You are likely to get most accurate results if you close down other applications, make sure that there is nothing running int eh backgrounds (no benchmarks etc.) and disconnect your network for good measure. Basically make sure that the benchmark application is the only one running.
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.
Cinebench R23 with support for Apple M1 SoC is now out. Geekbench scores up too.
- Thread starter EugW
- Start date
- Sort by reaction score
You are likely to get most accurate results if you close down other applications, make sure that there is nothing running int eh backgrounds (no benchmarks etc.) and disconnect your network for good measure. Basically make sure that the benchmark application is the only one running.
16GB
But of corse, it's different hardware. The balance between RAM and computational resources is going to be different.
Thanks. Is there a benchmark that will show the difference between my current 2018 i7 MacMini 32GB RAM vs the new 16GB M1 RAM MacMini, during a RAM intensive task (like pre-rending in After Effects)?
The 2017 27” 5K iMac in my signature got 1025 single core and 3915 multi-core - That is just plain bad in comparison to the M1.
Am i not looking in the right place. I can only seem to find the iGPU score for my Intel Core i7-8700B. Cant see the CPU scores?;
www.cpu-monkey.com
Intel Core i7-8700B - Benchmark, Test and Specs
Intel Core i7-8700B - Benchmark, Geekbench 5, Cinebench R20, Cinebench R23, Cinebench R15 and FP32 iGPU (GFLOPS) benchmark results
The results I posted are not M1. They appear to be A12Z (which would make sense, since M1 machines aren't not out in the wild yet).
No, because no benchmarks of M1 exist. Wait a week or two until reviews are out.
Yup, I get that, see my posts above. I am saying that I am already under what the A12Z did on Cinebench, so the M1 is going to destroy my iMac scores
Just ran the test on my iMac Pro 10 core.
I haven't installed yet, release candidate 2. I'm on the version right before RC. I don't know if that will affect the results significantly, but here goes.
Multi-core: 11303
Single Core: 996
M1 will be faster than the DTK but it looks like we're not on the iMac Pro/ Mac Pro territory yet. Single core we're there but multicore there's some catching up to do. Hope we will get there next year.
I haven't installed yet, release candidate 2. I'm on the version right before RC. I don't know if that will affect the results significantly, but here goes.
Multi-core: 11303
Single Core: 996
M1 will be faster than the DTK but it looks like we're not on the iMac Pro/ Mac Pro territory yet. Single core we're there but multicore there's some catching up to do. Hope we will get there next year.
To be clear, Apple did not claim they were there with multi-core. They only mentioned M1 single-core being fastest, and that's probably true considering A12Z is already almost as fast single-core as yours.
Duh, it's a quad core chip after all. Those things are fast but they are not magical. An 8-core chip mobile however will floor even the current 12-core Xeons.
I believe A12Z is 8 core and not 4 but yeah it's definitely fewer cores than the iMac Pro
Four of those cores are so called efficiency cores which are somewhere between 1/3 and 1/4 performance of the main cores. Those efficiency cores are just there to run non-critical tasks such as backups or email fetching with spending as little power as possible.
In benchmarks like Cinebench, almost all the work is done by the main performance cores (which are 4). The efficiency cores are contributing, but only so much (under 20%). In contrast, Intel can run two hardware threads for each of it's cores.
The Verge referred to the unified memory as DDR4X 4266MHz, but the Finnish apple store page was mentioning HBM2. Personally, hoping for the latter, HBM2 allowed the Radeon VII to dominate same period NV cards on raw compute due to memory bandwidth.
It’s LP-DDR4X. I read somewhere there was some translation issue for the Finnish page.
I also have a Vega 64 10-core iMac Pro, and it is downright sluggish editing some 4k 10-bit 4:2:2 codecs in FCPX. This is a decode bottleneck, not CPU or GPU. It is conceivable the new M1 laptops (basic as they are) could produce better editing performance due to improved video acceleration. This wouldn't be a stretch because the iPad Pro running the LumaFusion NLE can already handle 4k 10-bit HEVC faster than a top-spec iMac Pro.
So there is a lot more to performance than synthetic benchmarks. Fortunately we don't have long to wait. Probably within two weeks there will be extensive 3rd-party tests of the new Apple Silicon Macs -- using real apps not canned benchmarks.
The OS matters too. In benchmarks, Linux boxes often do several % better than windows or macOS with the same hardware.
Even with different cache sizes? Is that possible? Wouldn’t that require a physically different chip? I could be wrong but I thought things like L3 cache (could be wrong here so don’t bite my head off) are a physical difference.
That is an 8th generation Coffee Lake processor.
Why do you think they didn't compare it to the 10th Generation based MacBook pro system that they sell?
Because the MacBook Pro with M1 replaces the 2-port MacBook Pro which has 8th gen CPU. The one with 4-port and 10th gen CPU still sells for a higher price.
Most probably those will be replaced by 14" MacBook Pro expected in 2021.
No. Cinebench R23 was built to support M1 and later. However, as mentioned, no M1 devices are available yet. However A12Z dev kits are available, and as mentioned in the original post, that’s what was benched.
Just ran it on my 6 core 2018 i5 mac mini.
Single: 1046
multi: 5606
Looks like Its faster than the a12z, but the m1 will handily beat that by quite a margin.
Looking at multi-core (systems resulsts already pre-populated, that fastest Tiger Lake get 4904 for multicore - so only a little bit higher than the a12Z
MacBookAir10,1 - Geekbench
Benchmark results for a MacBookAir10,1 with an Apple Silicon processor.
browser.geekbench.com
1687 / 7433
My cinebench R23 scores from 10min default running. (Multicore)
iMac 2020 i7-10700K : 11823
Macbook Pro 16" i9-9880H : 9012
iMac 2020 i7-10700K : 11823
Macbook Pro 16" i9-9880H : 9012
So if performance is proportional...
MBA 2020
geekbench multi : 7433 (65% gain from A12Z)
cinebench R23 (estimated) : 7500ish
Devkit A12Z
geekbench multi native : 4510
cinebench R23 : 4530
Of course, one cannot compare thermally limited passive cooled MBA to a desktop with dedicated cooler.. I gather Mac Mini with M1 will yield better geenbench score, and hence much better cinebench R23 result for Mac Mini is expected.
So it’s obvious why Apple only introduced entry machines as to maximize comparative impact?
Anyway, it’s pretty phenomenal for 1st generation mac chip.
Can’t wait to see a real mac class chip unlike A14Z varient like chip like M1!