Note that M1 Pro would have performed the same but they didn't include it in the chart.
Last edited:
What?I feel sad that Apple will never make a desktop SoC.
he is saying the same thing...those 20/64 SoC is an 2x M1 maxWhat?
They will. Starting with the Mac Pro SoC 40/128 SoC. Apple will most certainly chop that in half and offer a 20/64 SoC.
And they might put that 20/64 SoC into the iMac Pro.
thats the rumor but for the mac pro maybe apple will allow a 1 M1 max not to be limited at 30W for the cpu and 60 for the gpu 3.2ghz cpu freq...so we will never know until 1 year from nowFor Conclusion & Summary
I feel sad that Apple will never make a desktop SoC that has Intel Core i9 or AMD Threadripper TDP.
The best we can hope for is a multi SoC configuration of the M1 Max on the iMac Pro or Mac Pro.
What?
They will. Starting with the Mac Pro 40/128 SoC. Apple will most certainly chop that in half and offer a 20/64 SoC.
And they might put that 20/64 SoC into the iMac Pro.
thats the rumor but for the mac pro maybe apple will allow a 1 M1 max not to be limited at 30W for the cpu and 60 for the gpu 3.2ghz cpu freq...so we will never know until 1 year from now
For those of us who arent that articulate or in the know what does Andrei mean?Also, for those who were stressing over the Geekbench GPU benchmark not being as high as it should:
View attachment 1877822
Andrei is the author of the article.
He means that the benchmarks are too short, so the "subtests" are finished before the GPU workload is high enough for it to hit the 1.29GHz clock speed it is capable of.For those of us who arent that articulate or in the know what does Andrei mean?
For those of us who arent that articulate or in the know what does Andrei mean?
This is it. Anandtech is unlikely to do something more in-depth.I look forward to a more in depth future Anandtech M1 Max & M1 Pro review.
Wallet on standby...continue waiting for the Mac mini with M1 Max!For Conclusion & Summary
I feel sad that Apple will never make a desktop SoC that has Intel Core i9 or AMD Threadripper TDP.
The best we can hope for is a multi SoC configuration of the M1 Max on the iMac Pro or Mac Pro.
He means that the benchmarks are too short, so the "subtests" are finished before the GPU workload is high enough for it to hit the 1.29GHz clock speed it is capable of.
Note that M1 Pro would have performed the same but they didn't include it in the chart.
Wow... that thread title is... really misleading at best...
Yes, the M1 Max (& likely the M1 Pro) achieve higher MT SPEC floating point performance scores than a 5950x and that is incredibly impressive, but that is primarily due to having roughly 10x(!) the available memory bandwidth which allows it to destroy the competition in memory bound portions of the test. It's certainly cool to see, and will likely put the M1 Pro/Max in a class of their own for certain specific tasks, but that kind of performance scaling is not where the majority of apps in the real world are going to land.
Sure. My point wasn't that the M1 Pro/Max (and even M1) aren't floating point monsters (they are, and they're awesome) but that outside of a limited subset of applications, expecting 5950X level MT performance out of an M1 Pro/Max is unrealistic (something that, to be fair, the article itself never claims.)Not only that. M1 simply has more flexible floating point execution. It offers four independent FP ALUs (128bit each) while Zen3 has more limited execution (if I understand it correctly it can execute four FP ops per clock but only under specific circumstances). Since not all code can be vectorized, real world FP workloads do end up doing quite a lot scalar operations and M1 will retire them quicker, especially combined with its huge caches.
This massive fp performance combined with fast access to GPU resources without worrying about copying data back and forth between CPU and GPU could unlock a lot of new algorithms for scientific computing. But yeah, not for everyone/every workload.Sure. My point wasn't that the M1 Pro/Max (and even M1) aren't floating point monsters (they are, and they're awesome) but that outside of a limited subset of applications, expecting 5950X level MT performance out of an M1 Pro/Max is unrealistic (something that, to be fair, the article itself never claims.)
It looks like M1 Pro/Max are worse for playing games. Cant even perform well.
M1 Max TDP is solidly in the desktop territory. Anandtech:For Conclusion & Summary
I feel sad that Apple will never make a desktop SoC that has Intel Core i9 or AMD Threadripper TDP.
The best we can hope for is a multi SoC configuration of the M1 Max on the iMac Pro or Mac Pro.
The laptop they tested it against hit a 220W package power on the same workload.M1 Max TDP is solidly in the desktop territory. Anandtech:
Finally, stressing out both CPU and GPU at the same time, the SoC goes up to 92W package power and 120W wall active power. That’s quite high, and we haven’t tested how long the machine is able to sustain such loads (it’s highly environment dependent), but it very much appears that the chip and platform don’t have any practical power limit, and just uses whatever it needs as long as temperatures are in check.
The entire SOC? Sure, but the only thing comparable today would be the PS5/Xbox Series X/S as the last PC part with a GPU anywhere near this fast would be the 100W TDP i7-8xxxG with Radeon™ RX Vega graphics. The M1 Pro/Max CPU appears to draw ~43W max for the CPU by itself putting it solidly in mobile territory.M1 Max TDP is solidly in the desktop territory. Anandtech:
Finally, stressing out both CPU and GPU at the same time, the SoC goes up to 92W package power and 120W wall active power. That’s quite high, and we haven’t tested how long the machine is able to sustain such loads (it’s highly environment dependent), but it very much appears that the chip and platform don’t have any practical power limit, and just uses whatever it needs as long as temperatures are in check.