MacBook Pro's M1 Max GPU is Over 3x Faster Than M1 in First Metal Benchmark

[Spidder]

It’s not crazy to think that they are capping frequencies and allowing full power in this mode with only the 16” because it can be done without throttling.

Technically possible, but that would be borderline false advertisment.

 

[priamXus]

Technically possible, but that would be borderline false advertisment.

That’s why I said ill complain about it. It’s a serious offense in Quebec.

There’s def a cap:

Getting to the end of the video there’s uncertainty yet. Guess that we will have to wait to confirm.

 

[PPietra]

I don’t know. The other way has sense too:

- Apple states that this mode will only be available in 16”
- Geenkbench frequencies at 1000MHz
- MBP 16” with M1 Max weights 100gr more.
- Subjective: everyone was expecting more in Metal score and thinking about something weird / wrong. Also, WTF with that 140w power adapter.
- Possible: Benchmarks during presentation were ran in this mode for the M1 Max.

It’s not crazy to think that they are capping frequencies and allowing full power in this mode with only the 16” because it can be done without throttling.

Geekbench is hardly reliable with those sorts of information, it is always presenting the wrong information. By the way, that is the same frequency that it shows for the M1 Pro and the M1, not exactly an SoC that would need to be throttled.
We still don’t have enough information about what is the score for the 32-core GPU - we cannot trust the info in geekbench.
If it is about throttling then it would not be about capping the frequency, because you can always achieve the frequency if you are allowing it to throttle afterwards. The fact is that there isn’t nothing substantial about the extra 30W for the additional 16 cores in the M1 Max.

 

[StoneJack]

RX580 is an old budget GPU from 2017 so you're GPU limited. Hop on amd.com and grab a 6900xt on their next drop.

6900xt Vulkan
https://browser.geekbench.com/v5/compute/3564577

6900xt OpenCL
https://browser.geekbench.com/v5/compute/3564587

Oh I see you learned to google

 

[CosmoPilot]

Good question. I've given up on trying to guess what Apple will do. As of this point I am leaning towards "M" standing for mobile. The M1 was clearly designed for laptops, but had enough oomph to stick in other low-end desktops. And they said the M1 Pro/Max were specifically designed for high-end ("pro") laptops. Those could very well end up in the first iteration of the new iMacs as well.

Why would any of these companies want to deal with Apple and their secretive nature. Aside from iphone games, please Apple systems are not game developer friendly at all and hence nobody wants to make games for it.

Profit is always the incentive for making a product, obviously these companies see there's no value in doing it. We can debate it all day long but this has been written in stone for so long. This is all apples fault, they are the owners of the platform, they are the secretive ones or highly restrictive when it comes to driver's etc. It's not a conducive environment for any decent size studio. Just make a iPhone game and be done with it.

This dance has been going on for so long with Apple that it's utterly pointless. I used to have great gaming times on my Mac back in 2000, it is about 22 years later and you could not get me to care about Matt games at all. I have a series X and A PS5 why would I care about what apple is doing?

Im not a gamer and I mean no disrespect. But saying it’s Apple’s fault is a bit short sighted IMO. The iOS App Store is a Fortune 500 company be itself. Developers have made HUGE fortunes developing for iOS.

So why not the Mac? Market share! Not many people own Macs as a percentage as own iPhones. Why? Because Macs are premium products that command a premium price to entry. Therefore business choose to stay on the Windows platform where they can mass purchase $300 computers for their employees. That translates to people needing Windows applications on their home rigs.

Also, everyday peeps would rather spend $300-500 on a windows crap computer to browse the web and check email than pay the premium to own a superior platform (Mac).

So developers have to make a choice….target the 60-70% market or the 15-20% market. I don’t think Apples business practices are the reason as iOS has been widely successful for developers. It’s just the overwhelming majority of people don’t own Macs for reasons stated.

As Steve Jobs once said, we can make crappy systems, but we choose to build products we would want our friends and families to own.

 

[StoneJack]

Im not a gamer and I mean no disrespect. But saying it’s Apple’s fault is a bit short sighted IMO. The iOS App Store is a Fortune 500 company be itself. Developers have made HUGE fortunes developing for iOS.

So why not the Mac? Market share! Not many people own Macs as a percentage as own iPhones. Why? Because Macs are premium products that command a premium price to entry. Therefore business choose to stay on the Windows platform where they can mass purchase $300 computers for their employees. That translates to people needing Windows applications on their home rigs.

Also, everyday peeps would rather spend $300-500 on a windows crap computer to browse the web and check email than pay the premium to own a superior platform (Mac).

So developers have to make a choice….target the 60-70% market or the 15-20% market. I don’t think Apples business practices are the reason as iOS has been widely successful for developers. It’s just the overwhelming majority of people don’t own Macs for reasons stated.

As Steve Jobs once said, we can make crappy systems, but we choose to build products we would want our friends and families to own.

Actually you can't play AA games on shtty 300 dollar Windows computers. Graphic card alone would cost more than 1k dollars.

 

[CosmoPilot]

Actually you can't play AA games on shtty 300 dollar Windows computers. Graphic card alone would cost more than 1k dollars.

True!

 

[priamXus]

Im not a gamer and I mean no disrespect. But saying it’s Apple’s fault is a bit short sighted IMO. The iOS App Store is a Fortune 500 company be itself. Developers have made HUGE fortunes developing for iOS.

So why not the Mac? Market share! Not many people own Macs as a percentage as own iPhones. Why? Because Macs are premium products that command a premium price to entry. Therefore business choose to stay on the Windows platform where they can mass purchase $300 computers for their employees. That translates to people needing Windows applications on their home rigs.

Also, everyday peeps would rather spend $300-500 on a windows crap computer to browse the web and check email than pay the premium to own a superior platform (Mac).

So developers have to make a choice….target the 60-70% market or the 15-20% market. I don’t think Apples business practices are the reason as iOS has been widely successful for developers. It’s just the overwhelming majority of people don’t own Macs for reasons stated.

As Steve Jobs once said, we can make crappy systems, but we choose to build products we would want our friends and families to own.

“Premium” almost every notebook nowadays with similar specs are priced the same. Even Anthony said that the other day in LTT

 

[mjtomlin71]

Unless Geekbench’s frequencies are correct. Remember; this is not available yet.

The M1 is available and every OpenCL compute test says "maximum frequency: 1000Mhz" We know for a fact that is wrong.


We also know for a fact that there are M1 MacBook Air GPU's with 7 compute units, yet there's not a single one in Geekbench results... they all say 8, which is odd.

There is definitely something wrong with Geekbench when it comes to the M1 SoC.

 

[mi7chy]

Oh I see you learned to google

Why invest all that time setting up a hackintosh then using something as old as dirt?

 

[PPietra]

That’s why I said ill complain about it. It’s a serious offense in Quebec.

There’s def a cap:

Getting to the end of the video there’s uncertainty yet. Guess that we will have to wait to confirm.

Actually I noticed one interesting thing, and who knows if this is relevant or not. In the GPU charts that Apple showed the M1 Max doesn’t double the power consumption over the M1 Pro, nor double the performance (+90% only). So maybe the High Power mode might give the extra 10% in performance still missing.

 

[theotherphil]

Yeah, that's the Max and only in the 16" which makes sense since there's a larger thermal envelope in a bigger chassis.

That could mean two different things...
1. it has higher clock frequencies that it can use when needed (this means it has a "Turbo" mode like Intel/AMD)
2. it has a higher thermal threshold (this means it runs at max speed longer).

The 16" MBP in Max configuration is heavier than the non-max per the specs. This doesn't happen in the 14" so I would suspect that the 16" Max models have a different cooling system to allow the "High Performance" mode.

Therefore, I suspect both 1 & 2 are correct. I also suspect that Apple only quoted performance figures without this "High Performance" mode being used...because they showed the performance of the M1 Max and didn't specifically state the M1 Max in 16" model under High Performance conditions. This implies this baseline benchmarked performance is available on both the 14" and 16".

I further suspect that this "High Performance" mode will be only available whilst plugged in....hence the 140W power brick.

 

[priamXus]

The 16" MBP in Max configuration is heavier than the non-max per the specs. This doesn't happen in the 14" so I would suspect that the 16" Max models have a different cooling system to allow the "High Performance" mode.

Therefore, I suspect both 1 & 2 are correct. I also suspect that Apple only quoted performance figures without this "High Performance" mode being used...because they showed the performance of the M1 Max and didn't specifically state the M1 Max in 16" model under High Performance conditions. This implies this baseline benchmarked performance is available on both the 14" and 16".

I further suspect that this "High Performance" mode will be only available whilst plugged in....hence the 140W power brick.

If it’s like this I’m fine with that which would return me to my purchase decision; avoiding the 16” just for that brick.

 

[Botts85]

If this thing works well for ethereum mining, we're in for a bad time.

That said, I wouldn't mind if it made me $5 a day while idling on my desk

I do want one soon though, not in 2 years after the miners swipe all of them.

 

[StoneJack]

Why invest all that time setting up a hackintosh then using something as old as dirt?

Maybe because I don't want to pay 2k dollars for the latest 6800 card (miners, thank you)? and RX580 is just fine for video editing and gaming in FHD (which is my monitor resolution). It is similar to 1050ti, which is on steam is a current most used videocard.

 

[travelsheep]

IMO you need the fans for FCP.

I was thinking about that too... where I live the outside temp is usually around 32°C daytime (28°C nighttime), aircon not always on, I guess fans would be needed to stay cool and keep throttling below 100%.

 

[Velli]

Assuming this is a 24 core and scaling everything up, 4C Die, the rumored top spec Mac Pro coming, would be about the same as dual Pro W6900X, with teice the CPU of 28 core. IF (a big if) Apple scales the price semi-resonably, this should be around 3-4k dollars up from M1 Max. Compared to pricing on upgrading current Mac Pro, this is WAY cheaper! Surely there is a limit to how much Apple could upcharge once they integrate it all in an SOC (right? RIGHT??)

 

[mjtomlin71]

The 16" MBP in Max configuration is heavier than the non-max per the specs. This doesn't happen in the 14" so I would suspect that the 16" Max models have a different cooling system to allow the "High Performance" mode.

Therefore, I suspect both 1 & 2 are correct. I also suspect that Apple only quoted performance figures without this "High Performance" mode being used...because they showed the performance of the M1 Max and didn't specifically state the M1 Max in 16" model under High Performance conditions. This implies this baseline benchmarked performance is available on both the 14" and 16".

I further suspect that this "High Performance" mode will be only available whilst plugged in....hence the 140W power brick.

I would venture to guess that extra tenth of a pound is for a bigger copper cooling bar to carry more heat away from the SoC.

And I agree, Apple's stated metrics in their marketing implies being used normally and not in a "High Performance" mode. Otherwise that would potentially make the 14" MBP 32c Max almost a waste of money to upgrade to and piss off a lot of people if they don't see the performance gains.

The Max should have a TDP of around ~100W which is comparable to the highest end Intel MBP w/5600M which at normal load, together hit 95W (45W/50W). The reason for fans turning into jet engines is because the Intel alone will double its wattage when running in Boost mode (95W @ 5.0GHz). I couldn't find any info on the AMD, but I'd guess it boosts to around 55-60W. Now we know the thermal control in the Intel MacBookPros were not very good due to the fact they were designed for a much more efficient chip that Intel never delivered on. So, if they designed this MBP correctly (and sounds like they did), it should have plenty of room to "boost" without needing to severely increase the fans.