For Blender workload, the M3 just outright beats the M2 Ultra with all the GPU advancements and ray-tracing. Not everything is about video editing. You would think Apple's best of the best would be.....best.
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.
M3 Ultra Chip is Only 10% Faster Than M4 Max in Benchmark Results
- Thread starter MacRumors
- Start date
- Sort by reaction score
Yes, that’s true, with each additional core that’s added, there’s a performance strain, so you will generally never see a linear path. A lot of that is usually due to needing to lower the clock speed of the cores to stay under a targeted thermal ceiling. Apple’s cores though are extremely efficient and they shouldn’t need to throttle the clock speed.
There’s also the case that p-core to e-core ratio is much higher in the M3 (3:1), whereas the M2 was 2:1. That makes this lower score even more “suspect”. there are 8 more p-cores in the M3 Ultra. A doubling of cores for a 33% gain just doesn’t make sense.
Anyway, just saying.
Not everything about Blender either. If you look at the Artisright's comparisons M2U is still at the top of the list in most various creative software Only the most powerful M4M specs started to overtake it. That's the problem with such a drastic increase in performance from generation to generation, that mid chips will jump the top tiers faster than some would like.
Probably very few people should buy Ultra chips. Only those who absolutely know that such a machine will pay for itself in a year. Most will definitely get a better deal when buying Studio with Max chip.
The problem here is that with chips like Intel and AMD the top of the line desktop chip was the best solution for everything compared to what we have with Apple. I fully expect the M5 Max to be better than the M3 Ultra in several situations.
Comparing the Geekbench results it looks like the M3 Ultra with 80-core GPU is the only one that's faster than Apple's previous fastest standalone GPU, the AMD Radeon PRO W6900X, with scores ranging 93% to 126% as fast. It looks like the M4 Max GPUs are about 78% and 79% as fast, respectively.
So the GPUs on the unbinned M3 Ultra are about 13%-40% faster than the M4 Max?
Just out of curiosity what Nvidia card would an M4 Max 16core and an M4 Ultra 32core be roughly equivalent to?
M3 Ultra, M4 Max will be around 3070 (or around PS5 Pro GPU performance)
Wow the ThreadRipper packs a nice punch in these tests! Power consumption much higher of course.Here’s a couple test results from 3D rendering apps that corroborate my guess of up to 50% better performance on the M3 Ultra vs M4 Max. From Tom’s Hardware
Not sure where you're getting your numbers from, but the official user generated Blender benchmarks are very good real world approximations for AS M series graphics power.
The M4 Max 16/40 is not too far behind an RTX 4080 and higher than a AMD Radeon RX 7900 XTX. So massively better performance than the 3070 which scores 2202.47
The M3 Ultra 32/80 is almost equivalent to an RTX 4090. I expect the 28/60 to be equivalent to a RTX 5000.
The PS5 you refer to –and fun fact, also the Xbox Series X– both use a similar AMD GPU to the RX 6800 XT which scores 1966.09 in the same Blender benchmarks.
Blender - Open Data
Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
opendata.blender.org
EDIT: Just noticed you actually mentioned a PS5 Pro, which looks like the equivalent to a RTX 4070 [source: https://www.pcguide.com/gpu/what-gpu-beats-ps5/] which scores 3435.67 in the Blender benchmarks.
Last edited: