Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
You say they do this “often”, but I’ve never seen it.
Are you able to provide a link?
You've never seen it? They do it all the time. Like here:


The M4 chip features a powerful 10-core CPU, an up to 10-core GPU, and support for up to 32GB of unified memory, making the new MacBook Air up to 2x faster than the M1 model.1

M1: https://browser.geekbench.com/v6/cpu/10912973

M4: https://browser.geekbench.com/v6/cpu/10890985
 
Last edited:
  • Like
Reactions: Wando64
I look at it in terms of video playback speed where 1x is normal and 2x is double. In this case, 1.5x is 50% faster and not 150% faster playback.

Keep in mind, I understand mathematically that a 100% improvement is consistent with doubling but when reading Apple's marketing claims I perceive 1.5x like the video playback analogy above.
 
You've never seen it? They do it all the time. Like here:


The M4 chip features a powerful 10-core CPU, an up to 10-core GPU, and support for up to 32GB of unified memory, making the new MacBook Air up to 2x faster than the M1 model.1

M1: https://browser.geekbench.com/v6/cpu/10912973

M4: https://browser.geekbench.com/v6/cpu/10890985
Oh wow, I’d never noticed it before.
Thanks
 
Something is off with this reported result. That benchmark represents a huge drop in performance gains for the M3 Ultra. (GeekBench scores)…

M1 Max (12639) -> Ultra (18368) +45%
M2 Max (14877) -> Ultra (21372) +44%
M3 Max (20949) -> Ultra (27749) +33%

I’d argue that score is on the very low end for the M3 Ultra. Just based off past gains we should be seeing benchmarks around 30,000. Definitely need to see more results.
 
Something is off with this reported result. That benchmark represents a huge drop in performance gains for the M3 Ultra. (GeekBench scores)…

M1 Max (12639) -> Ultra (18368) +45%
M2 Max (14877) -> Ultra (21372) +44%
M3 Max (20949) -> Ultra (27749) +33%

I’d argue that score is on the very low end for the M3 Ultra. Just based off past gains we should be seeing benchmarks around 30,000. Definitely need to see more results.
I thought I saw a score somewhere > 29000 but maybe not. I cannot find it.

However, diminishing returns. MT CPU scores tend to level off as number of cores and performance per core increase.

M1 Ultra - 20 cores
M2 Ultra - 24 cores
M3 Ultra - 32 cores
 
Last edited:
Apple often writes say "2X faster" when it means "1X faster".
something multipied by 1 is always the same thing unless it's a 0.
Maybe Apple shouldn’t build “ultra” chips but should find ways to make the most of the Max versions?
That's exactly what they're doing. Ultra is two Max chips fused together.
 
I look at it in terms of video playback speed where 1x is normal and 2x is double. In this case, 1.5x is 50% faster and not 150% faster playback.

Right, because that UI means “1.5 times as fast”, “1x as fast”, etc. That's the difference.
 
Something is off with this reported result. That benchmark represents a huge drop in performance gains for the M3 Ultra. (GeekBench scores)…

M1 Max (12639) -> Ultra (18368) +45%
M2 Max (14877) -> Ultra (21372) +44%
M3 Max (20949) -> Ultra (27749) +33%

I’d argue that score is on the very low end for the M3 Ultra. Just based off past gains we should be seeing benchmarks around 30,000. Definitely need to see more results.
GeekBench also got a lot of software improvements it is possible that M1 and M2 got this extra +10% from pure software upgrades inside system/benchmark overtime which isn't happening anymore as much.
 
Something is off with this reported result. That benchmark represents a huge drop in performance gains for the M3 Ultra. (GeekBench scores)…

M1 Max (12639) -> Ultra (18368) +45%
M2 Max (14877) -> Ultra (21372) +44%
M3 Max (20949) -> Ultra (27749) +33%

I’d argue that score is on the very low end for the M3 Ultra. Just based off past gains we should be seeing benchmarks around 30,000. Definitely need to see more results.
I have seen a benchmark myself that got a very similar result in the 27000 range. It’s super strange.
What’s even more strange is that metal and open cl scores were below even the m2ultra..
 
Last edited:
I have run a benchmark myself and got a very similar result in the 27000 range. It’s super strange.
What’s even more strange is that er hot metal and open cl scores below the m2ultra!
You already have M3 Ultra?

Edit:
- either you're writing nonsense,
- confusing M3U with something else,
- or you actually really have it, which means you probably shouldn't write it here yet 😁
 
Last edited:
A M4 Max Mac Studio with 128GB memory and 8 TB SSD lists for $5,899.00 and a nano screen Studio Display lists for $1,899 for a cost of $7,798 with no taxes or Apple Care. The 16" M4 Max Mac Book Pro with nano screen, 128GB ram and 8 TB SSD lists for $7349.00 without taxes or Apple care. The additional $$449 over the mobile use any where option with same performance numbers and supports multiple external monitors gets an immobile bigger screen and some extra ports on a non-mobile computer box.

Interesting trade offs....
 
The move to use an M3 ultra in the studio is to differentiate the Mac pro when it launches later this year, plus provide that incentive to spend big $$ on the latest and greatest you can’t acquire any other way. The new Mac pro will indeed have a more powerful chip than the M4 max, but it won’t be the M3 ultra or an M4 ultra based on Apple’s comments about chips recently. It will be based on M4 gen, but expect more cpu and gpu cores than the max and better performance than the m3 ultra. I believe there’ll be another surprise in there too, likely related to ai performance. Let’s wait and see.
 
You already have M3 Ultra?

Edit:
- either you're writing nonsense,
- confusing M3U with something else,
- or you actually really have it, which means you probably shouldn't write it here yet 😁
You’re right, I don’t personally have it. 😇
 
Having read the avalanche of guesstimates and doubting Thomas comments about the M3 Ultra Mac Studio and even though I like the memory and SSD updates, I will await next years Ultra Mac Studio which might have a non-kludged processor. The M3 Ultra seems to be resurrected from the M3 grave and tweaked. The tweaks are not in the original DNA of the chip which is a worry.

There is no verifiable data or information that Apple is actually using this specific chip in their AI computer farm(s).

It is not a burden to use what I already have and is fully paid for. 🤓
 
something multipied by 1 is always the same thing unless it's a 0.

That's exactly what they're doing. Ultra is two Max chips fused together.
That’s their current approach to the Ultra. But it isn’t easy to “fuse” 2 Max chips together. Could they do a half step and find a way to add more cores or boost the speed of the Max cores for a future “Ultra”?
 
😉
What else interesting would a person who theoretically had access to such an M3U notice?
This person sadly hasn’t run real tests like after effects or resolve on it. But in a blender test the new machine beat the m3max with 50s render time over 70s. Which is still not what you would hope for with twice the number of cores..
 
  • Love
Reactions: pksv
I'm sure some people will have a workload which justifies it, but that's gotta be a hard sell for many.
Anyone who does video professionally will benefit from the Ultra variants.

How many people on this planet fall under "video professionals" I do not know, but if YouTube is any indication, with millions and millions of videos per week added, and the global market for niche networks apparently still growing... and then there are major studios... all together it's a considerable number of people.
 
The M-next Max is very close and sometimes outright beating the Ultras.
No, dude, no. The Ultras are used for video production. The M2 Ultra still outperforms the M4 Max for video production (see ArtIsRight's comparisons.)

The M3 Ultra will be the top tool for video producers, for a desktop, for some time.
 
  • Like
Reactions: pksv
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.