Snapdragon 8cx Gen 3 vs Apple M2: ARM-based ThinkPad X13s Geekbench records show generational improvement but still years behind Apple silicon

[Redaktion]

The ARM-based Lenovo ThinkPad X13s is finally available for purchase, but the Windows laptop with a Snapdragon 8cx Gen 3 processor has not managed to catch up with Apple in terms of Geekbench performance. The results for the Snapdragon 8cx Gen 3 are not only well behind those of the Apple M2, but they also lag behind the M1 SoC from 2020.

https://www.notebookcheck.net/Snapdragon-8cx-Gen-3-vs-Apple-M2-ARM-based-ThinkPad-X13s-Geekbench-records-show-generational-improvement-but-still-years-behind-Apple-silicon.629767.0.html

[Sportbike Mike]

I already knew it wouldn't be able to compete with the m1 or m2, so I'm less interested in that comparison. I'd like to know how it compares to intel.  I have a 7cx Chromebook with a realistic 14 hour battery life, but I know that thing doesn't have the power for Windows.  I'd like to know if the latest and greatest Snapdragon does.

[gerdya]

Why is this compared to the M1/M2 and not against other CPUs, which you can find in other fan-less Windows laptops? It is not that going to buy a Mac, when i am looking for a Windows device.

[Marlo]

Isn't the score under Windows (Parallels) of the M1 much lower than 7400? I believe it makes only sense to compare   with other Windows scores. Windows Geekbench scores are typically lower than Linux/MaxOS Geekbench scores - so what you compare here is not really valid.

[AndreaZ]

I can assure that in single core 8cx Gen 3 is faster than an i7 tiger lake-U with turbo boost disabled.

[Jan Onderwater]

Isn't the score under Windows (Parallels) of the M1 much lower than 7400? I believe it makes only sense to compare   with other Windows scores. Windows Geekbench scores are typically lower than Linux/MaxOS Geekbench scores - so what you compare here is not really valid.

When Apple scores better in Geekbench, this is not valid, Geekbench in Worthless etc etc
When Apple scores less in Cinebench, Apple is slow, lags behind, you see, you see

[Marlo]

Isn't the score under Windows (Parallels) of the M1 much lower than 7400? I believe it makes only sense to compare   with other Windows scores. Windows Geekbench scores are typically lower than Linux/MaxOS Geekbench scores - so what you compare here is not really valid.

When Apple scores better in Geekbench, this is not valid, Geekbench in Worthless etc etc
When Apple scores less in Cinebench, Apple is slow, lags behind, you see, you see

My comment has nothing to do with Apple, but it is a fact, that the same machine scores lower in Geekbench under Windows. If we assume this is the case, why would you be against comparing at least Windows vs Windows scores or alternatively MacOS vs Linux scores? The 8CX Gen 3 will still not beat an M1 but the difference is much smaller.

In any case using Cinebench is likewise invalid, because Apple/ARM is at a big disadvantage here, because Cinebench is using the Intel Embree library, which has a NEON implementation just wrapped around SSE/AVX intrinsics.

You might start to understand, that i am trying to be objective here.

[Tridents]

Isn't the score under Windows (Parallels) of the M1 much lower than 7400? I believe it makes only sense to compare   with other Windows scores. Windows Geekbench scores are typically lower than Linux/MaxOS Geekbench scores - so what you compare here is not really valid.

When Apple scores better in Geekbench, this is not valid, Geekbench in Worthless etc etc
When Apple scores less in Cinebench, Apple is slow, lags behind, you see, you see

My comment has nothing to do with Apple, but it is a fact, that the same machine scores lower in Geekbench under Windows. If we assume this is the case, why would you be against comparing at least Windows vs Windows scores or alternatively MacOS vs Linux scores? The 8CX Gen 3 will still not beat an M1 but the difference is much smaller.

In any case using Cinebench is likewise invalid, because Apple/ARM is at a big disadvantage here, because Cinebench is using the Intel Embree library, which has a NEON implementation just wrapped around SSE/AVX intrinsics.

You might start to understand, that i am trying to be objective here.

The difference in Geekbench between macOS and Windows on the same Intel Mac is almost negligible.
The score under Windows (Parallels) on a M1 Mac is affected by running over an hypervisor and not having full access to all cores. Funny enough, even when running windows under Parallels the M1 gets a better score than this new Snapdragon.

[Marlo]

The difference in Geekbench between macOS and Windows on the same Intel Mac is almost negligible.
The score under Windows (Parallels) on a M1 Mac is affected by running over an hypervisor and not having full access to all cores. Funny enough, even when running windows under Parallels the M1 gets a better score than this new Snapdragon.

And yet, if you run the 8CX under Linux the score will be higher - even if Linux is running in a VM - and thus the effective difference to the M1 smaller compared to the numbers presented in this article. Not sure why you are opposing a fair comparison.

[Tridents]

The difference in Geekbench between macOS and Windows on the same Intel Mac is almost negligible.
The score under Windows (Parallels) on a M1 Mac is affected by running over an hypervisor and not having full access to all cores. Funny enough, even when running windows under Parallels the M1 gets a better score than this new Snapdragon.

And yet, if you run the 8CX under Linux the score will be higher - even if Linux is running in a VM - and thus the effective difference to the M1 smaller compared to the numbers presented in this article. Not sure why you are opposing a fair comparison.

Why? "Maybe" because you said that Windows scores are lower than on macOS (not true), and maybe because a fair comparison would imply running the same OS natively (no hypervisor) on each machine. And since you cannot yet run Linux natively on a M1 with full hardware support you are left with running the M1 on macOS, hence a comparison with Windows since Windows and macOS had shown similar results on the same Intel Mac.