New iPad Pro Has Comparable Performance to 2018 15“ MacBook Pro in Benchmarks(macrumors.com)
macrumors.com
New iPad Pro Has Comparable Performance to 2018 15“ MacBook Pro in Benchmarks
https://www.macrumors.com/2018/11/01/2018-ipad-pro-benchmarks-geekbench/
69 comments
i also feeling something fishy here. i remember geekbench reporting different memory latencies than AIDA64 a while ago.
maybe apple took geekbench and optimized their silicon for the test load(crypto)? or maybe geekbench for arm and geekbench for x86_64 may be doing the different tasks?
geekbench web site have the list of benchmarks( http://support.primatelabs.com/kb/geekbench/interpreting-gee... ), but is light on testing methodology and there is no way to verify what/how benchmarks are performed.
Unrelated: Linus Torvalds trashing gb3 in mailing list, a long thread https://www.realworldtech.com/forum/?threadid=136526&curpost...
Edit: added the link to gb4 benchmarks list
maybe apple took geekbench and optimized their silicon for the test load(crypto)? or maybe geekbench for arm and geekbench for x86_64 may be doing the different tasks?
geekbench web site have the list of benchmarks( http://support.primatelabs.com/kb/geekbench/interpreting-gee... ), but is light on testing methodology and there is no way to verify what/how benchmarks are performed.
Unrelated: Linus Torvalds trashing gb3 in mailing list, a long thread https://www.realworldtech.com/forum/?threadid=136526&curpost...
Edit: added the link to gb4 benchmarks list
Even if the absolute value of the Geekbench score isn’t useful, the fact that the delta between the MBP Geekbench score and the iPad Geekbench score has narrowed from huge to zero would still be meaningful.
The delta between an iPad and a $5000 server is zero, but the iPad is $700 and draws only a dozen watts! That's why Google is throwing out all their servers and filling their datacenters with iPads.
I know I'm being flippant, but I'm sick and tired of talking about Geekbench.
I know I'm being flippant, but I'm sick and tired of talking about Geekbench.
What's a better metric?
Do you believe that the 24-core $1500 server processor can beat the iPad in single to minimum threaded workload?
In practice I think the iPad would throttle. But let's assume it has adequate cooling.
The 7401 is clocked at two GHz. It's by no means a single-threaded monster. So maybe.
The 7401 is clocked at two GHz. It's by no means a single-threaded monster. So maybe.
Benchmarks often don't tell the whole story. I remember a similar headline from a couple years ago lauding the Nexus 9 for having comparable benchmarks to a Macbook Pro:
https://9to5google.com/2014/10/16/nexus-9-benchmark-is-compa...
The device still got mixed reviews.[0][1][2]
[0]https://arstechnica.com/gadgets/2014/11/nexus-9-review-a-pre...
[1]https://www.androidpolice.com/2014/11/03/nexus-9-review-a-li...
[2]https://www.engadget.com/2014/11/03/google-nexus-9-review/
https://9to5google.com/2014/10/16/nexus-9-benchmark-is-compa...
The device still got mixed reviews.[0][1][2]
[0]https://arstechnica.com/gadgets/2014/11/nexus-9-review-a-pre...
[1]https://www.androidpolice.com/2014/11/03/nexus-9-review-a-li...
[2]https://www.engadget.com/2014/11/03/google-nexus-9-review/
A major difference is in sustained performance, not peak performance. Mobile CPUs often have cores these days that are comparable to desktop/server CPU cores but they lack the cooling to sustain peak performance for very long. They're thermally throttled very rapidly.
That being said you could take such a CPU and given it desktop-like power and cooling and it could perform like a desktop chip.
That being said you could take such a CPU and given it desktop-like power and cooling and it could perform like a desktop chip.
Yeah it would be interesting to see how the apple tv performs compared to an ipad over sustained loads. It has wired power and I think the 4K even has a cooling fan.
1. That was using Geekbench 3, which according to Linus, "showing partiality for ARM64 devices and being a bad performance measure of x86-based system." Geekbench 4 released in 2016 is much better.
2. The "2012" Mac Pro stated is actually using 2008 CPU, with the lowest configuration.
2. The "2012" Mac Pro stated is actually using 2008 CPU, with the lowest configuration.
I honestly think they are going in the right direction and are purposely taking their time to perfect each development before over complicating iOS. They have an opportunity with iOS to start from scratch and develop new ways of using and interacting with a computer and have decided not to rush it.
My iPad Pro wish list:
- A “Proper” file system
- Ability to run my own code written on it (and leave it running in background, e.g. a development web server)
- Terminal access
- A detachable keyboard with touchpad (far better ergonomics than a touch screen when typing at a desk)
- Support for a mouse (less important)
Steve Jobs said they would never do a stylus, now they have one. Just wait to see what they do next.
My iPad Pro wish list:
- A “Proper” file system
- Ability to run my own code written on it (and leave it running in background, e.g. a development web server)
- Terminal access
- A detachable keyboard with touchpad (far better ergonomics than a touch screen when typing at a desk)
- Support for a mouse (less important)
Steve Jobs said they would never do a stylus, now they have one. Just wait to see what they do next.
With the support of external 4K Monitor from USB-C, all I wanted now is Wireless Keyboard and Mouse.
Give Xcode, and lots / most dev can switch over to programming on iPad.
We can start having / playing PUBG, C&C, Online Games with Keyboard and Mouse.
And hopefully some day they will have at least 8GB Memory on iPad Pro, or even 16G.
The one thing that worries me though is the insane amount of money required for storage. And I don't want to rely on iCloud all the time. I just wish they do Time Capsule Type Network Storage devices again.
Give Xcode, and lots / most dev can switch over to programming on iPad.
We can start having / playing PUBG, C&C, Online Games with Keyboard and Mouse.
And hopefully some day they will have at least 8GB Memory on iPad Pro, or even 16G.
The one thing that worries me though is the insane amount of money required for storage. And I don't want to rely on iCloud all the time. I just wish they do Time Capsule Type Network Storage devices again.
It seems like half (all) of those could be solved with a VM. On paper, the iPad has the storage (1TB!) and horsepower now. I wonder how difficult it would be to implement a hypervisor on iOS.
Don't all those things on your wish list just make it like every existing OS/machine?
Yeah, sounds great doesn't it? That great hardware with useful and flexible OS!
Have you considered buying a small laptop???
I use my iPad purely for consuming, and the laptop for all that you mention.
I do agree that there is a gap between the two though, an Apple product similar to the Microsoft Surface (pro). But I think this is a separate product to the iPad.
I use my iPad purely for consuming, and the laptop for all that you mention.
I do agree that there is a gap between the two though, an Apple product similar to the Microsoft Surface (pro). But I think this is a separate product to the iPad.
No, because they don’t have that brilliant touch screen.
Wouldn't it be easier to add the touch screen to an existing form factor and OS than to add all those things to an iPad?
No, because I (and I expect others) want an iPad that works as an iPad in every way and doesn't have a permanently attached keyboard, but has these features for when you want them.
So if Apple relented and added a touchscreen to their laptops you'd be all set?
Kind of related: Although not a real mouse, if you are on the latest iOS you can hold down the spacebar while using the keyboard and get a mousepad/cursor which is convenient for selecting text.
True, and that works well when using the on screen keyboard but not when you sing a detachable one.
My wish list is short:
- Ability to share my iOS software with my own customers without Apple deciding if it's politically correct or benefits Apples profit sheet
I'm not holding my breath on that one
- Ability to share my iOS software with my own customers without Apple deciding if it's politically correct or benefits Apples profit sheet
I'm not holding my breath on that one
I think there's a large number of people who feel they are gladly paying for this restriction in the sense of "if I buy this I'm less likely to get "hacked".
(but TBF I only share my software with my nerd friends who recompile it themselves so I admit I am not suffering at the coal face on this one as some developers are.)
(but TBF I only share my software with my nerd friends who recompile it themselves so I admit I am not suffering at the coal face on this one as some developers are.)
What is the difference between your ideal iPad Pro and a normal MBP?
Everything an iPad can do with that amazing touch screen.
>> A detachable keyboard with touchpad (far better ergonomics than a touch screen when typing at a desk)
I’d rather Apple sufficiently improve the software keyboard so we don’t need a detachable hardware keyboard. There are keyboard cases for the iPhone but they are still niche because the iPhone keyboard has always been good enough.
I’d rather Apple sufficiently improve the software keyboard so we don’t need a detachable hardware keyboard. There are keyboard cases for the iPhone but they are still niche because the iPhone keyboard has always been good enough.
A train also has a more horsepower than an SUV, but it can only go back and forth along tracks.
And sometimes that's all you need.
You can sit back and enjoy the ride, not care for traffic, carry much larger load more efficiently, and so on...
You can sit back and enjoy the ride, not care for traffic, carry much larger load more efficiently, and so on...
... but to go really deep, you have to realize that horsepower is just a marketing term that is mostly meaningless. It was used to sell vehicles to Americans all through the 80s and 90s until now.
It's torque you actually want - i.e. turning force.
Horsepower is just torque X rpm.
The best example of how meaningless horsepower is was the original Honda S2000, at the time, the world's highest specific-power engine (horsepower per liter of engine size). It was revvy as hell, and only developed power when really high in the rev range.. i.e. 6k-9k rpm, so many people hated driving it. You had to rev the crap out it to go anywhere.
A few years later, the next revision had a bigger engine, with less horsepower. But it accelerated faster and was way better to drive.
Why? The bigger engine had more torque.
The horsepower figure is mostly irrelevant to drive-ability and acceleration.
Interestingly, a steam locomotive has maximum torque at 0 rpm, a complete contradiction for an internal combustion engine.
It's torque you actually want - i.e. turning force.
Horsepower is just torque X rpm.
The best example of how meaningless horsepower is was the original Honda S2000, at the time, the world's highest specific-power engine (horsepower per liter of engine size). It was revvy as hell, and only developed power when really high in the rev range.. i.e. 6k-9k rpm, so many people hated driving it. You had to rev the crap out it to go anywhere.
A few years later, the next revision had a bigger engine, with less horsepower. But it accelerated faster and was way better to drive.
Why? The bigger engine had more torque.
The horsepower figure is mostly irrelevant to drive-ability and acceleration.
Interestingly, a steam locomotive has maximum torque at 0 rpm, a complete contradiction for an internal combustion engine.
"you have to realize that horsepower is just a marketing term that is mostly meaningless"
No it's not. Same torque on a single cylinder engine going up to 5000 rpm is a drastically different thing than on a 8 cylinders engine going up to 18000 rpm, and this makes your reasoning pointless. Race engines usually deliver most of torque at higher RPMs, and that's just how physics work. You might enjoy engines that yield torque early, but that's just a subjective personal matter that does not influence definitions and value of torque and hp as indexes of performance.
No it's not. Same torque on a single cylinder engine going up to 5000 rpm is a drastically different thing than on a 8 cylinders engine going up to 18000 rpm, and this makes your reasoning pointless. Race engines usually deliver most of torque at higher RPMs, and that's just how physics work. You might enjoy engines that yield torque early, but that's just a subjective personal matter that does not influence definitions and value of torque and hp as indexes of performance.
> No it's not. Same torque on a single cylinder engine going up to 5000 rpm is a drastically different thing than on a 8 cylinders engine going up to 18000 rpm, and this makes your reasoning pointless.
You're using examples that barely exist in the real world, so your reasoning is pointless.
a) Show me a single cylinder engine that makes the same torque from idle to 5000rpm. No such thing exists.
b) The only 8 cylinder engine ever made that revved to 18,000rpm was for Formula 1, and they have not used 8 cylinders for many years. I would propose there is not a single 8 cylinder engine being made in the world today that can rev to 18,000rpm.
> Race engines usually deliver most of torque at higher RPMs, and that's just how physics work
Can you link to a torque curve that shows that? I've never seen an engine where torque goes UP with rpm. It always falls off as volumetric (in)efficiency fails to get the air in and out of the cylinder.
You're using examples that barely exist in the real world, so your reasoning is pointless.
a) Show me a single cylinder engine that makes the same torque from idle to 5000rpm. No such thing exists.
b) The only 8 cylinder engine ever made that revved to 18,000rpm was for Formula 1, and they have not used 8 cylinders for many years. I would propose there is not a single 8 cylinder engine being made in the world today that can rev to 18,000rpm.
> Race engines usually deliver most of torque at higher RPMs, and that's just how physics work
Can you link to a torque curve that shows that? I've never seen an engine where torque goes UP with rpm. It always falls off as volumetric (in)efficiency fails to get the air in and out of the cylinder.
"You're using examples that barely exist in the real world"
No I'm not. I have a 110hp motorcycle with 4 cylinders (CBR 600RR), and a 30hp motorcycle with one cylinder (Royal Enfield Continental GT). They behave mostly the same, with comparable performance and torque, up to 6000rpm. Then the 1 cylinder reaches to top rpm. The 4 cylinders one starts to deliver the extra 3/4 of torque from 7000 rpm. I enjoy riding both, in different moods.
No I'm not. I have a 110hp motorcycle with 4 cylinders (CBR 600RR), and a 30hp motorcycle with one cylinder (Royal Enfield Continental GT). They behave mostly the same, with comparable performance and torque, up to 6000rpm. Then the 1 cylinder reaches to top rpm. The 4 cylinders one starts to deliver the extra 3/4 of torque from 7000 rpm. I enjoy riding both, in different moods.
What's the displacement of the Royal Enfield?
> They behave mostly the same, with comparable performance and torque, up to 6000rpm
...so , you are confirming what I said. A bike with 30hp performs similarly to a bike with 110hp. - UNTIL the Enfield hits it's rev limit, at which point you can't really compare them because one physically can't do what the other does.
> They behave mostly the same, with comparable performance and torque, up to 6000rpm
...so , you are confirming what I said. A bike with 30hp performs similarly to a bike with 110hp. - UNTIL the Enfield hits it's rev limit, at which point you can't really compare them because one physically can't do what the other does.
What's the displacement of the Royal Enfield?
535cc vs 600cc of the CBR.
"...so , you are confirming what I said. A bike with 30hp performs similarly to a bike with 110hp."
Look bro, I don't want to argue indefinitely. They really do not perform similarly. My point is that HP is not a perfect index of performance, but being it TxRVM, it includes rough information about power snapshots (torque) and power range (RVM), and it is therefore more informative, and therefore more appropriate to describe engine performance that torque only. If you are arguing that before buying a vehicle you should take a look to its power delivery curve first, yep, that's what's more informative than torque, and hp. In absence of the curve, and being in need of indexing with a single number, hp (number) is more informative than torque (number).
535cc vs 600cc of the CBR.
"...so , you are confirming what I said. A bike with 30hp performs similarly to a bike with 110hp."
Look bro, I don't want to argue indefinitely. They really do not perform similarly. My point is that HP is not a perfect index of performance, but being it TxRVM, it includes rough information about power snapshots (torque) and power range (RVM), and it is therefore more informative, and therefore more appropriate to describe engine performance that torque only. If you are arguing that before buying a vehicle you should take a look to its power delivery curve first, yep, that's what's more informative than torque, and hp. In absence of the curve, and being in need of indexing with a single number, hp (number) is more informative than torque (number).
Ever been on the Eurostar? The experience is spectacularly good when compared to driving or flying...
I have been; for me flying wins. Compared to driving the Eurostar is indeed far better on those distances.
At this point it's a good bet the next plain 'MacBook' will be an A series processor.
We're looking at the leading edge of an eventual transition to Apple's full hardware stack using its own IP top to bottom, though I expect a long Intel tail for 'Pro' users.
We're looking at the leading edge of an eventual transition to Apple's full hardware stack using its own IP top to bottom, though I expect a long Intel tail for 'Pro' users.
I wonder how long it'll take before `brew install whatever` works reasonably well on ARM chips.
It's no harder than having Debian for x86, x64, arm, arm64, etc.
Intel is toast.
Why? Apple is just a small buyer of Intel products compared to the server, workstation and PC market in general.
Wake me up when I can run VirtualBox on these fancy iPad Pros and be able to run 1-2 virtual machines for development purposes.
id love to see a "what percent of devices are sold to developers" statistic. youre not their target market.
you can already run VMs on azure or aws and access them with rdp.
you can already run VMs on azure or aws and access them with rdp.
With more and more transition to cloud platforms this might be a short sleep.
Though I still prefer a local environment for quick iterative development I often can't justify the time to spin up/manage a VM due to how quick I could otherwise do cloud native development in a browser.
Though I still prefer a local environment for quick iterative development I often can't justify the time to spin up/manage a VM due to how quick I could otherwise do cloud native development in a browser.
How about when customers with niche use cases like that are sufficient to sustain Intel at its current/similar size?
Not saying I necessarily agree with the top-level comment, but that is hardly a convincing argument against it.
Not saying I necessarily agree with the top-level comment, but that is hardly a convincing argument against it.
You may never be able to run VirtualBox on an iPad Pro but there is a good chance VirtualBox will run on an Apple chip in the near future.
You would dare to run arbitrary code not throughly checked for appropriate content on iOS?!
An IPAD can't even support a mouse.
A ARM Macbook will, soon. That is what I'm reading from this story.
If Apple is getting this kind of performance out of their CPU/GPUs, especially the "Xbox One X level performance" they mentioned about their iPad Pro, curious if they'll turn Apple TV into a gaming console.
If only you could plug an Android phone in desktop mode (Samsung's Dex or Huawei's projection) to power such a perfect mobile touchscreen and keyboard.
Android chips lag behind, but they're good enough for occasional development.
Android chips lag behind, but they're good enough for occasional development.
Surely this has to fuel the flames on the "2019 MacPro will be ARM based" fire.
At this point, it seems Apple could pull it off if they wanted to, while thumping Intel in the process.
At this point, it seems Apple could pull it off if they wanted to, while thumping Intel in the process.
I wish the iPad had a useful OS rather than just a magnified iPhone. It should be the apple equivalent of the surface.
Apple will keep iterating it until it's almost as good as a real OS but they get to control the walled garden and collect the 30% tax from the entire ecosystem inside it. They'll get there eventually, it's a very profitable long term strategy. They have absolutely no interest in encouraging adoption of a free and open internet where they don't get to collect their rent.
Right. The fact that iOS is so crippled compared to OSX is putting me off buying one.
CPU architecture I assume is the reason these things cannot run OSX.
CPU architecture I assume is the reason these things cannot run OSX.
CPU arch. would be the least of the problems. They can do roughly the same trick they did with the PowerPC -> Intel switch years back.
The problem is that you cannot just put a keyboard/mouse centric on a touch device. The entire UX will crumble down. Same story the other way around.
Small things like buttons and textfields need to be bigger and more spaced out. Entire panels need to be redesigned as the control density is lower. To get the same quality of UX you need to design for both interface methods.
iOS suffers the same when attaching a keyboard to an iPad. A lot of small things like workflow shortcuts and navigating the control space with just a keyboard just don't work (yet). Forcing you in either a touch-only mode or having to switch between keyboard and touch constantly.
The problem is that you cannot just put a keyboard/mouse centric on a touch device. The entire UX will crumble down. Same story the other way around.
Small things like buttons and textfields need to be bigger and more spaced out. Entire panels need to be redesigned as the control density is lower. To get the same quality of UX you need to design for both interface methods.
iOS suffers the same when attaching a keyboard to an iPad. A lot of small things like workflow shortcuts and navigating the control space with just a keyboard just don't work (yet). Forcing you in either a touch-only mode or having to switch between keyboard and touch constantly.
Nothing in the CPU architecture would prevent porting MacOS if they wanted to. However I suspect that Apple has a long-term plan to phase out MacOS and the Macintosh hardware line. They'll enhance iOS and iPad to make that platform meet the needs of content creators, not just content consumers. That would allow them to simplify their product catalog rather than maintaining two totally separate lines.
>The fact that iOS is so crippled compared to OSX is putting me off buying one.
That's its main allure.
If one just wants a computer to run arbitrary OS with desktop-like experience there are tons of them, it doesn't even have to be a tablet.
That's its main allure.
If one just wants a computer to run arbitrary OS with desktop-like experience there are tons of them, it doesn't even have to be a tablet.
I too see this as an advantage; sometimes I want to develop and tinker, sometimes I need to to get work done that does not require that freedom but does require: stability, focus, long battery life etc and for that the iPad pro with keyboard is perfect (in my opinion). I had a Surface Pro and for that ‘just work’ kind of experience it sucks in my experience; updates, crashes, battery life etc. It turns out 80% of my work can be done really well on the iPad Pro and I indeed find myself using it far more than my laptop these days. I think a combination has to screw up that experience as it has for all other more free devices besides maybe Android but there is no iPad quality hardware for Android.
> CPU architecture I assume is the reason these things cannot run OSX.
Not, the fact that it's taboo to control every single app binary and extract 30% tax on everything on macOS is what's making Apple not port OS X.
Not, the fact that it's taboo to control every single app binary and extract 30% tax on everything on macOS is what's making Apple not port OS X.
how does it compare in performance/watt?
Zzzzzzzz geekbench.
Intel's gross margins are high enough (especially on the higher end chips) that it makes sense for Apple to design their own chips for Macs. Rumors are pointing to 2020 for when it will first happen.
Macs could then be cheaper and better than the competition (or significantly more profitable).
Macs could then be cheaper and better than the competition (or significantly more profitable).
https://browser.geekbench.com/processors/2134
If you believe a tablet can beat a 24-core $1500 server processor in a highly threaded workload, you may be interested in my upcoming ICO!
Geekbench is only good as a punchline.