The M1 is pretty close to my i7-10700 system which is 8 cores/16 threads so I generally think of one M1 core = 2 i7 cores. 8 AS performance cores would make me very, very happy as it would be double what I have now and what I have now seldom gets saturated. Could I use 12 performance cores? I have the feeling that I wouldn't see any difference between 8 and 12. 4 and 8? Maybe.
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.
Disrupting the desktop market
- Thread starter Bodhitree
- Start date
- Sort by reaction score
16 performance cores would be a massive overkill when you take into account that the quad-core M1 performs very close to the 8-core SMT chips from other manufacturers. Apple doesn’t need hyperthreading.
Extrapolating from M1, a 16-core firestorm config would outperform a 28-core Xeon…
compared to my 1950x, which is 16 core 32 thread, the per core performance is double on the M1, but I use all the cores, and the 1950x despite being very old is still double the M1, but for a lot of things, it is slower due to the slower per core performance.
compare that to its replacement which has a 5950X, per core, at worst is equal to the 4 performance cores, so it destroys the M1.
for what I use it for, the M1 is a great notebook, but it’s not even close to ready to replace my desktop.
if it wasn’t for the bad graphics performance, it would replace my UX581GV
i am really hoping for 8 performance cores and about 4x the GPU performance and it would be my perfect notebook replacement.
but theres a long way to go until they are competitive with a proper workstation.
to replace my desktop, it needs to be double that again.
compare that to its replacement which has a 5950X, per core, at worst is equal to the 4 performance cores, so it destroys the M1.
for what I use it for, the M1 is a great notebook, but it’s not even close to ready to replace my desktop.
if it wasn’t for the bad graphics performance, it would replace my UX581GV
i am really hoping for 8 performance cores and about 4x the GPU performance and it would be my perfect notebook replacement.
but theres a long way to go until they are competitive with a proper workstation.
to replace my desktop, it needs to be double that again.
The 5950x's single core performance is not equal to 4 m1 performance cores, or am I reading your message wrong?
In Geekbench 5 the 5950x at stock clocks leave it's single core performance 50 points below that of the M1. it's multi-core score is of course significantly higher (2.1x as many points) in but its really not that impressive considering it has twice as many cores, quadruple as many threads and ten times the power draw.
The 1950x is also only 33% faster than the M1 in multi-core, not double
At the bottom you said double that again which I read as double 8 performance cores? It wouldn't need that, at least regarding CPU performance, 8 performance cores alone would come very close to matching the 5950x (probably around 14000 GB5 compared to 16500 on the 5950x). These are pretty much our expectations for m1x.
Aggregating across all the known benchmarks, I think one can confidently claim that an Apple Firestorm core running at 3.2Ghz (power consumption of 5 watts) is equivalent performance-wise to a Zen3 running at 4.9ghz (power consumption of 20 watts).
The 5950x has four times as many cores as the M1, since efficiency cores barely count. So it's a 4-core M1 (+maybe 1 core worth of performance from the efficiency cores) vs a 16-core 5950x (+SMT).
I do 3D rendering often with CPU so I'm hoping for as many CPU cores as possible.
Of course I want more GPU power too, the M1 is slow on that front compared to my 2019 iMac.
I'm hoping for 4 efficiency and 12 performance cores and at least 16 GPU cores.
Of course I want more GPU power too, the M1 is slow on that front compared to my 2019 iMac.
I'm hoping for 4 efficiency and 12 performance cores and at least 16 GPU cores.
single core vs single core, and actual performance does not directly relate real world performance differences.
the reality of the multi core results from Geekbench are far from real world to be totally honest, whereas single core i can agree with more.
at the end of the day, I am comparing real hardware against real workloads, and not synthetic benchmarks.
The reason for double again is not actually the CPU performance, but the GPU, unless they add support for discrete graphics cards, the GPU is a very poor performer. Certain workloads such as video it’s incredible at, but for compute or CAD it’s very bad.
when we are talking about wages vs cost of power, unfortunately the inefficiency of x86 and discrete graphics means absolutely zero. the M1 “is” my main notebook, I love it for all the same reasons everyone else does.
which is why i so badly want to see it compete enough I can replace my desktop heaters!
In real workloads (at least ones I care about), M1 can match or even surpass my 16” i9 (a CPU with twice as many cores and 4x higher sustained power consumption). It especially excels at software development, number crunching and database work.
M1 GPU is certainly far performance-wise from large desktop GPUs, but it’s a big step up from the models it replaces and gives the entry-level Apple laptops a big performance boost. We still don’t know it’s performance on the iMac, but I suppose benchmarks will be out soon enough.
As to use case you mention, I would expect M1 to do very good in CAD, since it should be much more efficient at drawing wireframe models and thin triangles than traditional IMR GPUs. Is there any CAD suite that is specifically optimized for Mac to look at? Compute… M1 is not too bad here, but it’s ALUs are definitely memory bandwidth starved for more trivial tasks. Again, are there any real-world Metal-optimized applications that one can use for comparisons? GPU-utilizing photo editors like Pixekmator seem to do great on M1…
Unfortunately you missed the point. Where I am looking to what apple silicon needs to do in the future to replace a workstation, you know, like the market the Mac Pro is aimed at.
not a replacement for XPS13 or intel MBP.
there’s no doubt it smashed everything in its class, the fact it’s comparable to the latest zen CPU’s is amazing, given how bad intel have been looking compared to AMD just makes intel look pathetic now, and given Apple were in intel park, this is a massive step up.
but there is so many things the M1 cannot do, and for it to replace my desktop, they are the numbers I would want to see. This is my workload, not yours. And if people like me didn’t exist products like the Mac Pro wouldn’t exist. I don’t own one because they are kinda crap machines compared to what I can build myself for much less. But let’s assume the M1 scales fairly linearly, and is worth about $500 for the processor and ram, well, I would happily pay the difference and the ‘uncommon product tax’ for such a processor.
12 performance cores and 4 efficient core, 64 GPU cores, 64gb of ram, I’d be happy to pay 5k for a machine wth such a cpu.
That would disrupt the market pretty seriously if it was in a 5k machine. If it was in a 10k machine, well, can build a lot more for a lot less using nvidia/amd.
Apple don’t need to address this market, and I have absolutely no doubt it will not exist this year, but hopefully late 2022 as their final product to cut over being the Mac Pro, this becomes reality.
Ah, you are taking about the Mac Pro. I think we might be pleasantly surprised. But then again, Apple is Apple, so they just might stop at "good enough". The litmus test will be the prosumer chips announced this year — this will give us an idea whether Apple is committed to go the extra mile. I won't expect a new Mac Pro until late 2022 though.
I’m counting on two further processors, one prosumer and one for the Mac Pro. I suspect from the prototype rumours we have seen that there will be a 16-core, 12+4 configuration. We will have to wait and see, but I’d be surprised if the Mac Pro will be later than Spring 2022. Apple said that it would take them 2 years at WWDC 2020, and they are usually pretty good at executing.
But I do think they have to be careful with pricing. Apple has never been cheap, but they have the opportunity to offer value with these new machines in a way that they haven’t done before. This is something they did right with the M1 MacBook Air. It’s an 1100 euro machine, not cheap when the cheapest Windows laptops are 400 euro’s. But it performs like a 1800 euro top-of-the-line Windows business laptop, and that’s what I mean by offering value.
You can say that Apple have always charged a premium for their design, and I think that the 24” iMac is a bit guilty of that. The M1 doesn’t perform well enough as a desktop chip to maintain the computing value argument that the MacBook Air started, it’s a 1600 euro machine which is not more than acceptably fast in its price class. It really needed an 8+4 configuration and a Geekbench 5 multi-core score of around 13,000 at that same price point to hit it out of the park and to make the same argument of computing value relative to machine price.
For prosumer machines I would make the argument that you need to target the best Ryzen processors in multi-core performance, which score about 18,000 in Geekbench 5. That would mean 12 performance and 4 efficiency cores in a 12+4 setup.
But I do think they have to be careful with pricing. Apple has never been cheap, but they have the opportunity to offer value with these new machines in a way that they haven’t done before. This is something they did right with the M1 MacBook Air. It’s an 1100 euro machine, not cheap when the cheapest Windows laptops are 400 euro’s. But it performs like a 1800 euro top-of-the-line Windows business laptop, and that’s what I mean by offering value.
You can say that Apple have always charged a premium for their design, and I think that the 24” iMac is a bit guilty of that. The M1 doesn’t perform well enough as a desktop chip to maintain the computing value argument that the MacBook Air started, it’s a 1600 euro machine which is not more than acceptably fast in its price class. It really needed an 8+4 configuration and a Geekbench 5 multi-core score of around 13,000 at that same price point to hit it out of the park and to make the same argument of computing value relative to machine price.
For prosumer machines I would make the argument that you need to target the best Ryzen processors in multi-core performance, which score about 18,000 in Geekbench 5. That would mean 12 performance and 4 efficiency cores in a 12+4 setup.
Last edited:
I agree about the 4+12 CPU being needed for 16” MBP and 32” iMac.
On the GPU front, the M1 is not as good as the AMD GPU in my 2014 iMac, and it needs to double in performance to catch up to my 2019 iMac. So I’d be looking for at least 16 cores there to put it in the same ballpark.
The larger issue I see going forward is the price-gouging on RAM and SSD.
On the GPU front, the M1 is not as good as the AMD GPU in my 2014 iMac, and it needs to double in performance to catch up to my 2019 iMac. So I’d be looking for at least 16 cores there to put it in the same ballpark.
The larger issue I see going forward is the price-gouging on RAM and SSD.
It’s particularly bad. Then again, with the current cluster**** of shortages unresolved it may soon be on par with prices
Enthusiasts and pros sure are ready to ditch their Ryzen, Threadripper and Epyc for M1X whatever with barely any software, no dGPU support, limit themselves to 16GB RAM instead of up to 8TB, give up ability to multiboot Linux/Windows/BSD/etc. for Big Sur full of subscription nagware and with memory mismanagement that makes 16GB on M1X like 4GB on Linux.
If you can challenge the value proposition within a certain price class by a big enough margin, for example by putting a top-of-the-range Intel Core i9 equivalent in a bottom-end iMac, then yes they will. For the enthousiasts and pro’s it is about computing performance per dollar, and that’s what made the low-end M1 laptops so succesful, they punched well above their price class in speed. And money talks.
It’s just a shame that Apple missed the opportunity to do it again with the iMac, by going for a lower-cost strategy of re-using what they already had and trying to milk the economies of scale. But then by the bright colours they were already signalling this wasn’t really a machine for the pro.
Not particularly. Consumers of the 27-inch iMac version may not be too happy about this new iMac. Perhaps it would be more accurate to say that it disrupted the 21.5-inch Intel iMac market. Which is no disruption at all...
The 21.5” model was a very popular home/office computer, and the M1 model is a great upgrade for that particular market.
The larger prosumer iMac is obviously still waiting for its own prosumer-grade chips.
I agree, it is definitely a great upgrade for these specific consumers and at this price point.
But far from being disruptive. Simply by being a great upgrade aimed at the audience of the 21.5-inch, while not disturbing other markets (such as the consumers of the 27-inch model), it does not cause any disruption.
I think we all agree that it's real workloads that count...unless there are people who only own a computer in order to run benchmarks
The current M1 GPU capability can be said to be a poor performer in absolute terms, when put against powerful dGPUs. Compared to other integrated GPUs it's a great performer, and as you say, it works well very certain graphically intensive jobs for which it has been optimised.
It will be interesting to see how far Apple can push GPU performance both within a single silicon die, and with a separate on-package die (or even separate package). They should be able to match the AMD APUs found in the XBox X and PS5, which are roughly equivalent to something between an NVidia RTX2080 and 2080Ti (or so it is reported).
This would probably be sufficient for a lot of people, but it won't be state-of-the-art. Who knows what the Mac Pro will use? Will Apple develop their own high power general purpose discrete GPU? Or use something similar to the Afterburner cards for specific workloads?
To address your points:
1) We don't know what limitations the next Apple Silicon SoCs will have, but I would be surprised if the machines with "Pro" in their name were limited to 16GB (the next smaller MacBook Pro, could conceivably have this limit)
2) The ability to multi-boot is more of an enthusiast requirement than a professional one these days. I used to do this all the time in order to have Windows & Linux on my Mac; now I just have separate machines (less fuss) and more commonly just run my workloads on cloud service instances (of which I can have as many as I need without rebooting). For better or worse, Apple has decided not to support it at the moment, and make M1 Macs "MacOS" only. Some of it may be technical limitations; or it may be a market position they're taking.
3) If your software is not available on MacOS, then it doesn't matter what capability the machine has; you're simply not going to choose a Mac. The Mac M1s have a lot of offerings in the software categories that I use (productivity, software development, document management, communications). Your needs may be different. If your software isn't supported (e.g. games, CAD etc.) then you will need a different machine.
4) What subscription "nagware" is there in Big Sur? I can't think of any off-hand.
5) Memory management on the M1 Macs seems to be pretty good compared to Intel Macs. I did have some issues with Safari hogging memory, and some cloud storage services seem to be unusually memory hungry (OneDrive, Google Backup & Sync), but overall it's not too bad. I'd agree that most Linux distros appear to be consume far less memory than MacOS or Windows. But again, how much of the software you need runs on Linux? There are a lot of useful utility apps, but not a huge offering of mainstream packages. Davinci Resolve and PixInsight are a couple that I use on Linux.
It sounds like you are really looking for a modular computer, with multiple OS support, on which to experiment with lots of different software.
<Jedi wave> Macs are not the computers you are looking for....
Just a personal opinion.
For me and my workflow I don't really want more cores, I want more singel threaded performance because that is what I would notice the most. Doubling the cores would have very little performance gain in my case.
Apple's current performance core design (Firestorm) outperforms pretty much everything else out there, and part of it is the Apple microarchitecture allows the processing of over double the instructions at the same time as the rest. Soon enough we should see the next iteration of the microarchitecture which will notch probably another 20% gain in performance.