Peak Peek - Apple officially announces new SE, new iPad Air, M1 Ultra, Mac Studio and Studio Display

[leadeater]

12 minutes ago, igormp said:

N6 is more of an updated N7, so a bit far from N5 in the end.

N6 is a N7 process improvement. 18% density with no claimed power or performance gains, due to the node. N6 also allows for more EUV usage.

[Dracarris]

2 hours ago, AluminiumTech said:

I'm aware. I just don't see that being an acceptable trade off, even for laptops. The ship has sailed on the phone and mobile tablet market but I disagree, it has not on laptop and it cannot be allowed to be normalized imo.

What Apple is doing here is not exactly your regular DDR RAM soldered to a laptop mainboard with no advantage other than slightly reduced height and the removal of a mechanical connector.

 

The RAM is simply part of the package of the SoC, it's not "soldered down to the computer". As others have already pointed out, this brings massive advantages. The most impressive one is imho that you can use a large part of that unified memory as VRAM while GPUs with over 24GB of VRAM alone cost more than a specced out M1 Ultra Studio, not to mention the access latency advantages, effectively completely doing away with DMA transfers. In no way will you get this architecture into a modular design (CPU, GPU, RAM replaceable).

 

As for anticonsumer, ewaste and so on: CPUs and DRAM are among the components that fail the least, they usually last for decades and only die long after they have become obsolete anyway.

[rikitikitavi]

9 hours ago, Kisai said:

The part on the left is the marketing, the part on the right is for nerds. Apple makes no distinction between why any part is better than the other because you don't get to choose. The Max, Pro and Ultra, are meaningful marketing-wise, but they don't really help understand why they are.

If Apple sold CPU's as more than a BYO option, then yes, you'd have something like this:

M1[P4E4G7M16]

M1[P4E4G8M16]

M1[P6E2G14M16] (Pro)

M1[P8E2G16M16] (Pro)

M1[P8E2G24M16] (Max)

M1[P8E2G32M16] (Max)

M1[P16E4G64M32](Ultra)

 

But Apple doesn't need to make that distinction because each BYO device only has like two options.

 

Then simplify it even more just to 1 option at each of 4 tiers - then no confusion at all why this 'Pro' is faster than that 'Pro' and no need to count cores.

[Video Beagle]

20 hours ago, FakeKGB said:

Yes...

...because those are for the regular consumer market. Y'know, the one that likes having things in fancy colors and want people to look at their things.

 

My point was, there was a specific call out to magic keyboard/mice that color match and compliment the new studio line. I didn't catch if they were new colors added to the line or not. And the studio line is being aimed at creatives...it's freakin' called Studio...and creative types do rather like creative workspaces.

 

19 hours ago, Dracarris said:

it was intentional. And Apple does this for a while now with their presentations. No matter what you think of it, IWD or not - it is a stark misrepresentation of reality. The percentage of female PoC or hijab-wearers in this industry is way lower.

So, the people shown in an Apple presentation represent the industry? Every tech company has an openly gay CEO?

 

[Video Beagle]

Oh, another thing I noticed that was...notable.... Apple made a specific call out to Sony with the bit about it's camera, which was shown on screen for some time, doing...something. (I didn't really pay attention to that and it was yesturday i saw it)....didn't seem to be an exclusive thing that particular camera alone did...but it was a specific mention.

 

Apple doesn't tend to make call outs to other companies..especially sometime competitors..for no reason. I wonder what that could signify.

[Lightwreather]

16 minutes ago, Video Beagle said:

 I wonder what that could signify.

/s

[LAwLz]

17 hours ago, leadeater said:

I did so for all of them, GB has a page/feature that can sort all the results and I chose the highest from those for all. Can't remember if I sorted by ST or MT.

Ah I see.

That's not a good idea to do with a processor like the 5900HS, because it varies greatly between power profiles.

If you just go by the highest score on GB, then you might be looking at a 6900HS using like 80 watts (yes, it does that high on some laptops by default). Meanwhile, the M1 Pro is pretty much always sitting at around 30 watts.

 

You might be comparing a 35 watt part (the 6900HS) running at 80 watts, vs a 35 watt part running at 35 watt (M1 Pro). You can't use that data and then pretend like the 6900HS used 35 watts in the test. It was most likely using way more power, and that's the reason why the benchmark you picked is significantly higher than the other 6900HS benchmarks.

The majority of benchmarks on Geekbench puts the 6900HS quite a bit below the M1 Pro, and that's probably because they ran it at 35 watts. And please note that the ~35 watts the M1 Pro uses also includes things like the storage controller, the RAM and a few other things, while the 35 watts for the Ryzen chip only includes the processor.

 

 

All proper apples to apples comparisons I have seen shows the M1 and M1 derivative chips to be significantly more power efficient than anything AMD has to offer, including their mobile offerings.

Tune both to the same low power TDP, isolate the CPU cores and how much power they use, and then run the same benchmark. The M1 will be significantly ahead.

[leadeater]

1 hour ago, LAwLz said:

That's not a good idea to do with a processor like the 5900HS, because it varies greatly between power profiles.

True but it's kind of their default thing on the page I was using, I don't remember actually seeing any lower ones. Quite possible it was exclusive to that laptop model, links off Google searches can take you to strange places heh

 

I.e. https://browser.geekbench.com/macs/macbook-pro-16-inch-2021-apple-m1-max

 

  

1 hour ago, LAwLz said:

If you just go by the highest score on GB, then you might be looking at a 6900HS using like 80 watts

The HX for sure, I don't think there are any HS that have sustained 80W configuration, the HX exists for that purpose.

 

The G14 is clearly higher power (Edit: confirmed 45W) configuration than the what was it MSI?

 

1 hour ago, LAwLz said:

M1 Pro is pretty much always sitting at around 30 watts.

Except when it's not...

 

34W CB23

36.9W 502 MT

40.9W 511 MT

43.9W 503 MT

 

Edit:

G14 power/clock graph, so 45W instead of 35W base spec.

 

[leadeater]

1 hour ago, LAwLz said:

All proper apples to apples comparisons I have seen shows the M1 and M1 derivative chips to be significantly more power efficient than anything AMD has to offer, including their mobile offerings.

Well I object to your "proper" because everything I linked is proper in the bounds of the test configuration used. I'm going to put aside your dismissal of HUB but it doesn't change much because you can get the same application benchmark data elsewhere so I'm not going to have a bias debate that doesn't even matter.

 

Something can be significantly ahead, or quite far ahead or w/e but that does not "spell the end for x86", if you catch my drift. And literally a large part of Apples lead is TSMC 5nm, like a very large part. That does not mean I am saying Firestorm is not a very good class leading arch either.

 

And PS all Ryzen Mobile are SoC too with more than just CPU cores so that "M1 has other things" point is moot and those are power gated very well when not being used anyway, for both of them.

 

 

Note things like Storage controllers, USB, Media Engines, Security Processors, GPU etc

 

To me it sounds like you've got a little bit of a Apple slant and ignoring factors that are the same between the two and only treating one as an SoC when in fact both are. The biggest difference is on package memory. If you want your cake and to eat it then you have to acklowdge all the other cakes in the bakery 

 

Also it's hardly Apples to Apples and fair if you package power cap the 6900HS to 35W and do a GB5 run and yet the M1 Pro/Max could be using more than 35W depending on specific sub test. We'd have to figure out how much of that is RAM ofc though which is essentially impossible without Apple only tools. Comparing between these simply is not that easy or basic and I will always fall back to the point that it does not matter. This is not what makes the M1 SoCs so good, there's much wider factors at play that are equally or more important and contribute to that overall product solution.

[Deadpool2onBlu-Ray]

I am disappointed they didn't release the se 2020 for 199$ like all the rumors said. My parents are rocking ORIGINAL SE's from like 2013 or whatever that was. 

[LAwLz]

31 minutes ago, leadeater said:

Except when it's not...

 

34W CB23

36.9W 502 MT

40.9W 511 MT

43.9W 503 MT

That's the power for the entire chip, not the CPU alone.

Also, not sure why you are quoting power consumptions for things like bwaves_r when we are talking about GeekBench.

 

When running GeekBench, the entire M1 Pro SoC uses 34 watts. That's including everything from RAM, the SSD controller, etc. 

If we subtract all of those things we are down to something like 30 watts.

 

It's very important that we make these distinctions and don't confuse things like power consumption when running one type of workload vs another type of workload. Otherwise we will just end up talking past each other.

 

 

17 minutes ago, leadeater said:

Well I object to your "proper" because everything I linked is proper in the bounds of the test configuration used. I'm going to put aside your dismissal of HUB but it doesn't change much because you can get the same application benchmark data elsewhere so I'm not going to have a bias debate that doesn't even matter.

All I did was ask if you knew how he measured the power consumption, because it does not align with other tests. Questioning what appears to be flawed data is not "being biased" or "dismissing".

 

 

19 minutes ago, leadeater said:

Something can be significantly ahead, or quite far ahead or w/e but that does not "spell the end for x86", if you catch my drift.

Did I ever say it did? Or are you making a strawman argument?

x86 is far from dead, and it certainly won't be the M1 that kills it.

 

 

21 minutes ago, leadeater said:

And literally a large part of Apples lead is TSMC 5nm, like a very large part.

Have I ever said it isn't? Again, are you making a strawman argument?

Anyway, does it really matter? People don't really care why something is faster/better than some other product. What matters is how it performs.

Back when Intel was stuck on 14nm nobody went "yeah AMD might get better efficiency and performance, but Intel is still better because if they had a better process node they might have been more equal, so therefore you should buy Intel processors!", but those are the vibes I am getting from you right now.

 

24 minutes ago, leadeater said:

And PS all Ryzen Mobile are SoC too with more than just CPU cores so that "M1 has other things" point is moot and those are power gated very well when not being used anyway, for both of them.

Ryzen mobile is an SoC, but the things I have specifically pointed out are not part of the SoC for Ryzen Mobile, but they are in the M1.

Ryzen mobile processors do not have the RAM on-package, the M1 does. 

When someone measures let's say 35 watts for the M1 it usually includes the power consumption of the RAM. 

When someone measures let's say 35 watts for a Ryzen Mobile processor it usually does not include the power consumption of the RAM.

 

Same goes for certain other things like the SSD controller (RAM and SSD controller are the two things I have mentioned so far).

 

 

27 minutes ago, leadeater said:

Note things like Storage controllers, USB, Media Engines, Security Processors, GPU etc

Are you pretending to not know the difference between an NVMe/SATA interface, and an SSD controller? Or do you genuinely not know the difference? I thought you worked with storage...

 

30 minutes ago, leadeater said:

To me it sounds like you've got a little bit of a Apple slant and ignoring factors that are the same between the two and only treating one as an SoC when in fact both are. The biggest difference is on package memory.

Ah yes... I am the one "ignoring factors".

Thanks for confirming exactly what I said by the way. That there is a difference such as the on-package memory. 

 

You sure love making strawman arguments. I am not sure if it's intentionally, if I am just bad at explaining my arguments, or if you are bad at understanding what I write. Maybe it's a mixture of both.

[leadeater]

32 minutes ago, LAwLz said:

Did I ever say it did? Or are you making a strawman argument?

Nope but remember where this conversation started, because that's actually important. Know the journey otherwise the path travelled gets lost.

 

32 minutes ago, LAwLz said:

Are you pretending to not know the difference between an NVMe/SATA interface, and an SSD controller? Or do you genuinely not know the difference? I thought you worked with storage...

No, but are you pretending a data interface also does not use power? The fact that their is an NVMe controller on an M.2 SSD or soldered SSD doesn't mean that data interface on the SoC for it does not use power.

 

32 minutes ago, LAwLz said:

Same goes for certain other things like the SSD controller

If it's not being actively used then it's mW so irrelevant. It was never worth bringing up in the first place unless you are running GB5 and a storage benchmark at the same time. The data interfaces on Ryzen also use mW when doing nothing also so also not relevant either. Hence my comment about power gating, none of these matter unless they are being used.

 

P.S. Just stop taking comments and conversations so personally, not everything is about you and what you have said specifically  If you want to join in on a conversation that started from hype train BS then please keep the origin in mind, because I may well be comment on that not you. 

[HRD]

40 minutes ago, leadeater said:

Well I object to your "proper" because everything I liked I proper in the bounds of the test configuration used. I'm going to put aside your dismissal of HUB but it doesn't change much because you can get the same application benchmark data else so I'm not going to have a bias debate that doesn't even matter.

 

Something can be significantly ahead, or quite far ahead or w/e but that does not "spell the end for x86", if you catch my drift. And literally a large part of Apples lead is TSMC 5nm, like a very large part. That does not mean I am saying Firestorm is not a very good class leading arch either.

 

And PS all Ryzen Mobile are SoC too with more than just CPU cores so that "M1 has other things" point is moot and those are power gated very well when not being used anyway, for both of them.

M1 has many extra features like the neural engine and memory which are not parts of the Ryzen mobile processors.

 

 

 

You always compare the advertised TDP not real-world use cases.

 Here the m1 pro has the same performance as the Ryzen 9 5900HX.

However, the r9 is using 271% more wattage. Ok, Apple has the 5nm advantage(Apple making better deals with TSMC is a bad job from AMD who could secure a better node) but let’s say it is fair to remove the improvements from the better node. According to TSMC 7 to 5 nm gives you a 30% efficiency advantage while offering the same performance. 

Let’s calculate it 26w*0.3= 7.8watt

26-7.8=18.2watt

Apple is still offering 160% better efficiency which is huge.

 

that's not to mention apple’s excellent idle power draw and the stand-by time.

this is not even apple’s latest architecture the A15 is already out and the A16 would most likely launch in September. By the time AMD&intel catch up to the m1 apple would be using the m3 or 4. 

Another example is Qualcomm which also makes ARM processors. Qualcomm is using a newer node but still, apple’s A15 chip offers better performance and uses lower power all while being on an older node!!!!!! Actually, apple chip is 6 core vs Qualcomm’s 8 and yet apple wins even in multicores performance.

[leadeater]

33 minutes ago, HRD said:

You always compare the advertised TDP not real-world use cases.

I've compared plenty, do you just intentionally not look at any of the real world application benchmarks I've posted?

 

33 minutes ago, HRD said:

Let’s calculate it 26w*0.3= 7.8watt

26-7.8=18.2watt

Apple is still offering 160% better efficiency which is huge.

Correct for single thread. Not for multi thread. However you could limit single thread boost power on Ryzen Mobile too, obviously performance will drop but since power and performance scaling are not linear that efficiency difference will be significantly less. Such is the nature of the differences between ARM and x86 right now, one supports clock boosting and the other does not, to the detriment of power efficiency of course.

 

If you are basically going to be complaining about cherry picking maybe don't also cherry pick too.

 

Not that this matters, I literally made this point but it seems you aren't reading everything anyway 

 

33 minutes ago, HRD said:

that's not to mention apple’s excellent idle power draw and the stand-by time.

Well fortunately that was already a thing when they were using Intel CPUs. Apple's/MacOS's idle and sleep/hibernate power is quite superior to that or Windows and PC laptops.

 

33 minutes ago, HRD said:

Apple making better deals with TSMC is a bad job from AMD who could secure a better node) but let’s say it is fair to remove the improvements from the better node.

Please for the love of everything do not call having more money to throw around as doing a "better job". Money does not make a better job, money gives an advantage and ability to do or get a better job done. Having less resources does not mean you have or are doing a "bad job".

 

[Paul Thexton]

Is it just me getting serious déjà vu from when Apple dropped the original M1 chip while reading all this? At some point people are going to start talking about (and fundamentally misunderstanding) HT again.

[Spindel]

1 minute ago, Paul Thexton said:

Is it just me getting serious déjà vu from when Apple dropped the original M1 chip while reading all this? At some point people are going to start talking about (and fundamentally misunderstanding) HT again.

To quote myself from this thread:

 

 

Quote

Man this is so fun.  

 

Currently exactly the same thing is happening with this GB, on PC-centric internet forums, that happened when OG M1 was released. 

 

A lot of denial, a lot of excuses. 

 

Man give it to Apple that they once again made tech forum debates interesting. 

 

[Paul Thexton]

Sorry @Spindel, I somehow missed you making the same point

[Commodus]

1 hour ago, Ryan829 said:

I am disappointed they didn't release the se 2020 for 199$ like all the rumors said. My parents are rocking ORIGINAL SE's from like 2013 or whatever that was. 

I didn't see any rumors that Apple would keep the 2020 model around for $199. I saw Bloomberg's Mark Gurman say he would like Apple to do that, but that's not the same as leaking it. (The original iPhone SE is from 2016.)

 

As it stands... $429 is definitely more expensive, but still a solid deal if your folks are willing to spend that much. Although I'd probably consider the slight price bump to 128GB to avoid future headaches.

[Deadpool2onBlu-Ray]

14 minutes ago, Commodus said:

I didn't see any rumors that Apple would keep the 2020 model around for $199. I saw Bloomberg's Mark Gurman say he would like Apple to do that, but that's not the same as leaking it. (The original iPhone SE is from 2016.)

 

As it stands... $429 is definitely more expensive, but still a solid deal if your folks are willing to spend that much. Although I'd probably consider the slight price bump to 128GB to avoid future headaches.

Oh. 2016? I was thinking 2013 because it has that same body style as the 5s released in 2013. 

 

The problem, in my opinion, is $429 is far too close compared to used prices for older/better devices. I got my current phone (11 pro max, 256gb) refurbished with 1 year warranty for $529. That phone has the new full screen form factor, OLED Screen, about twice the battery life. etc, all for $100 more. I know a lot of people are afriad of used, but that is the route I'd recommend to most people. The new SE is a decent deal though, I just wish they would use like the Iphone XR's body or something instead of the aging iphone 8 body. 

[Obioban]

6 minutes ago, Ryan829 said:

Oh. 2016? I was thinking 2013 because it has that same body style as the 5s released in 2013. 

 

The problem, in my opinion, is $429 is far too close compared to used prices for older/better devices. I got my current phone (11 pro max, 256gb) refurbished with 1 year warranty for $529. That phone has the new full screen form factor, OLED Screen, about twice the battery life. etc, all for $100 more. I know a lot of people are afriad of used, but that is the route I'd recommend to most people. The new SE is a decent deal though, I just wish they would use like the Iphone XR's body or something instead of the aging iphone 8 body. 

The body is one of the primary reasons for the SE-- old people can't learn how to use a new interface (lose the home button).

 

I very much doubt the iPhone 11 has twice the battery life of the SE3. The A15 was a huge improvement in battery life-- iPhone 13 mini has better battery life than iPhone 12 non mini!

[Commodus]

7 minutes ago, Ryan829 said:

Oh. 2016? I was thinking 2013 because it has that same body style as the 5s released in 2013. 

 

The problem, in my opinion, is $429 is far too close compared to used prices for older/better devices. I got my current phone (11 pro max, 256gb) refurbished with 1 year warranty for $529. That phone has the new full screen form factor, OLED Screen, about twice the battery life. etc, all for $100 more. I know a lot of people are afriad of used, but that is the route I'd recommend to most people. The new SE is a decent deal though, I just wish they would use like the Iphone XR's body or something instead of the aging iphone 8 body. 

That's true, and it's worthwhile if you can track down a good refurb. The iPhone SE is mainly appealing if you insist on the latest chip, don't like the thought of a used phone or want something a bit more compact.

 

My hunch is that this is the last iPhone to use the old form factor. The next model (2024? Hopefully sooner) will finally leap into a modern design.

[Deadpool2onBlu-Ray]

Just now, Commodus said:

That's true, and it's worthwhile if you can track down a good refurb. The iPhone SE is mainly appealing if you insist on the latest chip, don't like the thought of a used phone or want something a bit more compact.

 

My hunch is that this is the last iPhone to use the old form factor. The next model (2024? Hopefully sooner) will finally leap into a modern design.

Yeah. I imagine by 2024, they will use the iphone 11 "dated" design with the notch, and the flagships will have the pill shaped hole or hopefully, nothing on the screen at all. 

[Lightwreather]

29 minutes ago, Ryan829 said:

nothing on the screen at all. 

I sure hope not. I wouldn't want display capabilities to be a paid-for upgrade.

/s

[LAwLz]

4 hours ago, leadeater said:

-snip-

Lol, the nvme interface on the Ryzen 6000 does not use 3 watts of power. You can't just subtracting 3 watts of power from the 6900HS and say it uses 32 watts of power to make it seem more efficient than it really is. 

[hishnash]

6 hours ago, leadeater said:

Also it's hardly Apples to Apples and fair if you package power cap the 6900HS to 35W and do a GB5 run and yet the M1 Pro/Max could be using more than 35W depending on specific sub test. We'd have to figure out how much of that is RAM ofc though which is essentially impossible without Apple only tools. Comparing between these simply is not that easy or basic and I will always fall back to the point that it does not matter. This is not what makes the M1 SoCs so good, there's much wider factors at play that are equally or more important and contribute to that overall product solution.

There is a comparison for desktop systems you measure the aggregate power drawn at the wall for the entire test (number of Wh), since there are so many different choices that can impact power draw, its not just about the CPU cores compute, that is what matters when it comes to efficacy, different vendors might opt to target efficacy in different places but in the end the power draw from the wall is what matters.  (using the aggregate power draw also accounts for systems that solve the problem faster but use more power per second vs systems that are slow and use less power in the evaluation of efficacy that is the metric you need)... eg how many Wh does it need to do a given standard task.

There are other metrics, like how fast does it complete the task, how load does it get while running the task etc this is a multi dimensional optimisation target.

And I think we can all agree that there is no zen based system that can even come close to this on power usage.