[nerdsniper]

I do believe Apple are still the fastest single-core (M5, A19 Pro, and M3 Ultra leading), which still matters for a shocking amount of my workloads. But only the M5 has any noticeable gap vs Intel (~16%). Also the rankings are a bit gamed because AMD and Intel put out a LOT of SKU's that are nearly the same product, so whenever they're "winning" on a benchmark they take up a bunch of slots right next to eachother even though they're all basically the exact same chip.

Also, all the top nearly 50 multi-core benchmarks are taken up by Epyc and Xeon chips. For desktop/laptop chips that aren't Threadripper, Apple still leads with the M3 Ultra 32-core in multi-core passmark benchmark. The usual caveats of benchmarks not being representative of any actual workload still apply, of course.

And Apple does lag behind in multi-core benchmarks for laptop chips - The M3 Ultra is not offered in a laptop form-factor, but it does beat every AMD/Intel laptop chip as well in multicore benchmarks.

[deaddodo]

No, the AMD headliners still dominate for single-core performance[1]. Even if you normalize for similar/"same" chips; which really just means you have five cores each generation: AMD's, Intel's, Apple's, and ARM Cortex-A and Cortex-X.

Obviously it's an Apple-to-Oranges (pardon the pun) comparison since the AMD options don't need to care about the power envelope nearly as much; and the comparison gets more equal when normalizing for Apple's optimized domain (power efficiency), but the high-end AMD laptop chips still edge it out.

But then this turns into some sort of religious war, where people want to assume that their "god" should win at everything. It's not, the Apple chips are great; amazing even, when considering they're powering laptops/phones for 10+ hours at a time in smaller chassis than their competitors. But they still have to give in certain metrics to hit that envelope.

1 - https://thepcbottleneckcalculator.com/cpu-benchmarks-2026/

[nerdsniper]

I can't find which benchmarks those scores are from. It looks like sometimes they might have been comparing gaming FPS to AMDs paired with Nvidia 5090's? Something feels off about the site you linked - the methodology and scores aren't even cursorily explained, and gaming scores don't make sense. The 5600X doesn't even have an iGPU and the GFX card they had to have paired with it isn't listed.

What does "single core gaming performance" even mean for a CPU that doesn't have an iGPU? How could that not be a category error to compare against Apple Silicon?

I was looking at https://www.cpubenchmark.net/single-thread/

See also:

https://nanoreview.net/en/cpu-list/cinebench-scores

https://browser.geekbench.com/mac-benchmarks vs https://browser.geekbench.com/processor-benchmarks

[fluoridation]

>What does "single core gaming performance" even mean for a CPU that doesn't have an iGPU?

Just a guess, but I would interpret it to mean how fast the CPU can issue commands to the GPU (which is usually, though not always, done in a single thread). For example, that could be measured by choosing a graphically lightweight game at minimum settings together with the best possible GPU and measuring the framerate. I.e. Making sure the bottleneck is the CPU, how high does the framerate go?

Whether the package includes a GPU or not is irrelevant, because what is being compared is the CPU part of the package, not the GPU. Whether they both happen to live within the same package or even the same die is irrelevant.

[throwa356262]

Even at the time of announcement M5 was not the fastest chip. Not even on single core benchmark where apple usually shines due to the design choice of having fewer but more powerful cores (AMD for examples does the opposite). For example on geekbench Core i9-14900KS and Core Ultra 9 285K were faster.

The distance was not huge, maybe 3%. You can obviously pick and choose your benchmarks until you find one where "your" CPU happens to be the best.

[nerdsniper]

https://nanoreview.net/en/cpu-list/cinebench-scores

https://www.cpubenchmark.net/single-thread/

https://browser.geekbench.com/mac-benchmarks vs https://browser.geekbench.com/processor-benchmarks

Apple leads all of these in single core, by a significant margin. Even at geekbench.com (3398 for AMD 9950X3D vs 3235 for the 14900KS vs ~4000 for various Apple chips)

I'm not sure I could find a single core benchmark it would lose no matter how hard I tried...