Geekbench 5

[acnomad]

I just returned a mini with 8GB RAM and 512GB SSD in favor of one with 16GB RAM and 2TB SSD, primarily because I needed the larger local storage capacity. Surprisingly, the Geekbench results of the machine with 16GB RAM were slightly worse than the one with 8GB. Any thoughts?

 

[jdb8167]

I just returned a mini with 8GB RAM and 512GB SSD in favor of one with 16GB RAM and 2TB SSD, primarily because I needed the larger local storage capacity. Surprisingly, the Geekbench results of the machine with 16GB RAM were slightly worse than the one with 8GB. Any thoughts?

CPU lottery. What were the numbers?

 

[acnomad]

CPU lottery. What were the numbers?

M1 8GB: Single Core 1747, Multicore 7656
M2 16GB: 1302, 6022

Possibly worth noting that Geekbench 5 recognized the 8GB chip as Apple M1 3197 MHz (8 cores), whereas it thought the 16GB chip was a VirtualApple 2400 MHz (8 cores)

 

[ArPe]

M1 8GB: Single Core 1747, Multicore 7656
M2 16GB: 1302, 6022

Possibly worth noting that Geekbench 5 recognized the 8GB chip as Apple M1 3197 MHz (8 cores), whereas it thought the 16GB chip was a VirtualApple 2400 MHz (8 cores)

App bug, launch again or update it. Has nothing to do with the hardware.

 

[casperes1996]

M1 8GB: Single Core 1747, Multicore 7656
M2 16GB: 1302, 6022

Possibly worth noting that Geekbench 5 recognized the 8GB chip as Apple M1 3197 MHz (8 cores), whereas it thought the 16GB chip was a VirtualApple 2400 MHz (8 cores)

In that case your benchmark on the 16GB model was run under Rosetta 2 translation, running the Intel binary, and not the native Apple Silicon version of Geekbench

 

[acnomad]

App bug, launch again or update it. Has nothing to do with the hardware.

In that case your benchmark on the 16GB model was run under Rosetta 2 translation, running the Intel binary, and not the native Apple Silicon version of Geekbench

I reinstalled Geekbench 5 and ran it again. Results: Single core 1740, multi-core 7623. About the same as the 8GB RAM machine. I thought it would be faster...

 

[CMMChris]

Why should it? Geekbench doesn't take RAM size into account.

 

[jdb8167]

I reinstalled Geekbench 5 and ran it again. Results: Single core 1740, multi-core 7623. About the same as the 8GB RAM machine. I thought it would be faster...

RAM size has no affect on Geekbench. The benchmarks it uses are relatively small and will fit easily on an 8GB machine.

 

[casperes1996]

I reinstalled Geekbench 5 and ran it again. Results: Single core 1740, multi-core 7623. About the same as the 8GB RAM machine. I thought it would be faster...

There would be no reason to assume it would be any faster.

If you have three items you regularly use and a table that can hold three items, it does not make you work faster to get a table that can hold 5 items.
However, if you regularly use 5 items, your table that can only hold 3 items might result in you needing to go to your shed to cycle items in and out with the remaining 2; Then a bigger table would make you work faster, but only then.

Bit of a strained analogy but hope it makes sense.

macOS will use extra unused memory to cache files which will still give small speedups for some thing, but it's negligible, and memory you're not actively using generally doesn't add anything to you. However not having enough memory can severely slow you down

 

[acnomad]

macOS will use extra unused memory to cache files which will still give small speedups for some thing, but it's negligible, and memory you're not actively using generally doesn't add anything to you. However not having enough memory can severely slow you down

Thanks for the analogy. That does indeed make sense, and reminds me of a time that I once understood this sort of thing better. When 64KB of RAM in an Apple IIe was considered "industry standard."

 

[casperes1996]

Thanks for the analogy. That does indeed make sense, and reminds me of a time that I once understood this sort of thing better. When 64KB of RAM in an Apple IIe was considered "industry standard."

You're welcome
I have a lot of respect for the developers who had to fit their programs in that amount of space. A compiler I did recently is just under 2.5MB and that's just an executable. Though granted a decent chunk of that is runtimes and dependencies outside of my control that likely wouldn't be that big back then, but hey still.