Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.


macrumors bot
Original poster
Apr 12, 2001

Primate Labs today announced the release of Geekbench 5, the latest major version of its popular benchmark software.


For CPUs, Geekbench 5 features new benchmark tests and it also increases the memory footprint of existing workloads to more accurately account for the effect memory performance has on CPU performance:
The Geekbench 5 CPU Benchmark includes new benchmark tests that model the challenges your system faces when running the latest applications. These tests use cutting-edge technologies, including machine learning, augmented reality, and computational photography.

Geekbench 5 also increases the memory footprint of existing workloads to more accurately account for the effect memory performance has on CPU performance.

Finally, the Geekbench 5 CPU Benchmark includes new modes of multi-threaded benchmarks, allowing threads to work co-operatively on one problem rather than separately on different problems. With the addition of different threading models, Geekbench 5 better captures the performance of different multi-threaded applications on personal computing devices.
As for GPUs, the Compute benchmark now supports Vulkan in addition to Metal, CUDA, and OpenCL.

Geekbench 5 also has a refreshed user interface with full support for Dark Mode on macOS Mojave and later. Support for Dark Mode on iOS 13 will be available later this year, according to Primate Labs.

Geekbench 5 is available now for macOS, iOS, Windows, and Linux, with an Android version coming later this week. The software is 64-bit only, dropping support for 32-bit processors and operating systems.

Through September 10, Geekbench 5 and Geekbench 5 Pro for Mac are on sale for $7.49 and $49.99 respectively, down from $14.99 and $99.99. Geekbench 5 for iOS is also free until that time, while Geekbench 5 Pro for iOS has introductory pricing of $4.99, which will rise to $9.99 after the sale ends.

Article Link: Geekbench 5 Released With Improved Benchmark Tests, Dark Mode Support, and More
  • Like
Reactions: mudflap and mrjay


macrumors 6502
Apr 2, 2016
the only benchmark I really care about is FPS in Keyshot, camera demo... I always want Keyshot to be faster!


macrumors 6502
Mar 31, 2004
Tampa, FL
Does anyone really base their purchases off a benchmark?

Lets see some tests for what people really are waiting for, a better, longer lasting battery.

It's useful in-so-far as to compare two machines within the bubble that the test itself represents. Generation-over-generation performance improvement being one of the most common.

Ultimately, however, the individual's use cases have to be called into consideration, because a user with a primary application of probably wouldn't see much difference between a dual-core machine and a 20-core machine.


macrumors 6502a
Jan 5, 2017
I haven't tested it yet, but I trust the benchmark times will be longer? That was growing to be the primary concern with 4, it was over so fast everything would run it in boost/pre-throttling.

Looks like it brings the A12 even more above Android SoCs with the harder test from what I've seen so far.


macrumors regular
Sep 3, 2019
I think they have butchered the scores or something.
iPhone XR scores 1116 single and 2556 multi core. Not free or $1? Too bad.

Edit) It socred 1111/2563 on second run. Each test took 2min 30sec.
I am facing this similar problem too. Unusually low benchmarks.


macrumors 6502a
Jun 11, 2013
I think they have butchered the scores or something.
Yes, I think they did. I tested both 4 & 5 on my 6s and they yielded different results


  • 7D5F1F44-80E4-409D-BC9C-FAE83555D729.png
    91.2 KB · Views: 246
  • C03BC22F-01EF-406F-940F-D3F40D684CF3.png
    86.2 KB · Views: 219


macrumors 68000
Nov 5, 2004
With each version of the software, they choose a baseline machine and set the score to some number. Your device scores relative to that baseline device. With Geekbench 4, it was a Microsoft Surface Book with an Intel Core i7-6600U and the score was 4000. If you scored 2000, it was approximately half the processing power as that Surface Book. Likewise, 8000 would be twice as much computational horsepower.

Geekbench 5 scores are calibrated against a baseline score of 1000 (which is the score of an Intel Core i3-8100). Higher scores are better, with double the score indicating double the performance.

I use Geekbench to quickly relate potential performance between machines. For example, I see Apple has a new MacBook Air with a dual-core 1.6GHz 8th generation CPU. Well, how does that compare to my 2012 MacBook Pro with a 2.5GHz quad-core processor. With the Geekbench results, I can get a quick idea of the difference.


macrumors 68000
Jul 24, 2004
Benchmarks are for fake techies only. Benchmarks do little to predict actual real world performance where hardware and software interact in a myriad of ways to affect performance. Recent benchmarks have shown Apple’s hardware to be superior in many tests to the competition but that is meaningless too. What the user experiences when they are using the system is what counts. A benchmark gives a false sense of performance to anyone addicted to those tests. The tests show performance so they believe they are experiencing it, even if it’s just a few milliseconds. Human nature.


macrumors 65816
Jun 6, 2017
United States
It's nearly 2020 and, yet, peoples and it hardware websites still using this fake geekbench benchmark as reference! Unbelievable!

Why? Because it doesn’t show your phone at the top? What other benchmark works on multiple platforms and allows you to compare scores between them that is also better?
  • Like
Reactions: Precursor


macrumors newbie
Sep 3, 2019
the only benchmark I really care about is FPS in Keyshot, camera demo... I always want Keyshot to be faster!

As I tested now, with Geekbanch 5 it is even harder to see the difference between processor speed for rendering.

an i9 9900k gets 8000score in Geekbench 5
my TR 2970WX gets around 12000score, but in real life renders the AMD is 2x faster than the i9, cinebench R20 shows this too. 4900score vs 10k+


Oct 6, 2005
Never mind. Weird that’s version 4 is the first result. I had to scroll down quite a bit to find 5. It’s too early. Lol.


  • 07B17F17-12F4-44BB-A0F1-DD69BF02A2EC.jpeg
    398.2 KB · Views: 204
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.