'm sure it will probably handle my workload fine. What I'm getting at here has to do with longevity.

My beef with the A10X is that, in 2016, we had the iPad Pro 9.7" debut with an A9X, and about six months later the iPhone 7 came out with a roughly 10% faster A11 chip. This iPad Pro was understandably marketed to Pro users, with a premium price over the older model, and yet was bested by a phone within six months. Will the same happen with the A10X and the A11, but only after three months? I have to wonder.

Does anyone else feel like Apple held the iPad Pro 2 update for WWDC so they could get more people to buy them after showing off iOS 11 features? I wonder if they had this A10X ready a lot earlier. This cycle was longer than normal and seems weird. I get the sense that Apple probably had a good portion of this new iPad Pro multitasking and drag and drop stuff ready last summer, but it wasn't quite ready, so they held it. People who have worked inside Apple say they do this all the time, so it's not unusual, and explains why the early betas seem to work so well. So they spent all this effort on the iPad software, and in the meantime sales just kept tanking. When it came time for new hardware, they just modified the A10 to save money. If Apple seriously wants to go after Pro users, well—there's a lot of stuff they should do that I won't go into here and that was partially addressed with iOS 11—BUT what they should't do is let this alleged professional hardware get usurped by a phone within three months. That's all I'm saying.

It's a really fast device, and Apple makes the best chips on the market, but given the context of what the device is supposed to be, it's just really weird how it gets owned by a phone so quickly. Especially since traditionally the iPad was faster than the iPhone, and now the Pro model is slower after six months, and possibly 3 months? Hopefully it at least ships with a decent amount of RAM for working with bigger files and projects. If it still has 2GB I'm sending it back, lol, as even the iPhone Plus has 3GB now.

Alternatively, I wonder if there will be much in the way of speed improvements in the A11. They might have put a lot into power efficiency, as there is that new rumored (was it U series?) machine learning chip that might eat into the power budget of the A11. And probably that chip is integrated into the A11 so they didn't want to out it early with the iPad. Honestly, I wish they had just held the iPad Pro 2 until October if that were the case. It's just a very weird strategy, but I ordered one because I passed on the 9.7" with 2GB RAM and really want to use the Apple Pencil.

In your second paragraph, the iPhone 7 has the A10 Fusion chip, not the A11 (But that's coming).

Although, I do see what you're saying. And ultimately I think the A11 Chip would debut for the iPhone first over the iPad. Also, the iPad is updated less frequently and perhaps Apple see's the iPad getting the secondary internal hardware once the iPhone has its debut first. I know there was a lot of speculation the A11 was To be the most anticipated Chip, especially for upcoming capabilities.

But I do think the new 10.5 iPad will have at least 3 GB of Ram. I also read somewhere it was rumored to have 3 GB of Ram. I would like to think Apple would continually upgrade the Ram new versions. I was rather disappointed when the 9.7 Pro only had 2 GB at a $650.00 Price point, aside from Tru-Tone, quad speakers and Pencil Support.

I assume Geekbench results will release early next week with the new iPads, along with ifixits tear downs. Stay tuned. Much more to debut.

 

[BigAppleNYC123]

I think Apple has been very consistent in this pattern. Do you just take issue with it, or are you surprised also?

I'm trying to get a sense of what A chips had the biggest leaps over their predecessors. It is my impressions A8 to A9 is the biggest in recent years, while A9 to A10 is more comparable in impact to A7 to A8. Anyone have any thoughts ?

 

[profets]

Does anyone else feel like Apple held the iPad Pro 2 update for WWDC so they could get more people to buy them after showing off iOS 11 features?

That's an interesting theory. I think the iOS 11 improvements definitely sparked most of the excitement for iPad.

However, I also feel that if they had iPads ready earlier, they probably would have preferred to make them part of the (non) March event and let this keynote focus more on software.

It was a great keynote with a lot of new stuff, but when there's so much and you have to rush through everything, it kind of suffers as well.

After the March 2016 event with the 9.7" iPad Pro launch and iOS 9.3, it almost seemed like it was defining a new pattern for iPads - every spring (exactly half the year separated from iPhone launch) and a .3 iOS update for new iPad features (after .0 focuses on iPhone). But March 2017 & 10.3 came and went and it seemed like they weren't able to have the new Pros ready (or new iPad features for that matter).

 

[JPack]

Semiconductor process technology works in concert with SoC design. Some process nodes have bumps that slow yields and clock frequencies.

For example, the Samsung 14-nm process used on the Apple A9 has at least four generations:

• Low Power Early
• Low Power Performance
• Low Power Compact
• Low Power Ultimate

To get from LPE to LPC, it took about a year.

A9X or A10X doesn't just pop out of nowhere. It relies on mature and ready foundry processes.

 

[macduke]

In your second paragraph, the iPhone 7 has the A10 Fusion chip, not the A11 (But that's coming).

Although, I do see what you're saying. And ultimately I think the A11 Chip would debut for the iPhone first over the iPad. Also, the iPad is updated less frequently and perhaps Apple see's the iPad getting the secondary internal hardware once the iPhone has its debut first. I know there was a lot of speculation the A11 was To be the most anticipated Chip, especially for upcoming capabilities.

But I do think the new 10.5 iPad will have at least 3 GB of Ram. I also read somewhere it was rumored to have 3 GB of Ram. I would like to think Apple would continually upgrade the Ram new versions. I was rather disappointed when the 9.7 Pro only had 2 GB at a $650.00 Price point, aside from Tru-Tone, quad speakers and Pencil Support.

I assume Geekbench results will release early next week with the new iPads, along with ifixits tear downs. Stay tuned. Much more to debut.

Thanks, I fixed the error. So many freaking chip names going on!

I really hope it has 4GB. I didn't think 3GB was in the cards. 3GB would be a big "F U guys" IMO because they raised the price again. You would think the slightly bigger enclosure would mean bigger battery and they could afford it in the power budget, especially if a Plus phone can power 3GB, lol. Yeah, next week the Geekbenches will drop, if not earlier leaks late this week or over the weekend. Depends when the embargo goes up. But even the testers now will probably have some results going into the system that will be identified at some point. At least Geekbench has been updated to support testing for 4GB already, so it's just a matter of running the bench, right? At least that's the way it used to work.

I think Apple has been very consistent in this pattern. Do you just take issue with it, or are you surprised also?

I'm trying to get a sense of what A chips had the biggest leaps over their predecessors. It is my impressions A8 to A9 is the biggest in recent years, while A9 to A10 is more comparable in impact to A7 to A8. Anyone have any thoughts ?

This site gives a good sense: https://browser.geekbench.com/ios-benchmarks

As you can see, even the year old A8X barely bested the iPhone 6s. To me, that's about the way it should be. The iPad Air had the same chip as the iPhone 5s, but with faster clocks, and they came out at about the same time. I would consider that acceptable, but still not ideal with a "Pro" machine. In the 4th gen iPad, the A6X was basically the same as the A6, just clocked a little faster with a much better GPU that better handled the retina display than the third gen iPad that was released 6 months earlier. As you can see by the pattern, last year was the first year that the iPhone surpassed the iPad in the same year. Technically I suppose the A9X released in autumn 2015 in the 12.9" model about six months earlier than the 9.7" model. But even so, what we're looking at right now is an A10X that might get knocked out by the A11. Again, that might not be the case after all, but looking at the pattern, it seems likely, no? And very strange for a "Pro" device. It's basically like Apple releasing a MacBook Pro, then three months later releasing a MacBook 12" that is faster than the 15" quad-core Pro. It's just weird, lol. That being said, again, the fact that this iPad is faster than the MacBook is nothing to sneeze at either.

That's an interesting theory. I think the iOS 11 improvements definitely sparked most of the excitement for iPad.

However, I also feel that if they had iPads ready earlier, they probably would have preferred to make them part of the (non) March event and let this keynote focus more on software.

It was a great keynote with a lot of new stuff, but when there's so much and you have to rush through everything, it kind of suffers as well.

After the March 2016 event with the 9.7" iPad Pro launch and iOS 9.3, it almost seemed like it was defining a new pattern for iPads - every spring (exactly half the year separated from iPhone launch) and a .3 iOS update for new iPad features (after .0 focuses on iPhone). But March 2017 & 10.3 came and went and it seemed like they weren't able to have the new Pros ready (or new iPad features for that matter).

Yeah, you would think so, especially with iPad sales sinking. But considering the longer than usual upgrade cycle, one has to wonder. That and there was the rumored March event. Many rumors about this, and some of them came from rather reputable sources that were shocked when it didn't happen and we had that mini release. Really strange and again, leads me to think that maybe the A11 won't be a speed demon over the A10 and we probably won't see another iPad Pro update for 18 months or so. Maybe by then they'll get back on track with an A12X releasing not long after the A12, lol.
[doublepost=1496779556][/doublepost]

Semiconductor process technology works in concert with SoC design. Some process nodes have bumps that slow yields and clock frequencies.

For example, the Samsung 14-nm process used on the Apple A9 has at least four generations:

• Low Power Early
• Low Power Performance
• Low Power Compact
• Low Power Ultimate

To get from LPE to LPC, it took about a year.

A9X or A10X doesn't just pop out of nowhere. It relies on mature and ready foundry processes.

Thanks for your insight into how the chip process works. If that is the case, however, wouldn't it be best if Apple just waited until the A11 that they put in the iPhone, and package it with the extra graphics cores? The A10 did more with fewer cores than the A9X did with three. At least, I assume when you're talking about compactness you're talking about core density?

 

[007p]

Not sure why it matters if the A11 is faster? It will be in the phone. It's not like another iPad will come out in 3 months time? Look at how long the 12.9" lasted with the A9X, we could not see another iPad Pro until the A12x is ready at this rate.

 

[ABC5S]

Some of you guys just analyze things too much. I turn on my iPad and it works for what I need to do and that's just fine. I don't go and analyze anything that may come out just around the corner or many months from now. That's technology as I figure and I don't get bent out of shape for it.

 

[gobikerider]

First of all, I want to say that I think the A10X is an impressive chip—probably even the fastest ARM chip in the world right now. I'm sure it will probably handle my workload fine. What I'm getting at here has to do with longevity.

My beef with the A10X is that, in 2016, we had the iPad Pro 9.7" debut with an A9X, and about six months later the iPhone 7 came out with a roughly 10% faster A10 chip. This iPad Pro was understandably marketed to Pro users, with a premium price over the older model, and yet was bested by a phone within six months. Will the same happen with the A10X and the A11, but only after three months? I have to wonder.

Does anyone else feel like Apple held the iPad Pro 2 update for WWDC so they could get more people to buy them after showing off iOS 11 features? I wonder if they had this A10X ready a lot earlier. This cycle was longer than normal and seems weird. I get the sense that Apple probably had a good portion of this new iPad Pro multitasking and drag and drop stuff ready last summer, but it wasn't quite ready, so they held it. People who have worked inside Apple say they do this all the time, so it's not unusual, and explains why the early betas seem to work so well. So they spent all this effort on the iPad software, and in the meantime sales just kept tanking. When it came time for new hardware, they just modified the A10 to save money. If Apple seriously wants to go after Pro users, well—there's a lot of stuff they should do that I won't go into here and that was partially addressed with iOS 11—BUT what they should't do is let this alleged professional hardware get usurped by a phone within three months. That's all I'm saying.

It's a really fast device, and Apple makes the best chips on the market, but given the context of what the device is supposed to be, it's just really weird how it gets owned by a phone so quickly. Especially since traditionally the iPad was faster than the iPhone, and now the Pro model is slower after six months, and possibly 3 months? Hopefully it at least ships with a decent amount of RAM for working with bigger files and projects. If it still has 2GB I'm sending it back, lol, as even the iPhone Plus has 3GB now.

Alternatively, I wonder if there will be much in the way of speed improvements in the A11. They might have put a lot into power efficiency, as there is that new rumored (was it U series?) machine learning chip that might eat into the power budget of the A11. And probably that chip is integrated into the A11 so they didn't want to out it early with the iPad. Honestly, I wish they had just held the iPad Pro 2 until October if that were the case. It's just a very weird strategy, but I ordered one because I passed on the 9.7" with 2GB RAM and really want to use the Apple Pencil.

Edit: Wrong chip number, thanks Relentless!

Yah if history continues the A11 will be the next big leap in processor.

 

[JPack]

Thanks for your insight into how the chip process works. If that is the case, however, wouldn't it be best if Apple just waited until the A11 that they put in the iPhone, and package it with the extra graphics cores? The A10 did more with fewer cores than the A9X did with three. At least, I assume when you're talking about compactness you're talking about core density?

A11 will be built on a cutting edge 10nm FinFET process. iPad Pro needs a chip with low cost, high clock speed or core count, and high yields. You can't have all three at the same time on day one.

The foundry (TSMC) needs time pump out mature 10nm FinFET products. Then they can optimize the process to manufacture a large die or high speed A11X. If Apple wanted a performance version of A11 in Sept 2017, it would result in very low yields and high cost.

 

[jamesrick80]

At this point most ipads are fast and continue to have under-utilized power. No one should be conflicted about the processor speed. Some of us would prefer a much better display and design and they could have kept the same A9X in the latest ipad if they gave me OLED, trackpad/mousepad support, a real edge to edge design, a real new apple lighted smart keyboard and decreased pencil size, and an increased ppi.

I have too many devices now with OLED, no way I am going back to IPS displays with the exception of the chromebook plus (really lovely display).....

 

[Andres Cantu]

GPU is also a consideration (the AX chips have much more GPU power), but I do find it hard to believe that the A10X, with a whole extra core, only offers 30% improvement over the A9X. Perhaps the GeekBench numbers will tell a different story.

 

[macduke]

A11 will be built on a cutting edge 10nm FinFET process. iPad Pro needs a chip with low cost, high clock speed or core count, and high yields. You can't have all three at the same time on day one.

The foundry (TSMC) needs time pump out mature 10nm FinFET products. Then they can optimize the process to manufacture a large die or high speed A11X. If Apple wanted a performance version of A11 in Sept 2017, it would result in very low yields and high cost.

But they used to do simultaneous or near-simultaneous chip launches for both products. Is the complexity/density of the new designs to blame?
[doublepost=1496794977][/doublepost]

But I do think the new 10.5 iPad will have at least 3 GB of Ram. I also read somewhere it was rumored to have 3 GB of Ram. I would like to think Apple would continually upgrade the Ram new versions. I was rather disappointed when the 9.7 Pro only had 2 GB at a $650.00 Price point, aside from Tru-Tone, quad speakers and Pencil Support.

I assume Geekbench results will release early next week with the new iPads, along with ifixits tear downs. Stay tuned. Much more to debut.

Not sure if you saw it, but Serenity Caldwell on Twitter confirmed 4GB of RAM in both. I'll take that. At least the Pro machines are both on a more equal footing now from a developer perspective, which should open up more opportunities for advanced Pro applications to unleash the power.

I'll be interested to see how fast the NAND is. I still need to pick up a USB 3.0 SD adapter.

 

[thunng8]

GPU is also a consideration (the AX chips have much more GPU power), but I do find it hard to believe that the A10X, with a whole extra core, only offers 30% improvement over the A9X. Perhaps the GeekBench numbers will tell a different story.

Apple always quotes single core performance in their improvement figures. So 30% increase in single core and much higher for multi core.

Same thing happened with a8x to a9x. Apple quoted 80% improvement.
Also in their comparison of a8 (mini 4)and a8x (iPad Air 2) they only said a8x was less than 10% faster than a8 despite 2 vs 3 cores

 

[JPack]

But they used to do simultaneous or near-simultaneous chip launches for both products. Is the complexity/density of the new designs to blame?

Apple never used leading edge process tech until the A8/A9. Prior to that, Apple had always used n-1 or n-2 process nodes.

With the A11, they are using 10nm. Apple will be using the latest process node available.

 

[gobikerider]

GPU is also a consideration (the AX chips have much more GPU power), but I do find it hard to believe that the A10X, with a whole extra core, only offers 30% improvement over the A9X. Perhaps the GeekBench numbers will tell a different story.

I know it sounds small but actually is quite a nice speed bump that 30% is about 1400 extra on geekbench

 

[rui no onna]

Apple always quotes single core performance in their improvement figures. So 30% increase in single core and much higher for multi core.

Same thing happened with a8x to a9x. Apple quoted 80% improvement.
Also in their comparison of a8 (mini 4)and a8x (iPad Air 2) they only said a8x was less than 10% faster than a8 despite 2 vs 3 cores

In all fairness, few apps can fully utilize multiple cores. There's usually a task bottlenecked by single-threaded performance. Hence, aside from software optimization, better single-threaded performance is partly why Apple tri-core or dual-core chips tend to be faster than octacores used in Android devices.

 

[JulianL]

I wonder whether the absolute performance jump in A10 to A11 will be as much as people think. Don't forget that this is the point where iPhone is rumoured to jump from current whatever percent of front coverage the current iPhone 7 screen is to virtually 100% front screen coverage in the anniversary edition iPhone (or whatever it ends up being called). That means a lot more screen to drive with pretty similar internal volume available for battery. (There are also rumours that the screen assembly on the next iPhone will be thicker so the rumoured increase in case thickness may mostly go to accommodate the thicker screen assembly rather than a bigger battery.)

All this makes me think that, although I'm sure that the excellent Apple engineers will again push forward strongly in terms of performance per Watt on the A11 vs the A10, much of that improvement this time might go towards keeping down power consumption rather than looking for really big delivered performance gains so that the much bigger screen doesn't damage battery life. I actually hoping that Apple will even be aiming for a position to be able to boast of class-leading battery life even after the big screen size increase.

 

[DavesIknow]

Some of you guys just analyze things too much. I turn on my iPad and it works for what I need to do and that's just fine. I don't go and analyze anything that may come out just around the corner or many months from now. That's technology as I figure and I don't get bent out of shape for it.

iPad Pro 12.9 Version 2
IPad Pro 9.7
IPone 7 Black Matte 128 GB - Week 40

Whats an Ipone? Does it have the A11?

Sorry couldn't help my self

 

[macduke]

If this is real, this iPad Pro 2 benchmark is a pretty awesome result: http://browser.geekbench.com/v4/cpu/3036382

Multi-core: 9091
Single-core: 3832
RAM: 4GB

 

[Kal-037]

At the end of the day the iPhone will always get better specs as that is Apple's money maker... BUT, even though the A10 was clocked higher than the A9X, my iPP still worked better and without glitches. The iPhone 7 and A10 have been nothing but buggy and glitchy since day one, (and I don't think it's because I got a dud or anything, as it still works better than my older iPhones.) The A10X is the biggest beast Apple has made (obviously each new chip is,) but the X series always take it to a new level. Sure the A11 may get a higher clock speed for iPhone 10-year, but in day to day usage the iPad has greater performance IMO.

 

[rui no onna]

Some of you guys just analyze things too much. I turn on my iPad and it works for what I need to do and that's just fine. I don't go and analyze anything that may come out just around the corner or many months from now. That's technology as I figure and I don't get bent out of shape for it.

When I'm spending ~$1400 on something (after taxes), I kinda like to make sure I'm gonna be happy/satisfied with said purchase. That usually involves doing a cost-benefit analysis.

 

[gobikerider]

When I'm spending ~$1400 on something (after taxes), I kinda like to make sure I'm gonna be happy/satisfied with said purchase. That usually involves doing a cost-benefit analysis.

Preach it

 

[neutrino23]

It's not so easy to compare the iPhone CPU vs the X version used in an iPad. The iPhone may sprint ahead quickly for a short time. The X version has better heat sinks that should allow sustained performance. We don't know what Apple has done to the X versions, besides adding RAM, for the iPads.