I'm interested what chips they can make for those garbage bins

This should actually have some gaming performance. I'd rather buy a gaming laptop that looks like a 16" MacBook Pro than all the abomination looking things they are selling as gaming laptops. I'm willing to bet that price/perf. with these new chips isn't even going to differ much anymore. I appreciate the competition Apple is bringing and I'd love to see all the Apple haters' faces melt when someone plays the same games with more fps on their MacBook than on their fugly Alienware Gaming Laptop Xtreme Edition.

PowerPCThis should actually have some gaming performance. I'd rather buy a gaming laptop that looks like a 16" MacBook Pro than all the abomination looking things they are selling as gaming laptops. I'm willing to bet that price/perf. with these new chips isn't even going to differ much anymore. I appreciate the competition Apple is bringing and I'd love to see all the Apple haters' faces melt when someone plays the same games with more fps on their MacBook than on their fugly Alienware Gaming Laptop Xtreme Edition.

... they wont be able to play the same games, its a totally different architecture...

8 big cores are going to provide a considerable performance boost given that the ~5X multi core scaling M1 has is atrocious but, there's a problem. They'll be running out of space, quite literally, their cores are huge with massive system caches. Will there be space for bigger GPUs ? Wont power efficiency go downhill really fast ? What about cooling ? SoCs are fundamentally not meant to be high performance, for some reason Apple thinks otherwise, see there is no real point if you end up making a huge power hungry SoC for a mobile device.

Keep in mind that their non-X SoCs still only have 2 high performance cores with a couple of low performance (like, really low performance if you look at MT scaling) little cores that are apparently based on a 2012 design, that should give people an idea of just how bad they scale.

the problem is the memory soldered hate that

Vya DomusWill there be space for bigger GPUs ?

The GPU doesn't need to get bigger it just needs to get faster.

Vya DomusWont power efficiency go downhill really fast ?

No, probably not. Efficiency scaling is more than just power draw from the wall. If the chip doubles in power but quadruples in performance than efficiency went up, not down. Plus Apple has a lot of wiggle room in this regard since they aren't even at the limit of battery tech. If they double SoC power on full load and up the battery a bit it'll still equalize over 12hrs battery life which is one of the design goals.

Vya DomusWhat about cooling ?

Likely going to look very similar. The fan basically doesn't turn on with the current design. It's there just to assure that temps stay low in less friendly environments. Add a bit more finstack and it'll be fine.

Vya DomusSoCs are fundamentally not meant to be high performance

Zen is also an SoC design and targets high performance.

PerfectWavethe problem is the memory soldered hate that

Yes it's unfortunate that there's not a DRAM socket in existence that can handle the bandwidth that the M1 requires. The cores themselves would be happy with ~80GB/s from the DRAM landing but GPU, ML, and some FF blocks need 200-300GB/s and large data sets to be even remotely effective. Access latencies from end to end is also a huge concern for the chip, and adding mere centimeters of wire to the design could double the latency.

FouquinNo, probably not. Efficiency scaling is more than just power draw from the wall. If the chip doubles in power but quadruples in performance than efficiency went up, not down. Plus Apple has a lot of wiggle room in this regard since they aren't even at the limit of battery tech. If they double SoC power on full load and up the battery a bit it'll still equalize over 12hrs battery life which is one of the design goals.

Yes it is a win but you forget, this is still low power device so we all know that it may scale nice with power vs performance but if it does hit the ceiling of the power allowed, it doesn't matter how good it scales.

FouquinThe GPU doesn't need to get bigger it just needs to get faster.

Sure and we all know that faster, doesn't have anything to do with bigger. Especially in graphics.

FouquinZen is also an SoC design and targets high performance.

Zen is also an SoC design and targets higher power limit thus higher performance comes with it. That's how you should have phrased it. You can also say Zen is an SoC design and targets higher performance thus higher power comes with it.
You keep forgetting that we are talking about mobile here. With low power budget. Why do you even bring a desktop CPU into this?

PowerPCThis should actually have some gaming performance. I'd rather buy a gaming laptop that looks like a 16" MacBook Pro than all the abomination looking things they are selling as gaming laptops. I'm willing to bet that price/perf. with these new chips isn't even going to differ much anymore. I appreciate the competition Apple is bringing and I'd love to see all the Apple haters' faces melt when someone plays the same games with more fps on their MacBook than on their fugly Alienware Gaming Laptop Xtreme Edition.

Comedy in word's, when's cyberpunk 2077 out then on it?!.

FouquinThe GPU doesn't need to get bigger it just needs to get faster.

That's how GPUs get faster, they get bigger. More cores scale almost linearly in performance, increasing IPC is difficult and increasing clocks are a bad idea.

FouquinNo, probably not. Efficiency scaling is more than just power draw from the wall. If the chip doubles in power but quadruples in performance than efficiency went up, not down.

This wont quadruple in performance, it will increase by 1.5-1.8X at best but power will double, that will put it close to the 25W mark making it more or less about the same with any other comparable x86 SoC.

FouquinLikely going to look very similar.

At the expense of frequency, these SoCs are really dense, it's not a matter of spinning the fans faster the thermal limitations come from the lack of surface area and energy density.

FouquinZen is also an SoC design and targets high performance.

No it doesn't, 8 low clocked CPU cores and 8 GPU 4 year old Vega cores, you call that high performance ? A high performance SoC would be something like the one used for consoles.

ratirtSure and we all know that faster, doesn't have anything to do with bigger. Especially in graphics.

I sense the smarmy sarcasm but no, bigger isn't necessarily faster. Maxwell vs Kepler. Navi 10 vs Vega 20. Architecture means as much or more than just stapling units together.

ratirtYou keep forgetting that we are talking about mobile here. With low power budget. Why do you even bring a desktop CPU into this?

Zen is also in the mobile and embedded space, and doing quite well might I add. Targeting down to 10W typical power envelopes. The whole point of Zen's SoC design was that it enhanced scalability from top to bottom.

Vya DomusThis wont quadruple in performance, it will increase by 1.5-1.8X at best but power will double, that will put it close to the 25W mark making it more or less about the same with any other comparable x86 SoC.

I won't argue numbers because it was an example, not a stated fact. So sure, your made up numbers are more valid than my made up numbers. It doesn't actually matter if the chip exceeds comparable x86 cores in sustained power draw (it won't) if it still outperforms them. Don't get so hung up on power draw when the work done is more important. To reiterate: if it completes the task faster it still uses less power.

Fouquinyour made up numbers are more valid than my made up numbers.

They are not made up, twice the cores means a maximum theoretical scaling of 2X. Naturally that wont ever happen in the real world, so 1.5-1.8X is a perfectly valid number, you can look at every other multicore CPU that has ever existed and you'll see the same kind of scaling.

FouquinIt doesn't actually matter if the chip exceeds comparable x86 cores in sustained power draw if it still outperforms them.

Absolutely ridiculous idea, of course power matters no matter how fast something is, especially in a mobile device.

Fouquinif it completes the task faster it still uses less power.

What ? How did you come up with that.

Vya Domus8 big cores are going to provide a considerable performance boost given that the ~5X multi core scaling M1 has is atrocious but, there's a problem. They'll be running out of space, quite literally, their cores are huge with massive system caches. Will there be space for bigger GPUs ?

I expect the M1X to be a bit more flexible, and possibly to be less of a monolithic SoC. I.e., a CPU (maybe as small iGPU) chip with DDR4 support, and a discrete GPU (possibly on the same package with HBM or LPDDR).

Sykobeeand a discrete GPU (possibly on the same package with HBM or LPDDR).

That sort of defeats the purpose, the reason SoCs are so power efficient is because they are tightly integrate, as soon as you put the GPU outside the SoC power increases considerably. All of which is counter intuitive to what they're trying to achieve.

Mussels... they wont be able to play the same games, its a totally different architecture...

Games that are going to be portet, obviously.

Vya DomusThat sort of defeats the purpose, the reason SoCs are so power efficient is because they are tightly integrate, as soon as you put the GPU outside the SoC power increases considerably. All of which is counter intuitive to what they're trying to achieve.

What do you know about what Apple is trying to achieve? It should be obvious that they are aiming for full market coverage. It's not just for power efficiency. They'll have performance versions of these probably sooner than you can say "Mac Pro".

Mussels... they wont be able to play the same games, its a totally different architecture...

That remains to be seen with Rosetta 2 in the picture. Other than architecture, everything else is the same, like calls to the Metal and whatnot. It might do better than you would expect, particularly for games that aren't CPU bound.

Vya DomusThey are not made up, twice the cores means a maximum theoretical scaling of 2X. Naturally that wont ever happen in the real world, so 1.5-1.8X is a perfectly valid number, you can look at every other multicore CPU that has ever existed and you'll see the same kind of scaling.

Right. Because clock increases don't add performance. Architectural improvements don't add performance. Your focus is much too narrow.

Vya DomusAbsolutely ridiculous idea, of course power matters no matter how fast something is, especially in a mobile device.

And we're still talking about 25W from your own made up numbers so we're well within the power budget limitations for a mobile device targeting 12HR battery life.

Vya DomusWhat ? How did you come up with that.

This is basic mathematics. If chip A uses 12W to do a task in 10 minutes, and chip B uses 25W to do it in 3.5 minutes than chip B uses less power to complete that task. Watt hours is what matters, not watts.

PowerPCGames that are going to be portet, obviously.


What do you know about what Apple is trying to achieve? It should be obvious that they are aiming for full market coverage. It's not just for power efficiency. They'll have performance versions of these probably sooner than you can say "Mac Pro".

uhh no, ported games can't be directly compared either - they've gotta be pretty much redesigned sometimes. Moving from X86-64 to ARM is nowhere near the same thing... this is like expecting to run Crysis or something on an android phone easily... no chance. none.

anything for these new devices is going to be coded specifically for them, so they'll run well but you can't compare them to anything else because its literally apples and oranges

5900x is dead

FouquinRight. Because clock increases don't add performance. Architectural improvements don't add performance. Your focus is much too narrow.

There is no reason to believe this chip will feature architectural changes or any significant clock speed increase, that's why it's still called M1X.

FouquinAnd we're still talking about 25W from your own made up numbers

Again, they are not made up, it looks like M1 in the mac mini already peaks close to 25W, you don't have to make up anything to estimate the power consumption of a chip with twice the cores.

FouquinThis is basic mathematics. If chip A uses 12W to do a task in 10 minutes, and chip B uses 25W to do it in 3.5 minutes than chip B uses less power to complete that task. Watt hours is what matters, not watts.

What a bizarre thought process, chips are not under 100% load all the time. The chip that goes up to 25W will still be a lot worse in terms of battery life than the one which is 12W, you refuse to accept that the relationship between performance and power is not linear, the higher you push the performance the lower the efficiency.

Has the M1 actually been reviewed anywhere yet, let alone M1X?

Will we be able to use M1 to run anything outside of Apple propriety software? If not it's DOA for those of us who wouldn't want be "locked in", from a software perspective anyway.

FouquinI sense the smarmy sarcasm but no, bigger isn't necessarily faster. Maxwell vs Kepler. Navi 10 vs Vega 20. Architecture means as much or more than just stapling units together.

Sure but here you have RDNA vs RDNA2 like 5700xt and 6800. You have 2080 Ti and 3080. It can also go all the way around. Besides there are more things that come into play.
Just to be clear. You are talking about the die size or transistor count? These two can be misleading. Just like the 2080Ti and 3080 are a different node so 2080 Ti is bigger in die size but 3080 packs more transistors. Just like Kepler vs Maxwell. The RDNA and RDNA2 are a better comparison since these are the same node. 6800 is almost twice as big as the 5700XT

Vya DomusThere is no reason to believe this chip will feature architectural changes or any significant clock speed increase, that's why it's still called M1X.

There's no reason to believe it won't contain architectural changes. You didn't design it, and neither did I. So you know zero and I know zilch. You can't make any statement for or against the architecture of M1X.

Vya DomusAgain, they are not made up, it looks like M1 in the mac mini already peaks close to 25W, you don't have to make up anything to estimate the power consumption of a chip with twice the cores.

The Mac Mini isn't a laptop now is it? The M1 in the Air peaks around 11W. But again you can't validate any numbers for a chip that doesn't yet exist, so you're making it up.

Vya DomusWhat a bizarre thought process, chips are not under 100% load all the time. The chip that goes up to 25W will still be a lot worse in terms of battery life than the one which is 12W, you refuse to accept that the relationship between performance and power is not linear, the higher you push the performance the lower the efficiency.

The higher you push the performance the lower the efficiency? Somebody better tell AMD! Zen 3 must be the least efficient chip they've ever made, it's way higher performance than Zen 2! :kookoo:

We're talking about watt hours to complete tasks. A higher TDP doesn't mean higher watt hours if the higher watt chip is faster per watt. And don't pretend that it'll idle at higher power we're talking about a chip with E cores and core-off power states. The idle wattage will be identical.

ratirtSure but here you have RDNA vs RDNA2 like 5700xt and 6800. You have 2080 Ti and 3080. It can also go all the way around. Besides there are more things that come into play.
Just to be clear. You are talking about the die size or transistor count? These two can be misleading. Just like the 2080Ti and 3080 are a different node so 2080 Ti is bigger in die size but 3080 packs more transistors. Just like Kepler vs Maxwell. The RDNA and RDNA2 are a better comparison since these are the same node. 6800 is almost twice as big as the 5700XT

Are we just going to forget IPC, ILP and clockrate? On 28nm GM204 with 2B fewer transistors and 200mm smaller die overtook GK110B in performance. On 7nm RDNA took 3B fewer transistors on a 80mm smaller die to match and sometimes outperform the 13B transistors in GCN5.1. Architecture. It matters a lot.

Musselsuhh no, ported games can't be directly compared either - they've gotta be pretty much redesigned sometimes. Moving from X86-64 to ARM is nowhere near the same thing... this is like expecting to run Crysis or something on an android phone easily... no chance. none.

anything for these new devices is going to be coded specifically for them, so they'll run well but you can't compare them to anything else because its literally apples and oranges

Plenty of games already are running on OS X for x86_64 already. That's the hard part, not handling the CPU architecture differences. If you're talking about games that run on Windows using OpenGL, Vulkan, or DirectX, sure, but that's hard because you have to port it to OS X with a different graphics API. Even Rosetta 2 performance is actually pretty good most of the time. I suspect that most games already running on OS X pre Apple Silicon with Metal, already probably work on the M1. Phoronix has a series of benchmarks that compares native and rosetta performance of the M1 on the Mac Mini. The numbers are actually pretty decent.

FouquinYou can't make any statement for or against the architecture of M1X.

Of course I can by looking at the past and present, something which you don't or refuse to. The X variants of a chip have never ever featured a different architectures just more CPU and GPU cores.

FouquinThe Mac Mini isn't a laptop now is it? The M1 in the Air peaks around 11W.

It's not but the chip is the same, I know you're trying really hard to refute this but twice the amount of big cores will mean a lot more power.

FouquinThe higher you push the performance the lower the efficiency?

Yes it does, that's why CPUs drop their clock speed when multiple cores are used and if you disable the throttling power figures make your eyes water. The wider and the more cores you have the worse the power leakage gets and the more difficult it is to power gate effectively portion of the chips.

A first gen Zen 1 core topped out at around 8-9W, a Zen 3 core tops out at 20W. A Zen 3 core is about 1.8X faster but uses 2.5X times the power and keep in mind that's on a much better node. Still don't believe that power efficiency takes a hit the more performance you extract ?

FouquinWe're talking about watt hours to complete tasks.

No we are not, you are. No one uses a PC in the way you describe, I feel like I am repeating myself indefinitely, the bigger the chip and the higher the power consumption and the worse the battery life gets, even if performance/watt has increased. The only situation when that's not true is if the chip is used at maximum load until the battery dies, almost no one uses their device like that.

AquinusPlenty of games already are running on OS X for x86_64 already. That's the hard part, not handling the CPU architecture differences. If you're talking about games that run on Windows using OpenGL, Vulkan, or DirectX, sure, but that's hard because you have to port it to OS X with a different graphics API. Even Rosetta 2 performance is actually pretty good most of the time. I suspect that most games already running on OS X pre Apple Silicon with Metal, already probably work on the M1. Phoronix has a series of benchmarks that compares native and rosetta performance of the M1 on the Mac Mini. The numbers are actually pretty decent.

Yeah. In my previous job we ported GPGPU libraries from PC to mobile platforms. If it worked on a x86 Mac with Metal already, getting it to work on a iPhone is a breeze (assuming XCode didn't ***t itself, which happens more often than I like).