I went for the Mac Studio M1 Ultra with the 64-core GPU option but I'm now sort of regretting it. I'm upgrading to the Mac Studio from a 2013 Mac Pro (Trashcan). My main usage will be for Photo/Video work and I edit Blackmagic 6K BRAW and RED 8K footage often. How much of a difference should I see vs the base M1 Ultra 48-core GPU version? If its only going to be like 5-10% better then I think I'm going to cancel. It just sucks that the backorder is now approximately 3 months long.
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.
Difference between the Mac Studio M1 Ultra 48-core GPU and 64-core GPU options
- Thread starter ChromeHeart
- Start date
- Sort by reaction score
We will see when the tests are out - but I just ordered the 48 core version. I presume the difference is not all that realistically.
As far as I understand the price difference is because of a couple of things: The 64 core version is much harder to make since it needs to be a perfect assembled chip, whereas the 48 core version is the same but it isn't that necessary to have the same, extremely hig, standard in the soldering. I would think that the 48 core versions are the ones that are refused at 64 cores, and there is a sort of limiter that makes 48 cores work, not more. The second is that Apple charges that much more for the 64 core version just because they can.
As far as I understand the price difference is because of a couple of things: The 64 core version is much harder to make since it needs to be a perfect assembled chip, whereas the 48 core version is the same but it isn't that necessary to have the same, extremely hig, standard in the soldering. I would think that the 48 core versions are the ones that are refused at 64 cores, and there is a sort of limiter that makes 48 cores work, not more. The second is that Apple charges that much more for the 64 core version just because they can.
I think that for your work, pushing around RAW video files, probably better to err on the side of more, especially if you are working with multiple video streams. If not, and this is just a guess, the 48 would be fine.
Here's the main thing: don't cancel your order. Place ANOTHER order for the lower spec, then test out the first one and if you like it, keep it and cancel the other order. If not, you have 30 days to return it and accept delivery of the 48.
I ordered the Ultra, which may be overkill for the 4k editing and encoding I do, but it was too hard to resist.l
That’s online. Retail has a separate queue. So, if you live within driving distance of an Apple Store, you will find it in stock much sooner than that. High-demand products sell out quickly, though, so you might need to keep checking back each day.
Apple claims that the M1 Ultra GPU is 80% faster than the MacPro with a Radeon 6900 XT.
Geekbench shows this is not true! Or am I wrong? Even with my old 5,1 and a decent Radeon RX 580 I get 48800 for Metal Score.
Here the Open CL scores of the 48 core and the 64 core vs the last iMac Intel 27 inch with ATI Radeon 6900 XT.
M1 Ultra 48 cores: 77164
M1 Ultra 64 cores: 85800
Imac with Radeon 6900 XT: 134378 !!!
M1 Ultra 48 cores: 77164
M1 Ultra 64 cores: 85800
Imac with Radeon 6900 XT: 134378 !!!
Well are you using opencl?
Here is gfxbench and shows ultra ahead of 6900 xt. Don’t know if this is 48 or 64 core ultra
GFXBench - Unified cross-platform 3D graphics benchmark database
The first unified cross-platform 3D graphics benchmark database for comparing Android, iOS, Windows 8, Windows Phone 8 and Windows RT capable devices based on graphics processing power.
Interesting. The Geekbench 5 testing showed differently, but it wasn't a Metal test.. Time will tell.
I posted the metal test before. Same experience.
mac pro 178000
mac studio 103000
significantly lower
Yeah I couldn’t have seen the Ultra beating high end discrete cards outside of Apples contrived graphs.I posted the metal test before. Same experience.
mac pro 178000
mac studio 103000
significantly lower
I am not sure if I want to rely on those scores just yet as I am not sure if they are real. Some list core counts for the GPU others don’t, something sure seems a bit fishy.