App Store
The App Store gives people around the world a safe and trusted place to discover apps that meet our high standards for privacy, security, and content.
www.apple.com
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.
M1 Ultra Outperforms 28-Core Intel Mac Pro in First Leaked Benchmark
- Thread starter MacRumors
- Start date
- Sort by reaction score
I was expecting this. M1 scaled up to be comparable to the fastest multi-core x86 CPU, at a decent price and with way lower TDP probably.
Also, this is roughly 2X as fast as the M1 Max, which has half the cores. So a linear scale-up. Nice.
Also, this is roughly 2X as fast as the M1 Max, which has half the cores. So a linear scale-up. Nice.
Last edited:
Wicked expensive as well, $10k in 1990...!
Dash 30fx, a IIfx on Steroids
A company known as 68000 repackaged the Mac IIfx logic board in a huge, heavy metal case, overclocked the board to 50 or 55 MHz, replaced the 40 MHz CPU and FPU with the 50 MHz version, installed even faster RAM, and turned the “wicked fast” IIfx into something 25-35% faster!Agreed, I with in HPC and can’t wait to get my hands on one
Meh, GeekBench is effectively a useless benchmark. I fail to see why it's so widely used. I'll wait for some real-world benchmarks (though I expect the Ultra to do better in many of them)
Why is it useless? It tells you how fast the CPU is by itself, with no memory sharing. Real-world tests will be exercising multiple components together, but that's only useful if they happen to test your workload, at which point you might as well just try it yourself.
How about the M1 Galactus for the quad?
It's useless for all the reasons that followed your second sentence.
Most things? Certainly not ”most” software that is still supported and sold. Unless you are rocking a copy of CS6 or something, most of the best selling and most commonly used software is now Apple Silicon native. Still some holdouts, of course, and particularly for certain specialty workflows you may have to wait awhile.
Just when we thought wait the new M1 max wasn’t fast enough for our every day needs, it’s cheaper buying a 14” MacBook Pro with M1 max and use it in docking mode with a monitor that way you can a laptop as well, the speed of the M1 max MacBook be max studio is identical but one is portable and one is hard wired
We don’t actually know that the speed is identical until we get more testing results. In particular, the thermal solution on the max studio appears to be beefier than that on the 14” MBP, so you may see less thermal throttling (though there’s very little with the MBP as it is). In any event, these will sell very well for compute farms, business environments where it isn’t desirable for computers to get up and take a walk, etc.
According to geekbench, there is only 1 chip faster, and without the neural engine, graphics and encoder/decoder, and memory bandwidth
And threadripper 3990x with 64 cores
And threadripper 3990x with 64 cores
True for the m1 max, but you said not fast enough, the m1 ultra is 2x. Sorry if I misunderstood your post
It’s actually even worse for Intel because 4 of the 20 M1 cores are actually the smaller efficiency cores, IIRC.
Worse than usesless, it's misleading.
Most benchmarks are merely useless because they're generally measuring "turbo speed".
A fair benchmark would "warm up" for 2 minutes then run for 10 minutes and see how many loops (of slightly modified operations to prevent RAM cache cheating seen in FPS game benchmarks) the benchmark can perform.
I don't use Premiere every day but when I use it, I use it all day long. Any time I take a break from rendering, the first couple minutes are always faster than the next couple of hours because the machine heats up. Yet almost every benchmark I've ever seen completes within 5-90 seconds, rendering the whole simulation rather pointless.
Best yet, give me a benchmark that warms up for 30 minutes and then renders from minutes 31-60. That'd be more accurate and useful. The difference between 5 and 10 seconds being "twice as fast" is very little compared to something that takes 2 hours vs 4 hours or worse, 20 hours vs 40 hours.
GeekBench is just a glorified pissing contest.
Apps will come, because these computers give developers what they need. I think Apple finally realized that intel was playing them and limiting their development to keep their PC customers happy. They sell more chips to them. Now Apple can extend their capabilities to greater heights without the bottle neck that was intel.
When it comes to Apps one thing holds true, all Apple has to do is target an area that is being neglected and release 1st party software to address it and suddenly better software will be available. People always like to say Apple needs competition. Not really, they are generally competing with themselves to be better and better.
They introduce competition on the software side which crates the chase. Music production software was being neglected, they bought emagic and made logic. Suddenly all the players were scrambling to upgrade software they had neglected for years. Microsoft rarely updated Office for Mac until Apple surprised them with Pages, Numbers, and Keynote. Google had every excuse for keeping turn by turn off the iPhone for years while they promoted Android Phones as the alternative to iPhone and that feature as a signature difference. Until Apple surprised them with their completely in-house built Maps and stopped depending on them to update such an important app. The same week Google announce Google Maps for IPhone would have Turn by Turn. It still took them months to do it, but it came.
The thing about apps not being available, it's more about the developers not wanting to put money towards development costs. Any developer is able to reach out to Apple and ask for assistance in getting their apps to run native on Apple silicon.
I’m so split in what to think about this new machine; I would rather have my MacBook pro in dock mode with a monitor
Whut??
Apple's benchmarked their current processor against 3 years old processor and got better results?? OMG that's great. I bet it beats current lineup of quantum computers too.
Seriously, why not compare them against 3345, 3365 or 3375? Now, that might have been more interesting. Because anyone buying a workstation will not buy new intel chips, right? Wrong...
As for GPUs - they should compare these against rescent GPUs that matter at this price point... this is getting bizarre.
I am still laughing at their joke from previous event, when they've said that m1max was beating 3080.
Apple's benchmarked their current processor against 3 years old processor and got better results?? OMG that's great. I bet it beats current lineup of quantum computers too.
Seriously, why not compare them against 3345, 3365 or 3375? Now, that might have been more interesting. Because anyone buying a workstation will not buy new intel chips, right? Wrong...
As for GPUs - they should compare these against rescent GPUs that matter at this price point... this is getting bizarre.
I am still laughing at their joke from previous event, when they've said that m1max was beating 3080.
Let’s be real, if you’re in that type of environment they are probably using Windows. And if it’s a huge corporation (e.g. government) they probably have a Microsoft contract. I’ve been in government for over 20yrs working with various companies such as Raytheon, Harris, etc and have seen zero instances of MacOS being used.
hmmm, what is the bandwidth? > 800GB/s memory bandwidth? why not wait for gen 5. You do know that the only CPU that beats this beast (according to Geekbench) is the AMD 3990x with 64 cores, and not by much? Or are you saying that Apple didn't build enough peripheral bandwidth to accommodate Thunderbolt drives, up to 5 monitors, etc?