iPad Pro's M4 Chip Outperforms M3 by Up to 25%

[Zest28]

I'm not arguing explicitly arguing against an iPad Pro being able to "dual boot" or whatever - But, if there was more then niche appeal - wouldn't the x86 MS Surface Pro tables running desktop Windows made in roads, dented iPad sales, or taken the world by storm?

These people on MacRumors think they know better than the people working at Apple.

I had to install something on my Steam Deck OLED using the touch screen in Linux and it was a horrible nightmare. All the guides were saying do not dare to do this with the touch screen or the controls, and I fully understand why after that painful experience.

Using an OS that is designed for a keyboard and mouse for a touch screen device is a bad idea.

 

[m1mini]

Proud owner of an m3 pro mbp, it is still screamingly fast so can't even imagine how fast a m4 pro/max would be. can't wait for m5, m6 and onwards

 

[Tagbert]

Man i can’t believe people fall for this minor changes and what’s unbelievable is that people find excuses to buy this products and probably people that lives with a tight budget… smh

What would you consider to be a non-minor change?

It’s not like most people with an M2 iPad are going to just trash it and buy a new one just for the M4 chip upgrade. This is targeted at people with older iPads and people new to iPads. If someone is on a tight budget and they are trying to buy an iPad Pro, they need to reconsider their financial planning.

 

[thingstoponder]

• M4 - 3,695/14,550
• M1 - 2,272/8,208

fourth generation chip and less than 2x performance increase over the first gen chip? are my expectations to high? seems... meh?

Yes it’s not the 80s anymore.

Will an iPad user, pro or otherwise, ever make use of all that power?

Unless iPadOS 18 provides significantly more powerful features, it seems such a waste.

Don’t get me wrong, I’ve ordered an 11” M4 iPad Pro (256GB), especially for the OLED display, but the processor just seems like overkill given iPadOS’ limitations.

Everything you do benefits from a faster CPU. 5 year old devices are slower than they were 5 years ago because software (especially the web) gets more bloated.

 

[Rychiar]

Get me an m4 max Mac Studio and I’ll be happy

 

[MallardDuck]

The thing that amazes me is the single core performance. Makes me really think about upgrading my m1 max mbp when the new macs come out. It's just hard to justify given my current workloads.

It wouldn't surprise me if they're announced at WWDC. Otherwise knowing that chip exists is going to push off a lot of upgrades.

 

[magbarn]

Looking forward to seeing these chips in Macs. I was reading in some discussion threads that part of the CPU score bump could be related to the new ML accelerators (ARM's SME)? It seems like those are the tasks where performance score really increased, the other tasks were much more equivalent to M3 performance.

Per Tom's Hardware:

This is key right here. Definitely a YMMV moment. Not sure how much of the benchmarks that went up 200% in GB really equal in real world app improvement given the much smaller improvements in the rest of geekbench.

 

[John_san]

I just bought an M3 Max, because it's here now, and I know what it can do now rather than speculate about what something might do at some point in the future.

Doesn’t change the fact that M3 is already outdated, after only 6 months.
If you are happy with M3 here and now, happy for you.

 

[TheWraith]

You will be waiting for more WWDC. If you are expecting iPad OS to morph in to Mac OS, I doubt it will happen any time soon. It’s been going on for past 8-10 WWDC.

ding ding ding

 

[smulji]

I doubt it will happen any time soon

I doubt it will happen at all

 

[Daul]

“It’s going to impact Mac sales”
Meh. Every year Apple releases upgraded hardware. Savvy Apple consumers who care about having the latest hardware already know this and know to buy when a new version comes out rather than buying in the middle of a release cycle. Apple doesn’t have the same problem as Osborne did. They have been working with cyclical sales for many years and delayed purchases aren’t forgone purchases.

“My device is already obsolete!”
See above. You already knew this was going to happen within a year of the purchase. It doesn’t mean you need to buy the new thing. Either your current device meets your needs or it doesn’t.

 

[TigerNike23]

Nice upgrade for those on M1 (and older) Macs. Skip the M3. Wait for the M4.

Yeah, guaranteed they’ll talk about performance against M1 when they release this for the Mac.

Interesting to me is that we’re getting close to doubling the M1 numbers. Maybe the M5 will do it.

2X the performance every 5 years is a pretty good standard, in my opinion.

 

[MrGimper]

Doesn’t change the fact that M3 is already outdated, after only 6 months.
If you are happy with M3 here and now, happy for you.

M3 is ... Pro and Max aren't.

 

[crazy dave]

AMD isn’t. Look at their Zen processors. Granted, they don’t release a new one each year, however they get >10% IPC gains each time without fail. They are in fact due to repeat this with Zen 5 in Q3. I am not knocking Apple, the M series chips are incredible. I am just very curious what they plan to do next.

You could flip that on its head and argue that AMD have a lot of catching up to do with regards to IPC and Apple is catching up to AMD's clock speeds. After all M1 started off with 3.2 GHz and they're now at 4.4 GHz (and they did it without becoming power hungry too). Each design has its own areas where it can expand the fastest. Like AMD and Intel have hit a GHZ barrier, it is possible Apple has hit an IPC barrier but likewise that other designs will too - we don't know until others catch up. Note that Qualcomm who aquihired a lot of ex-Apple engineers from Nuvia didn't really do any better than the M2 Avalanche core and ARM's wide core didn't get better than it either. It could be that there is an intrinsic limit to IPC gains in "normal" code from going wider and everyone will hit that or there is some bottleneck that someone will solve and open it up again.

This is key right here. Definitely a YMMV moment. Not sure how much of the benchmarks that went up 200% in GB really equal in real world app improvement given the much smaller improvements in the rest of geekbench.

To be fair, most of the others still went up 10-20%.

 

[ric22]

Uh, hate to break this to you…

My point being there was no CPU differentiation on M1/2/3 before. It remains valid. GPU differences or higher spec Pro/Max chip CPU variations are a different thing. I was well aware.

 

[mashdots]

this is me being internet explorer-levels of slow, but I had no idea the A17 Pro had higher single core scores than the whole M2 series. Wild.

 

[Smittywerben]

• M4 - 3,695/14,550
• M1 - 2,272/8,208

fourth generation chip and less than 2x performance increase over the first gen chip? are my expectations to high? seems... meh?

The competition, AKA S9 Ultra is 2,069/5,408, so I would call that a clear win for Apple.

 

[gusping]

You could flip that on its head and argue that AMD have a lot of catching up to do with regards to IPC and Apple is catching up to AMD's clock speeds. After all M1 started off with 3.2 GHz and they're now at 4.4 GHz (and they did it without becoming power hungry too). Each design has its own areas where it can expand the fastest. Like AMD and Intel have hit a GHZ barrier, it is possible Apple has hit an IPC barrier but likewise that other designs will too - we don't know until others catch up. Note that Qualcomm who aquihired a lot of ex-Apple engineers from Nuvia didn't really do any better than the M2 Avalanche core and ARM's wide core didn't get better than it either. It could be that there is an intrinsic limit to IPC gains in "normal" code from going wider and everyone will hit that or there is some bottleneck that someone will solve and open it up again.


To be fair, most of the others still went up 10-20%.

Very valid points. It's just we've seen zero evidence of IPC uplifts (I think it's around <5% from M1 to M4), so it will be very interesting to see how Apple tackles it, especially are supposedly losing a lot of its chip talent.

 

[gusping]

Not sure where Apple will go next (unless losing a lot of silicon designers to Nuvia did hurt them a little) but this post is interesting

https://twitter.com/x/status/1788596920627118248

Yes, the last part is particularly interesting/worrying, whatever you want to call it.

 

[MrGimper]

You could flip that on its head and argue that AMD have a lot of catching up to do with regards to IPC and Apple is catching up to AMD's clock speeds. After all M1 started off with 3.2 GHz and they're now at 4.4 GHz (and they did it without becoming power hungry too). Each design has its own areas where it can expand the fastest. Like AMD and Intel have hit a GHZ barrier, it is possible Apple has hit an IPC barrier but likewise that other designs will too - we don't know until others catch up. Note that Qualcomm who aquihired a lot of ex-Apple engineers from Nuvia didn't really do any better than the M2 Avalanche core and ARM's wide core didn't get better than it either. It could be that there is an intrinsic limit to IPC gains in "normal" code from going wider and everyone will hit that or there is some bottleneck that someone will solve and open it up again.


To be fair, most of the others still went up 10-20%.

With a ~10% clock increase

 

[ric22]

Some people hate when computers get faster. It’s very odd.

Strangely a different set of people to those that hate when computers get "too much" RAM or "too much" storage. 😂🤦‍♂️

 

[DrWojtek]

Can’t believe there are people out there using M3 laptops in public, they’re garbage now. Should be thrown in that hydraulic M4 press IMO.


Lol.

 

[masotime]

@MacRumors/@jclo, friendly suggestion- your list of benchmark data is fairly hard to appreciate. A horizontal bar graph in a more logical order makes more sense (why is M1 right in the middle of the data, M4 at the top, M3 at the bottom?, A17 Pro second?)

I didn't verify if the chart was accurate but it seems mostly legit.

 

[Ursadorable]

Does that mean the MacMini will skip the M3 and go straight to the M4?

 

[NewOldStock]

How does this work when the end-user can`t get their hands on them yet?
How can you test something not in the wild?