Fake tests or as other mentioned overclocked/nitrogen.
Here’s the official Geekbench list complied/averaged from thousands of tests with outliers removed.
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.
M5 Chip Achieves Impressive Feat in 14-Inch MacBook Pro Speed Test
- Thread starter MacRumors
- Start date
- Sort by reaction score
Fake tests or as other mentioned overclocked/nitrogen.
Here’s the official Geekbench list complied/averaged from thousands of tests with outliers removed.
Metal score is good only for specific apps which is meaningless. Besides, almost all GPU intensive software are CUDA or Nvidia based.
Geekbench is a highly useful/accurate benchmark.
All software is comprised of various basic functions (sorting, compression, encryption, integer/floating point math and so on).
Geekbench does a set of tests using the most common “building blocks” that software developers use and then tests the processor using these tasks to give a final overall performance score.
The only time it would be inaccurate is if you have a very specific application (like rendering) that only did one specific task. In that case you’d want to run a specific benchmark more in line with your workload (like Cinebench or Blender).
For the majority of applications/users Geekbench will be an accurate indicator of system performance
Tell me if M4 series are even close to RTX 40 series, they aren't especially since Apple cant even make RTX 4090 or 5090 grade GPU. Do you even know that Apple advertised that M1 Ultra is as good as RTX 3090 which turns out to be false?
If you think Apple Silicon's GPU is great, you better check Nvidia who is dominating GPU market by more than 90%.
You will be happier with a PC then. I'm sure Jensen awaits your money. If you're still caring about CUDA on Mac ten years since Apple kicked Nvidia to the curb, I would recommend a new hobby.
This only disapproves your point. The truth hurts, huh?
Is this the same Geekbench benchmark that the phones run? My iPhone 17 Pro is pulling 3902 on that benchmark.
Yeah Apple needs to do something about graphics. CPU is impressive and all but a lot of professional workloads benefit massively from strong graphics.
I know the Pro and Max options juice the GPUs but they're tied to the more powerful CPU, driving up total system cost. One wonders why Apple isn't looking at multi-die approaches to chip design like the other big players - hubris?
I know the Pro and Max options juice the GPUs but they're tied to the more powerful CPU, driving up total system cost. One wonders why Apple isn't looking at multi-die approaches to chip design like the other big players - hubris?
That sounds about right. The performance cores are the same as in the M4, so single-core should come out pretty close.
iPhone CPUs are stupid fast.
Not really in the context of this article. All the chips in a series use the same E and P core. So any M4, M4 Pro, M4 Max should have the same single-core performance. Same across the M5 line. It would be a big deal if the base M5 outperforms an M4 Pro or something like that.
Supposedly this is coming with the M5 Pro and M5 Max. I guess you haven't seen the rumors. https://www.tweaktown.com/news/1081...ocks-for-unique-soc-configurations/index.html
Instead of two chips in separate areas on the board, it's going to be 2 independently made chips for GPU and CPU which are then horizontally joined into one larger module. The same thing Wii U was doing in 2012.
Oh, great, so basically what AMD and Intel have been doing for a little while now. Glad Apple isn't too far behind here. This'll be especially important if Apple wants to take gaming seriously like they always claim they do.
The truth is Mac lacks GPU intensive software and now, Mac specific apps such as FCPX are joke. All GPU intensive software are CUDA and Nvidia based/friendly. It's still silly to say Mac's GPU is great while Mac lacks GPU intensive software.
Apple does not make their own processors. The choice is ship them as TSMC has them ready in quantity, or ship nothing until they are all ready. Other than your personal esthetics why is the second option better?
There are LOTS of erroneous results on Geekbench on the top list, so it takes a bit of mental filtering when comparing processors. For example, you'll find one dude in China with an Android rocking 5000 in the single core... is this real, no. Just getting your name on the 'high score' like an old arcade game by faking values. Or someone in a lab nitro-cooling a chip to see how much voltage they can shove into it before it blows (which it likely became a crisp two seconds after completing the test).
An easy way to tell if something is real is just look at the amount of results with similar values of a specific chip. Then you'll see the real speed of a chip. The M5 is the highest single for consumer chips.
Running a 5090 in a laptop or a small form factor desktop is also joking. If you believe that Apple should make their own 5090 equivalent for the Mac Pro that's a perfectly valid argument. Now you are down to a market demand argument which is not the one Apple is making.
It almost feels like overkill to put that much power in an iPad.
Just ran GB6 on my M1 Max:
SC: 2322
MC: 12183
Looks like the M6 series will get us to full 2x on single core. M5 is close to 2x but not quite. (We're already there on MC of course).
Thank you. Wowzers!There are actually more results than just this, including GPU and AI results.
Metal GPU: 76727 vs ~56989 for M4 (just picked a random representative result)
CoreML GPU: 13172 (single precision), 24682 (half precision), 23672 (quantized) vs 8408, 10078, 9166 for M4
Very nice improvements!
Mac17,2 - Geekbench
Benchmark results for a Mac17,2 with an Apple M5 processor.browser.geekbench.com
Mac17,2 - Geekbench
Benchmark results for a Mac17,2 with an Apple M5 processor.
So the M5 has a significantly faster CPU and GPU than the M1 Pro. Time to upgrade!
Is dominating Gpu market but Apple cannot do that since they do gpu just for themselves while nvidia did it even for apple and the other OEM, or maybe you are talking about strictly performance wise, then its true
Apple can't make the GPUs or hasn't because Apple also values efficiency?
"Do you even know that Apple advertised that M1 Ultra is as good as RTX 3090 which turns out to be false?"
Can you link to those ads?
Apple's marketing was that it matched higher end RTX GPUs at a certain Wattage level, not that it was as fast as one without any constraints in power.
Further, limitations in M1 Ultra performance are also tied to optimization. Powerful hardware will run poorly optimized software poorly.
"If you think Apple Silicon's GPU is great, you better check Nvidia who is dominating GPU market by more than 90%."
Market share is necessarily a metric of quality and performance? By that logic Windows is vastly better than macOS and Linux.
Apple's GPUs are phenomenal. How do Nvidia's integrated options stack up?
I think in ray tracing the M5 is faster than M2 ultra, but i need to test...for me these are just numbers until the M5 hit the customer market and test in apps that i'm interested personally
Awkward math. Couldn't you simply divide low by high? 3784/4263 = 0,888 thusly ≈ 12.5%