Has anyone got to run Cinebench 2024 or 2026 yet and compared it to Tahoe / Sequoia results?
I'm really curious since this is now an ARM only OS.
I'm really curious since this is now an ARM only OS.
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.
MacOS 27 Beta CPU performance?
- Thread starter greenmike
- Start date
- Sort by reaction score
kharys415
macrumors member
Just tried Cinebench 2024 in multicore on M4. 948 points, last run on 26.6 beta ended at 961, so about the same.
Games are running worse than they did in 26.6 though.
Also tested the GPU render. 3840 points vs 3719 (26.6), this is for 8-core M4 GPU.
Games are running worse than they did in 26.6 though.
Also tested the GPU render. 3840 points vs 3719 (26.6), this is for 8-core M4 GPU.
Last edited:
Thank you, that's very insightful. Welp, I expect something like a 10% jump but I guess Apple Silicone was already well optimized in their dual architecture OSs.
I guess I'll stick with Sequoia then still. Definitely not going to switch to Tahoe.
in the end this is their first developer beta and improvements will most likely span across all devices in coming weeks
Anything improves over time, usually, with continuous refinement, but I specifically referred to now the first MacOS without containing Intel Mac code.
I'm just saying that I expected more from the get go. I guess they already optimized it well enough.
The improvements might be noticeable when you are running many different heavy processes. A single process running 10 threads it's not too hard to schedule.
You can expect early betas to have debugging code. If your benchmarking is that beta is similar to latest Tahoe - then that is a good sign. I would wait until August releases for meaningful benchmarking comparison.
From WWDC, the most obvious performance change is (will be) in app activation. I have seen nothing to suggest significant change in CPU or GPU intensive tasks.
Yeah. I was wondering how much debugging code is in the betas. When I was a kernel developer we certainly had plenty of sanity checks (assert statements) in the code that were stripped out in later builds and testing those assertions definitely adds at least some overhead at quite low levels in the kernel.
I'm also wondering at what point in the beta cycle all debugging code is removed. Clearly (I think) it would be gone when they get to RC1 since a release candidate is intended to be bit-identical to what would be released if that candidate is deemed good enough to actually release but as an ex kernel developer I think I'd prefer to remove all debugging code at least a beta or two before RC1 because I'd rather see any race conditions early rather than right at the end of a beta/RC cycle and debugging code can potentially hide race conditions.
Probably too late unless there is applicable data that could be found from posts made about this time last year but a more valid comparison would probably be how this beta compares with the first developer beta of Tahoe since that would at least to some extent control for debug code in the first developer beta.
Developers have had 6 years to port all of their apps over to apple silicon. Rosetta 2 and the development of emulation software has made it easier one way or another to transfer code from intel to apple silicon.
keksikuningas
macrumors 6502
kharys415
macrumors member
Native, NTE and Control specifically. For x86, I'm playing Mass Effect 2 with Crossover right now and it seems good but it's not a difficult game to run.