M1X worst case scenario

[casperes1996]

Do they have PL2 on mobile chips? I was shocked the first time I saw those thing where TDP doesn't seem to mean anything at all.

Yes. However, what the different PL levels means and when they trigger is kinda defined by the logic-/motherboard so you can't really rely on it to mean one thing either. Intel has a guideline specification for how long it should turbo for example - But a lot of devices just ignore it and do their own thing

 

[leman]

Did you guys notice how vague the A15 presentation was? I take it as a further evidence that the new prosumer Macs will be based on similar technology. They don’t want to spill the beans until the Mac event.

 

[senttoschool]

Did you guys notice how vague the A15 presentation was? I take it as a further evidence that the new prosumer Macs will be based on similar technology. They don’t want to spill the beans until the Mac event.

Or that the year and year gains are nothing to write home about which is why they're comparing the performance to the competition instead of the A14.

 

[pshufd]

Or that the year and year gains are nothing to write home about which is why they're comparing the performance to the competition instead of the A14.

Was there anything in the iPad mini presentation on performance?

 

[Andropov]

Did you guys notice how vague the A15 presentation was? I take it as a further evidence that the new prosumer Macs will be based on similar technology. They don’t want to spill the beans until the Mac event.

I wouldn't be too optimist. Comparing the A15 against 'competition' instead of the latest iPhone for the first time is a bad signal. I'm guessing they didn't get the 20% single core increase in performance this time. The benchmarks will be here anyway before the new Macs keynote anyway so no reason to be vague.

 

[CWallace]

Did you guys notice how vague the A15 presentation was? I take it as a further evidence that the new prosumer Macs will be based on similar technology. They don’t want to spill the beans until the Mac event.

Reminded me of the presentations for the earlier models (A13, A14). I imagine they don't feel the need to go into a technical deep dive during a marketing event.

I presume Apple will either release said deep dive later this week or they will let a site like ArsTechnica perform that.

I remain convinced the Macs to be announced next month will be on "M1X" and use the A14 as their foundation. I do not expect to see an A15-based "M2" until 1H 2022 at the earliest.

 

[CWallace]

Was there anything in the iPad mini presentation on performance?

They did compare it against the iPad Mini 5 so there were significant gains due to the Mini 6 being on A15.

 

[thenewperson]

Or that the year and year gains are nothing to write home about which is why they're comparing the performance to the competition instead of the A14.

They actually did that for the A14 as well last year.

 

[Andropov]

They actually did that for the A14 as well last year.

Did they? That would be good news. And would explain them taunting the competition at the start with that 'competitors are still playing catch-up with our CPUs from two years ago'.

 

[leman]

Or that the year and year gains are nothing to write home about which is why they're comparing the performance to the competition instead of the A14.

I wouldn't be too optimist. Comparing the A15 against 'competition' instead of the latest iPhone for the first time is a bad signal. I'm guessing they didn't get the 20% single core increase in performance this time. The benchmarks will be here anyway before the new Macs keynote anyway so no reason to be vague.

Ah, but the answer to those concerns is already in the presentation. I think it’s almost clear that they decided to focus on battery life instead. Their hardware is already two years ahead of anyone else, why bother making even faster phones? At some point it will get silly anyway. Dropping the clocks a bit, increasing the battery life - that’s where noticeable usability improvements lie. You are still way faster than the competition, but you improve the battery life as well, what’s not to like?

 

[leman]

I remain convinced the Macs to be announced next month will be on "M1X" and use the A14 as their foundation. I do not expect to see an A15-based "M2" until 1H 2022 at the earliest.

We‘ll see. I am fairly confident that the new prosumer chip will have little in common with M1…

 

[senttoschool]

Ah, but the answer to those concerns is already in the presentation. I think it’s almost clear that they decided to focus on battery life instead. Their hardware is already two years ahead of anyone else, why bother making even faster phones? At some point it will get silly anyway. Dropping the clocks a bit, increasing the battery life - that’s where noticeable usability improvements lie. You are still way faster than the competition, but you improve the battery life as well, what’s not to like?

I hope you're right. We're going to get benchmarks in about 10 days.

 

[pshufd]

I hope you're right. We're going to get benchmarks in about 10 days.

The mini ships Friday if you order it today. So perhaps results a bit sooner.

Edit: it delivers Friday or pickup in store on September 24.

Delivers: Thu, Sep 16
Pickup:Sep 24

 

[senttoschool]

The mini ships Friday if you order it today. So perhaps results a bit sooner.

Might get some leaks too.

 

[leman]

I hope you're right. We're going to get benchmarks in about 10 days.

I expect the benchmarks to show only modest improvements in performance, but also lower power consumption.

 

[senttoschool]

The mini ships Friday if you order it today. So perhaps results a bit sooner.

Edit: it delivers Friday or pickup in store on September 24.

Delivers: Thu, Sep 16
Pickup:Sep 24

How did you get September 16th delivery?

Also, LOL at the undefined. Come on Apple.

 

[pshufd]

Ooops. The Apple Pencil and other addons is on the same order page - the Friday deliver date is for the pencil.

 

[Macintosh IIcx]

Did you guys notice how vague the A15 presentation was? I take it as a further evidence that the new prosumer Macs will be based on similar technology. They don’t want to spill the beans until the Mac event.

That was extremely lackluster. There was info about double cache but now I’m concerned that it the only performance improvement part to the design. Oh well, we shall see. 😐

 

[Andropov]

Ah, but the answer to those concerns is already in the presentation. I think it’s almost clear that they decided to focus on battery life instead. Their hardware is already two years ahead of anyone else, why bother making even faster phones? At some point it will get silly anyway. Dropping the clocks a bit, increasing the battery life - that’s where noticeable usability improvements lie. You are still way faster than the competition, but you improve the battery life as well, what’s not to like?

Possibly. Their hardware has been two years ahead of anyone else for a few years now, I wonder why this year would be different.

On the other hand, it's true that some iPhones (specially the Mini) could really make use of more battery life. So maybe that's it. We shall see, IIRC Anandtech makes a energy efficiency estimate benchmark for the SoCs they test.

 

[MacCheetah3]

To add a a few additional takes on what could be the M1X worst case scenario (in addition to what others have said before):

1. hotter and louder than the performance increase justifies
2. heavy performance throttling under load
3. disappointing real-world performance increase from the extra CPU/GPU cores

Not strictly M1X problems, but the machines it goes in are key too:
5. poorly redesigned laptops with major issues (e.g. something akin to a butterfly keyboard or stage light)
6. new features that don't work as intended, either buggy software or hardware problems
7. limited availability and high price

I'm pretty confident Apple won't muck this up, but I would also be surprised if M1X wows in quite the same way as M1 did.

I think the most likely areas of disappointment will be price and heat/noise, but mostly price.

These are some of the reasons why I shake my head reading comments hoping Apple halves the volume of the Mac mini enclosure because there’s no need for “wasted space” or iMac chin whining. Think about it this way, if Apple keeps the designed for Intel cooling systems, there’s a lot of thermal headroom for Apple SoC increases.

While under heavy load (e.g., Handbrake), the M1 mini in an ambient temp of >=24ºC seemingly exhausts as much heat as my 2012 Mac mini at low to no load.

 

[MacCheetah3]

Na, I wouldn’t worry about that. Intel‘s TDP are mostly marketing numbers, they do not reflect real power dissipation over most of relevant scenarios. Remember how people were complaining that the Air with a 7W Intel CPU didn’t have proper cooling? Well, the same air with M1 somehow runs perfectly fine.

To be fair to Intel, their TDP numbers actually sort of do reflect power usage... Under specific circumstances.
No AVX workloads
Base clock only, and as long as there's headroom for it it'l not run at base

To be fair, TDP and similar measurements are mostly marketing for every company. In fact, you shouldn’t even use it to compare across an entire brand. Such measurements are realistically only useful for comparing models within one or two generations of a product family.

 

[pshufd]

To be fair, TDP and similar measurements are mostly marketing for every company. In fact, you shouldn’t even use it to compare across an entire brand. Such measurements are realistically only useful for comparing models within one or two generations of a product family.

I use it to spec out the PSU by adding up the TDP of CPU, GPU, MB and other components and then adding a percentage. So I expect the numbers to be accurate. That's why I was so surprised when I started seeing PL2 numbers.

 

[casperes1996]

To be fair, TDP and similar measurements are mostly marketing for every company. In fact, you shouldn’t even use it to compare across an entire brand. Such measurements are realistically only useful for comparing models within one or two generations of a product family.

Correct. But at least Intel's equation for computing TDP actually contains wattage in the formula. AMD's doesn't. Although it often more accurately reflects chip power draw in real world scenarios regardless

 

[leman]

I use it to spec out the PSU by adding up the TDP of CPU, GPU, MB and other components and then adding a percentage. So I expect the numbers to be accurate. That's why I was so surprised when I started seeing PL2 numbers.

They are not innacurate: Intel defines the TDP as the power dissipation of the chassis required to sustain at least the base CPU clock when running a prolonged complex workload on all cores. It’s basically a promise to the system maker, if you can dissipate X watts than the CPU will sustain at least Y ghz. It’s just not that useful to a user. Intel system run hot and noisy not because the Mac chassis is not enough to take care of the TDP (they can dissipate much more heat in fact), but because a) the CPU will often run way above its TDP, hearing the system up and b) even the smallest burst workload will push the CPU frequency to its max, again heating the CPU up. Disk indexing? Heat. Backup? Heat. Running syntactic analysis on a code file? Heat, heat, heat. The system is more often at its peak thermal load than not.

Apple Silicon is very different. You only get to the peak thermal load when you actually run a demanding workload, often one that utilizes multiple clusters (like CPU+GPU) Background things like indexing or backups don’t even show up because the E-cores will take care of them. And even burst workloads require a fraction of energy than with a x86 CPU. That’s why M1 macs are so cool and silent even though the sustained power dissipation is the same as their Intel counterparts. Of course, with M1 sustained dissipation is the peak dissipation. With Intel, peak dissipation can be 100W for a 15W TDP.

 

[jdb8167]

Was there anything in the iPad mini presentation on performance?

During the iPad mini reveal they said it has a 40% faster CPU and 80% faster GPU over the previous model. It uses a 6-core CPU with 2 performance and 4 efficiency. It also has a 4-core GPU. The A15 specs were not part of the iPad mini presentation but it was part of the iPhone 13 reveal. I assume the A15 in the iPad mini and iPhone 13 is the same silicon.