Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Anandtech: M1 Max multithread performance is like Ryzen 5950x
- Thread starter senttoschool
- Start date
- Sort by reaction score
This has nothing to do with my posts. We know that MBPs are more power efficient than their PC counterparts.
Luke Miani only said a few words so he will have to elaborate. I expect to see detailed battery reviews from lots of reviewers in the next two weeks. There are reviewers that specialize in the various niches of Mac reviews and I keep an eye on what they come out with.
Mine arrives tomorrow and I can look at my impressions on battery life.
"Desktop territory" starts with the M1 Mac mini. It's such a wide territory that it's a rather useless term.
280W for the chip vs 140W for the MBP 16" charger.
The point I am getting across is that for the power consumption of a whole MBP 16" it outperforms 1 component that uses more than 140W.
Throw in the Amazon MiniPCs and you can get “desktop territory” Celerons.
I assume that 140 watts will also power USB-C devices, the screen, etc. The on-die power usage should include SSD power which is not typically the case in laptops. Maybe the same is true for RAM.
The flipside there is that if we were to cap both to 65W, Apple would still have a lot more headroom, and the M1 Pro would be at the point where it might not need to throttle at all.
The i7 and i9 chips Apple was using before was also known for high levels of power consumption. Intel claimed 45W TDP, but the i9 in particular was known for pulling well beyond TDP even before the dGPU was taken into consideration.
They still scale amazingly well. Keep in mind that full CPU load should be around half the power of the Intel 16" MBP, with close to double the perf. Fans should be fairly quiet when I'm sitting in Xcode, I suspect. It's more that instead of using the opportunity to pull total peak consumption down, they just kept scaling the GPU higher until they ran out of thermal headroom again.
As long as the fans/heat are mostly when I'm under that full load of CPU+GPU, it's still a massive improvement, IMO. I'll find out tomorrow when my unit arrives. I find it hard to leave power on the table with laptops, but I may consider dropping down to the M1 Pro version if the Max's GPU is just too hungry.
When the M1 came out I expected the MBP 14" & 16" SoC to equal HPC desktops.
My worry now is that the very same M1 Pro & M1 Max SoC without a TDP increase will find itself into the iMac 27", iMac Pro & Mac Pro without a TDP change.
Just like how they did the iMac 24".
The only change would be multi SoC configurations.
Would they only deploy the M1 Max the Pro desktops?
Andrei of Anandtech is experiencing unusual screen issues
Considering I use a black background and dark windows, this concerns me. I will wait to see if it's something I notice too much in person tomorrow.
Gruman said that the larger iMacs will come with 14/16c+64c options and for the Mac Pro, 32/40c+128c.
About the TDP change question, I think the M1 families are already operating near the top of the power/performance curves. This has been the case for Apple chips since 2015, when A9X was the last chip from Apple to clock at a significant higher frequency than its phone counterpart (A9 1.8GHz vs. A9X 2.26GHZ) for the Twister cores. Ever since then it is only 5% yearly lift in frequency, more cores, wider cores, and same clocks on the phone chips and iPad chips.
Take a look at Anandtech's A15 review and you will see despite a inter-node improvement, A15's performance cores was only able to run at same frequency as M1's firestorm cores:
https://www.anandtech.com/show/16983/the-apple-a15-soc-performance-review-faster-more-efficient/2
The primary reason behind: For their TSMC N5P chips (A15, M1 Pro/Max) Apple uses low voltage & high density logic and SRAM cell library and that's precisely why you see them cramming 57Bn transistors into a 520mm^2 chip while the similar sized AMD Navi 21, built on TSMC N7, only has 26Bn Txtors. TSMC themselves claim N5 is 1.84x more dense than N7, yet Apple's use of high density libraries make M1 Max 2.2x dense as Navi 21, not to mention an SOC integrating Analog IP etc. is supposed to be less dense than a GPU if compared on the same node.
This also in part gives rise to the fact AMD's N7 CPUs clocking up to 5 GHz, Navi clocks up to 2.5GHz, while Apple's particular architectures clocks at 3.2GHz and 1.2GHz respectively. Wider, denser, lower frequency and power.
https://en.wikichip.org/wiki/5_nm_lithography_process#TSMC
Giving SRAM as an example here: "Two 6T SRAM bitcells were disclosed by TSMC. The high-performance cell is 0.025 µm² while the high-density cell is 0.021 µm²."
Different Top-level design philosophy, different core/IP design, different node and transistor library, different frequency/voltage curves.
Nobody know it Apple will follows up their current high-density low-power design with the 27" iMac and Mac Pro chips; I suspect they will. If they continue to do, then expect 2x and 4x max package power of M1 MAX, that is around 200W and 350W respectively. For the latter, the 2013 Trash Can's 460W power supply should greatly suffice.
Last edited:
The issue with Gruman's forecast is whether he took into consideration how low a volume these Pro desktops are.Gruman said that the larger iMacs will come with 14/16c+64c options and for the Mac Pro, 32/40c+128c.
About the TDP change question, I think the M1 families are already operating near the top of the power/performance curves. This has been the case for Apple chips since 2015, when A9X was the last chip from Apple to clock at a significant higher frequency than its phone counterpart (A9 1.8GHz vs. A9X 2.26GHZ) for the Twister cores. Ever since then it is only 5% yearly lift in frequency, more cores, wider cores, and same clocks on the phone chips and iPad chips.
Take a look at Anandtech's A15 review and you will see despite a inter-node improvement, A15's performance cores was only able to run at same frequency as M1's firestorm cores:
https://www.anandtech.com/show/16983/the-apple-a15-soc-performance-review-faster-more-efficient/2
The primary reason behind: For their TSMC N5P chips (A15, M1 Pro/Max) Apple uses low voltage & high density logic and SRAM cell library and that's precisely why you see them cramming 57Bn transistors into a 520mm^2 chip while the similar sized AMD Navi 21, built on TSMC N7, only has 26Bn Txtors. TSMC themselves claim N5 is 1.84x more dense than N7, yet Apple's use of high density libraries make M1 Max 2.2x dense as Navi 21, not to mention an SOC integrating Analog IP etc. is supposed to be less dense than a GPU if compared on the same node.
This also in part gives rise to the fact AMD's N7 CPUs clocking up to 5 GHz, Navi clocks up to 2.5GHz, while Apple's particular architectures clocks at 3.2GHz and 1.2GHz respectively. Wider, denser, lower frequency and power.
https://en.wikichip.org/wiki/5_nm_lithography_process#TSMC
Giving SRAM as an example here: "Two 6T SRAM bitcells were disclosed by TSMC. The high-performance cell is 0.025 µm² while the high-density cell is 0.021 µm²."
Different Top-level design philosophy, different core/IP design, different node and transistor library, different frequency/voltage curves.
Nobody know it Apple will follows up their current high-density low-power design with the 27" iMac and Mac Pro chips; I suspect they will. If they continue to do, then expect 2x and 4x max package power of M1 MAX, that is around 200W and 350W respectively. For the latter, the 2013 Trash Can's 460W power supply should greatly suffice.
It's very optimistic that they made more than 1 million per year. Will Apple actually create a specific SoC with those number of CPU & GPU cores?
That's why I am think it b more likely a multi SoC config based on the M1 Max would be used with the iMac Pro's 500W PSU or the Mac Pro's 1.4kW PSU.
The idea is to take two or four M1 MAX chips and stitch them together to get those core counts. The core counts are nicely multiples of ten.
And here I am looking at just buying the base model iMac 27" replacement to replace my 2012 iMac 27" Core i7 BTO.
Just hope it does not exceed $2.5k
I expect M1 Pro/Max 27 inch iMacs along with the bigger 20c/40c iMac Pro type desktops. I have the feeling that these are going to come out to more than $2,500 given that you're getting a big screen and all of the MacBook Pro screen goodies.
Yup, that's roughly the idea. Put multiple M1 Max SoC on a logic board to increase CPU & GPU core counts and unified memory.
During the PowerPC days multiple CPUs were placed on the same logic board for a primitive dual core processor.
This would be the only economical way to do it.
Unless of course Apple would instead put the M1 Max die destined for single SoC package and then put the multiple M1 Max dies into a multi SoC package instead?
Last edited:
You are certainly right on this regard. M1 Max's size is already really extreme on N5P; an 16+64 SOC as predicted for the iMac will easily exceed the reticle limit for the manufacturing process and simply be unfeasible.
However, there is currently no clear evidence regarding the existence of high-speed intra-chip connections on the M1 Pro/Max, which is crucial for a multi-processor heterogeneous computer designed The Apple Way. Simply put, it looks like M1 Max can't be easily packaged by two or four to form a bigger chip just yet.
So my prediction would be some large package with CPU&GPU chiplets the AMD Ryzen way. But this makes designing the memory subsystem really difficult, itself being a big selling point of the M1 family architecture. 4 cpu chipsets with 10 cores each and 4 GPU chips with 32 cores each to form a 40+128 config, with memory/IO crammed into another chiplet with 2048 bit LPDDR5??? This would be a nightmare to design with all the cachelines and packaging technologies.
Ahh, sure, there is the Magical, twice-delayed Intel Ponte Vecchio...
https://www.tomshardware.com/features/intel-ponte-vecchio-and-xe-hpc-architecture-built-for-big-data
But is it just me who thinks M1's design topology is too exotic and expensive to scale beyond M1 Max? Like, craaaazy cool for a super low volume halo Mac product but way too expensive to not go into a multi-billion dollar super computer cluster...
You are certainly right on this regard. M1 Max's size is already really extreme on N5P; an 16+64 SOC as predicted for the iMac will easily exceed the reticle limit for the manufacturing process and simply be unfeasible.
However, there is currently no clear evidence regarding the existence of high-speed intra-chip connections on the M1 Pro/Max, which is crucial for a multi-processor heterogeneous computer designed The Apple Way. Simply put, it looks like M1 Max can't be easily packaged by two or four to form a bigger chip just yet.
So my prediction would be some large package with CPU&GPU chiplets the AMD Ryzen way. But this makes designing the memory subsystem really difficult, itself being a big selling point of the M1 family architecture. 4 cpu chipsets with 10 cores each and 4 GPU chips with 32 cores each to form a 40+128 config, with memory/IO crammed into another chiplet with 2048 bit LPDDR5??? This would be a nightmare to design with all the cachelines and packaging technologies.
Ahh, sure, there is the Magical, twice-delayed Intel Ponte Vecchio...
https://www.tomshardware.com/features/intel-ponte-vecchio-and-xe-hpc-architecture-built-for-big-data
But is it just me who thinks M1's design topology is too exotic and expensive to scale beyond M1 Max? Like, craaaazy cool for a super low volume halo Mac product but way too expensive to not go into a multi-billion dollar super computer cluster...
A few months before the M1 there was news of an ARM-based supercomputer running Linux.
Would not be surprised if Apple would transition their data center, cloud, iCloud, App Store, etc to a M1 Max-based server farm.
These operations would benefit the most from the performance per Watt of Apple Silicon.
That would appear to be a surprise to me since the whole point of the leaving of Apple's Chief Architect Gerald William III in 2019, is to form Nuvia where they can actually build Server chips, which Apple allegedly had no plans for.
Key iOS Chip Architect Gerard Williams III Departs Applehttps://www.macrumors.com › 2019/03/30 › gerard-willia...
Now Nuvia is acquired by Qualcomm who failed to commercialize their ARM server business a decade ago... Oops.
Doubtful, Apple is a primarily consumer-focused company with weak enterprise support. They’d have to have a major tonal shift to decide they want to sell servers again.