October 18th Apple Event - Unleashed - Apple Silicon, MacBook Pro upgrades, HomePod mini, AirPods 3rd Generation

[leadeater]

1 hour ago, Imbadatnames said:

The selling point is battery though and a lot of manufactures like to use 200+W for the performance figures off the wall and then use 20W low power mode with 10% screen brightness for the battery life. Ultimately a lot of the more powerful laptops aren’t laptops at all they’re a workstation with a screen built in because you can’t actually do work on them for more than an hour on battery. 

It's a selling point for these Apple silicon laptops correct, and it's a huge benefit to not have a performance difference, however that's not a reason to exclude a test configuration when comparing against other laptops.

[Imbadatnames]

44 minutes ago, leadeater said:

It's a selling point for these Apple silicon laptops correct, and it's a huge benefit to not have a performance difference, however that's not a reason to exclude a test configuration when comparing against other laptops.

I was talking about other laptops. They will all boast battery life at x hours and performance numbers as whatever times x hardware which is a lie. “High end” laptops are basically not functional as a laptop. 

[WolframaticAlpha]

On 12/17/2021 at 7:41 PM, Imbadatnames said:

Tbh every laptop review should be on battery only 

Depends really. An ultrabook makes sense to be on battery, but a portable workstation does not.

 

On 12/17/2021 at 5:04 PM, saltycaramel said:

F its power is harnessed properly (e.g. native ARM app using Metal), the GPU in the M1 Max can be as fast as a lowish TDP mobile 3080. 

Any system can become faster if you put in time for optimizations. The fact is that I am not going to be buying a laptop because it can do great things if it is properly optimized. People will buy it depending on how it works for them, and it doesn't matter how fast it is in certain applications. The Microsoft SQ2 can be a great chip, but is being held back by crap software. By your reasoning we should only be using benchmarks specially optimized for it.

[hishnash]

25 minutes ago, WolframaticAlpha said:

Depends really. An ultrabook makes sense to be on battery, but a portable workstation does not.

 

does still make sense to consider it, even a portable workstation might end up not being always connected to power or be used in a situation were the wall socket can't provide more than 100W (Airplane and other transport provided sockets) 

[HenrySalayne]

2 hours ago, hishnash said:

does still make sense to consider it, even a portable workstation might end up not being always connected to power or be used in a situation were the wall socket can't provide more than 100W (Airplane and other transport provided sockets) 

We should also test for water tightness, because these machines might end up at the bottom of the ocean.

 

 

 

 

[Imbadatnames]

4 hours ago, WolframaticAlpha said:

Depends really. An ultrabook makes sense to be on battery, but a portable workstation does not.

 

Any system can become faster if you put in time for optimizations. The fact is that I am not going to be buying a laptop because it can do great things if it is properly optimized. People will buy it depending on how it works for them, and it doesn't matter how fast it is in certain applications. The Microsoft SQ2 can be a great chip, but is being held back by crap software. By your reasoning we should only be using benchmarks specially optimized for it.

Portable workstations aren’t consumer hardware though. 

[mr moose]

42 minutes ago, Imbadatnames said:

Portable workstations aren’t consumer hardware though. 

Professionals still need advice and reviews before they purchase.  Not every pro works for a company with an IT team dedicated to tooling.

[leadeater]

On 12/18/2021 at 5:27 AM, Imbadatnames said:

I was talking about other laptops. They will all boast battery life at x hours and performance numbers as whatever times x hardware which is a lie. “High end” laptops are basically not functional as a laptop. 

That wasn't the point, performance on power is still important because many will run it on power and it's still just as important to know what that performance is regardless. High end laptops work just fine and function as anyone should expect, if you load up a game and draw a huge amount of power then run time being short is not unexpected and much of a criticism. Anybody that knows even a small about computers knows this.

 

A good laptop with a decent battery and a high(er) end GPU can still render out most things in Blender on battery just fine without running out of battery life. Sure you can't do that 3 times on battery without plugging in but then you have to ask the question "how often are you truly away from a power source".

 

Are you really going to raise the argument that a 30+ minute render where you essentially won't use the laptop that you would forgo plugging it in to power if it were available. Not only would that not use the battery capacity it would also complete the task faster.

 

High end laptops aren't as useless as you are trying to make out. Very few portable devices have zero draw backs and offer the best performance in every task possible.

 

Every single hot desk at work has a laptop docking station that providers power, ethernet and dual monitors. You'd be very hard pressed to not be able to plugin at some point, some of the few times would be in meeting rooms.

 

The point is not doing on power performance test configurations isn't a good idea, more information is better not less.

 

1 hour ago, Imbadatnames said:

Portable workstations aren’t consumer hardware though. 

So then never review laptops that could be used professionally? Goodbye every single MacBook Pro review.

[mr moose]

35 minutes ago, leadeater said:

So then never review laptops that could be used professionally? Goodbye every single MacBook Pro review.

And in one sentence the entire argument was shown to be self defeating.     

[Imbadatnames]

1 hour ago, leadeater said:

That wasn't the point, performance on power is still important because many will run it on power and it's still just as important to know what that performance is regardless. High end laptops work just fine and function as anyone should expect, if you load up a game and draw a huge amount of power then run time being short is not unexpected and much of a criticism. Anybody that knows even a small about computers knows this.

 

A good laptop with a decent battery and a high(er) end GPU can still render out most things in Blender on battery just fine without running out of battery life. Sure you can't do that 3 times on battery without plugging in but then you have to ask the question "how often are you truly away from a power source".

 

Are you really going to raise the argument that a 30+ minute render where you essentially won't use the laptop that you would forgo plugging it in to power if it were available. Not only would that not use the battery capacity it would also complete the task faster.

 

High end laptops aren't as useless as you are trying to make out. Very few portable devices have zero draw backs and offer the best performance in every task possible.

 

Every single hot desk at work has a laptop docking station that providers power, ethernet and dual monitors. You'd be very hard pressed to not be able to plugin at some point, some of the few times would be in meeting rooms.

 

The point is not doing on power performance test configurations isn't a good idea, more information is better not less.

 

So then never review laptops that could be used professionally? Goodbye every single MacBook Pro review.

Workstation laptops aren’t aimed at the general public. They’re for companies and organisations with workloads that require portable compute power. They’re not for a CAD hobbyist or YouTuber. They’re for design firms or people doing field work. 

[Imbadatnames]

1 hour ago, mr moose said:

And in one sentence the entire argument was shown to be self defeating.     

MacBooks can be used for general compute, as a hobbyist or for work as a self employed startup requiring a high performance computer. Workstation laptops aren’t for that use case. 

[leadeater]

14 minutes ago, Imbadatnames said:

Workstation laptops aren’t aimed at the general public. They’re for companies and organisations with workloads that require portable compute power. They’re not for a CAD hobbyist or YouTuber. They’re for design firms or people doing field work. 

And yet there is very little difference between those and gaming laptops and that is still not a reason to not review them or review them in any different way to any other laptop. How are you supposed to get an informed picture of laptops if you test them all differently because you want to deem them as different classifications of devices and segment the testing.

 

How about both on power and on battery testing is performed and then nobody loses out and we get all the information not half of it.

 

Also what is a "workstation laptop", one that happened to have a GPU that is greater than a RTX 3050 and doesn't have RGB? Not all "workstation" laptops have "Quadros" in them. Not all "workstation" workloads need a GPU at all. Segmenting and isolating  products and treating them differently is a very Apple way of thinking where as simply finding the most suitable laptop for your needs and tastes is a very "PC" thing to do.

[Imbadatnames]

27 minutes ago, leadeater said:

And yet there is very little difference between those and gaming laptops and that is still not a reason to not review them or review them in any different way to any other laptop. How are you supposed to get an informed picture of laptops if you test them all differently because you want to deem them as different classifications of devices and segment the testing.

 

How about both on power and on battery testing is performed and then nobody loses out and we get all the information not half of it.

 

Also what is a "workstation laptop", one that happened to have a GPU that is greater than a RTX 3050 and doesn't have RGB? Not all "workstation" laptops have "Quadros" in them. Not all "workstation" workloads need a GPU at all. Segmenting and isolating  products and treating them differently is a very Apple way of thinking where as simply finding the most suitable laptop for your needs and tastes is a very "PC" thing to do.

Yes a workstation laptop does require a quadro as it’s a workstation along with a powerful CPU both with adequate cooling. If it does not have a quadro then it is inferior for workstation tasks and does not justify the cost. Unfortunately because they’re not high volume and not modular like actual workstations you only have the option of having both. I use workstation laptops daily there are things they do that consumer hardware just does not do very well. 

[leadeater]

25 minutes ago, Imbadatnames said:

Yes a workstation laptop does require a quadro as it’s a workstation along with a powerful CPU both with adequate cooling

No it doesn't and then additionally that invalidates your entire point then. Many laptops with moderate GPUs have vastly lower on battery run times when heavily utilizing them. A workstation laptop absolutely does not require a Quadro nor a GPU at all. Not every high end workload uses a GPU and not every high end workload benefits at all from a Quadro, it makes zero sense and is horrible advice to recommend buying a laptop with a Quadro class GPU when all that is required is an RTX 3050 and there would be minimal to no performance increase.

 

Lenovo W series and now P series laptops are specifically marketed as their workstation line, you can get non workstation class GPUs in them.

 

What you are doing is unnecessarily shoe horning extra requirements in to your argument which further detracts from and invalidates your original argument that laptops lose significant amounts of performance on battery which is true and not isolated to workstation laptops. Therefore "workstation" is not a limiter to the discussion at all.

 

You originally said laptops should only be reviewed on battery, I'm sorry but I strongly disagree with that statement. I agree that on battery testing should be done but not as a replacement of on power and both should be done. Yes it's unfortunate not every reviewer does on battery and on power testing but there are those that do right now.

 

On 12/18/2021 at 3:11 AM, Imbadatnames said:

Tbh every laptop review should be on battery only 

The long and short of this discussion is I'm never going to agree with your statement here. By saying only I cannot agree and never will.

[Imbadatnames]

10 minutes ago, leadeater said:

No it doesn't and then additionally that invalidates your entire point then. Many laptops with moderate GPUs have vastly lower on battery run times when heavily utilizing them. A workstation laptop absolutely does not require a Quadro nor a GPU at all. Not every high end workload uses a GPU and not every high end workload benefits at all from a Quadro, it makes zero sense and is horrible advice to recommend buying a laptop with a Quadro class GPU when all that is required is an RTX 3050 and there would be minimal to no performance increase.

 

Lenovo W series and now P series laptops are specifically marketed as their workstation line, you can get non workstation class GPUs in them.

 

What you are doing is unnecessarily shoe horning extra requirements in to your argument which further detracts from and invalidates your original argument that laptops lose significant amounts of performance on battery which is true and not isolated to workstation laptops. Therefore "workstation" is not a limiter to the discussion at all.

 

You originally said laptops should only be reviewed on battery, I'm sorry but I strongly disagree with that statement. I agree that on battery testing should be done but not as a replacement of on power and both should be done. Yes it's unfortunate not every reviewer does on battery and on power testing but there are those that do right now.

 

The long and short of this discussion is I'm never going to agree with your statement here. By saying only I cannot agree and never will.

Name a high end process that does not benefit from a quadro or a highly threaded CPU? 

[leadeater]

16 minutes ago, Imbadatnames said:

Name a high end process that does not benefit from a quadro or a highly threaded CPU? 

Why are you now adding on highly threaded CPU, I just said you are shoe horning extra requirements on which you just did again. I only said not every high end workload needs a GPU or benefits from a greatly more powerful one. So why are now treating my statement as now not also needing a high end CPU? I said no such thing.

 

This does not warrant a reply to your question. Not that I couldn't name multiple that don't need a GPU however you aren't conversing genuinely so are rapidly losing the right to receive a reply at all.

 

It also changes nothing about your original statement that I don't agree with.

[igormp]

2 hours ago, Imbadatnames said:

Name a high end process that does not benefit from a quadro or a highly threaded CPU? 

You can get workstations with a really powerful CPU and no dedicated GPU whatsoever.

[saltycaramel]

11 hours ago, WolframaticAlpha said:

Depends really. An ultrabook makes sense to be on battery, but a portable workstation does not.

 

Any system can become faster if you put in time for optimizations. The fact is that I am not going to be buying a laptop because it can do great things if it is properly optimized. People will buy it depending on how it works for them, and it doesn't matter how fast it is in certain applications. The Microsoft SQ2 can be a great chip, but is being held back by crap software. By your reasoning we should only be using benchmarks specially optimized for it.

 

Apparently

1) using the proper native architecture of the CPU (instead of relying on Rosetta)

2) using a graphical API Apple has been promoting for 7 years already (= all the warning in the world in advance) and that is abundantly used on 1 billion iPhones/iPads (not like it is some obscure API nobody uses)

 

can be labeled as “optmizations”.

Like we’re talking optimizing for the byzantine Playstation 3 architecture. 

 

That’s some twisting to justify the M1 Max = 3050 equation.

 

Anyway, these M1 Macs are only 1 year old.

Software that adequately leverages their hardware (including GPUs) will come (how do I know? because they’re too compelling machines overall not to attract it). It will only get better and better, exactly the opposite of Anthony’s closing statement about “consumers moving on, making the purpose-built hw dead weight”. 

[HenrySalayne]

4 hours ago, saltycaramel said:

Apparently

1) using the proper native architecture of the CPU (instead of relying on Rosetta)

2) using a graphical API Apple has been promoting for 7 years already (= all the warning in the world in advance) and that is abundantly used on 1 billion iPhones/iPads (not like it is some obscure API nobody uses)

 

can be labeled as “optmizations”.

This is by definition the meaning of the word "optimization". To improve something and make it more suitable.

And once again, AVX512 is a native part of all Intel desktop processors in recent years (even 12th gen), so why are you not complaining about that?

4 hours ago, saltycaramel said:

Software that adequately leverages their hardware (including GPUs) will come (how do I know? because they’re too compelling machines overall not to attract it). It will only get better and better, exactly the opposite of Anthony’s closing statement about “consumers moving on, making the purpose-built hw dead weight”. 

I think you completely missed his point here. He was not talking about M1 in general, but about the architecture relying heavily on ASIC blocks for a wide variety of tasks (like video playback). That's how Apple manages the impressive 22 h of video playback or the huge number of parallel video tracks. If Youtube changes their video codec - good luck getting your 22 h out of the machine.

The same is true for NVENC or the video encoders built into AMD or Intel CPUs, but to a much lesser degree. They generally use a wide-purpose approach.

 

Your entire argument is "only test specialized and optimized stuff" while PCs generally should do a wide variety of tasks. And that is exactly what Anthony did in his review. A wide range of software, showing strength and weaknesses of the M1 platform comparing it to a Intel/Nvidia platform.

[igormp]

4 minutes ago, HenrySalayne said:

The same is true for NVENC or the video encoders built into AMD or Intel CPUs, but to a much lesser degree. They generally use a wide-purpose approach.

They don't, if youtube releases a new codec, those older encoders are all useless.

[HenrySalayne]

41 minutes ago, igormp said:

They don't, if youtube releases a new codec, those older encoders are all useless.

My comment was misleading. I wanted to say that the en- or decoder is only a small part of the hardware. The rest is just general-purpose (or wide-purpose) hardware.

[Imbadatnames]

6 hours ago, igormp said:

You can get workstations with a really powerful CPU and no dedicated GPU whatsoever.

Show me a laptop with one 

[igormp]

15 minutes ago, Imbadatnames said:

Show me a laptop with one 

Sure, you can get a ZBook G8 without an nvidia gpu:

 

Or if you want a xeon CPU:

[hishnash]

1 hour ago, HenrySalayne said:

I think you completely missed his point here. He was not talking about M1 in general, but about the architecture relying heavily on ASIC blocks for a wide variety of tasks (like video playback). That's how Apple manages the impressive 22 h of video playback or the huge number of parallel video tracks. If Youtube changes their video codec - good luck getting your 22 h out of the machine.

The same is true for NVENC or the video encoders built into AMD or Intel CPUs, but to a much lesser degree. They generally use a wide-purpose approach.

So YouTube does not use H265 the video decode on YouTube is AV1 or V9 or H264 depending on the resolution and browser your using. And every cpu that has shipped in the last 4 years at least has had dedicated H264 decoding blocks. Having dedicated hardware decoders is not unique to apples products and is not the reason for big power savings. The key reason for large power savings is much more affiant cpu arc, much lower power memory for the bandwidth (that comes at the cost of not being modular). 

The NVAENC decoders and decoders are just the same as apples encoder/decoders the really is not difference apart from apple spending a lot more die area on encoders letting them do much more parallel encoding (not relevant at all for YouTube playback but very very relevant for video workflows).  
 

But even in situations were there are no hardware encoding blocks (for video codecs that doe not have any) these chips do very very well, the ability for the CPU and GPU to share memory addresses and read/write to them as if they are just 2 cores sharing memory really really really helps the video encode/decode pathways. These problems end up with a mistures of vector/matrix math (good for the gpu) and integer branching math (very very bad for gpu) by being able to share the memory (without needing to copy) between CPU and GPU these codecs can spread the workload across the gpu and cpu depending on the instruction pathways without needing to constantly wait for data to be copied back and forth.  This is also true for the hardware decoders/encoders they can access the memory just the same as a cpu core or gpu core so for tasks that can use them can also work with much much lower overhead. As a developer there is also a really big benefit to this model, typeicly for use to be able to make the most out of the GPU CPU combo we would need to do a lot of hard work (optimising) the task to figure out how we can send waves of GPU optimal instructions to the gpu while still having work to do in the cpu and then have new work ready for the gpu just as it finishes without waiting around for the CPU to finish its bit of work.  Having a unified memory address space makes things a lot LOT simpler. 

 

[Imbadatnames]

1 hour ago, igormp said:

Sure, you can get a ZBook G8 without an nvidia gpu:

 

Or if you want a xeon CPU:

See the graphics on the “not nvidia” model is interesting as the 11850H has an Intel UHD based iGPU so the Xe GPU in there looks to be a dedicated professional GPU unless the website is wrong. Oh and the Zbook isn’t actually one line it’s an overarching brand with separate product lines. Some of which are workstation platforms some aren’t.