Marco Arment

the CPUs by the Mac identifiers. You can find results. We'll put links in the show notes so you can look at the results for yourself. Again, they are preliminary with a few number of data points, so take that for the grain of salt. And as I always say, Geekbench, like all benchmarks, may or may not be representative of anything you care about.

the CPUs by the Mac identifiers. You can find results. We'll put links in the show notes so you can look at the results for yourself. Again, they are preliminary with a few number of data points, so take that for the grain of salt. And as I always say, Geekbench, like all benchmarks, may or may not be representative of anything you care about.

Making a good representative benchmark is very difficult. In the end, what you care about is, does it do the thing that I need it to do faster than my old thing? Whatever that thing is. Are you rendering things? Are you compressing audio? Are you compiling an Xcode? Whatever it is that you're doing, all you care about is how fast does it do that?

Making a good representative benchmark is very difficult. In the end, what you care about is, does it do the thing that I need it to do faster than my old thing? Whatever that thing is. Are you rendering things? Are you compressing audio? Are you compiling an Xcode? Whatever it is that you're doing, all you care about is how fast does it do that?

But things like Geekbench try to be a representative benchmark for the various parts of the computer so you have some way to compare it. So here they are with all those caveats in mind. There's some amazing stuff going on. We've talked before about the M4's single-core performance being amazing. We only really had measurements of that in the iPad Pro because that was the only thing the M4 was in.

But things like Geekbench try to be a representative benchmark for the various parts of the computer so you have some way to compare it. So here they are with all those caveats in mind. There's some amazing stuff going on. We've talked before about the M4's single-core performance being amazing. We only really had measurements of that in the iPad Pro because that was the only thing the M4 was in.

But anyway, now the plain M4s are out. The M4 Pro has the same amazing single-core performance as the M4, but a little bit better because it's probably clocked higher and has better cooling. So M4 Pro is reaching... $3,900 in the single-core Geekbench benchmark. And how does that compare to its predecessors? That is a 26% increase over the M3 Pro.

But anyway, now the plain M4s are out. The M4 Pro has the same amazing single-core performance as the M4, but a little bit better because it's probably clocked higher and has better cooling. So M4 Pro is reaching... $3,900 in the single-core Geekbench benchmark. And how does that compare to its predecessors? That is a 26% increase over the M3 Pro.

It is also a 26% increase over the M3 Max because single-core really doesn't change. Pro, Max, whatever, like maybe they'll be clocked higher or maybe they have better cooling for sustained throughput, but single-core is usually similar. This was true back in the Intel days too, by the way.

It is also a 26% increase over the M3 Max because single-core really doesn't change. Pro, Max, whatever, like maybe they'll be clocked higher or maybe they have better cooling for sustained throughput, but single-core is usually similar. This was true back in the Intel days too, by the way.

Very often, if you wanted to get the fastest single-core performance, you would get, like, an iMac or something with a smaller number of cores because the big, like, 24-core or 12-core or whatever it was back in the day, the big multi-core thing that was in the Mac Pro had lower single-core performance because they couldn't clock those single-cores as high because the chip was just so big and hot.

Very often, if you wanted to get the fastest single-core performance, you would get, like, an iMac or something with a smaller number of cores because the big, like, 24-core or 12-core or whatever it was back in the day, the big multi-core thing that was in the Mac Pro had lower single-core performance because they couldn't clock those single-cores as high because the chip was just so big and hot.

But anyway, single-core is still great. Versus the M2 Ultra. which is the fastest SoC that you can get in the Mac Studio and the Mac Pro, the M4 Pro single-core performance is 41% faster. Multi-core. Now we start to get interesting. Multi-core, remember the M4 Pro is 14 cores. It has 10 power cores and 4 efficiency cores. Multi-core is 22,000.

But anyway, single-core is still great. Versus the M2 Ultra. which is the fastest SoC that you can get in the Mac Studio and the Mac Pro, the M4 Pro single-core performance is 41% faster. Multi-core. Now we start to get interesting. Multi-core, remember the M4 Pro is 14 cores. It has 10 power cores and 4 efficiency cores. Multi-core is 22,000.

That is a 48% increase over its predecessor, the M3 Pro. The M3 Pro made some very different choices. It was the 6-6 arrangement, 6 efficiency cores, 6 power cores. So it's bound to not do as well in multi-core. The M3 Pro was just plain aiming for like, I guess... Better low power performance for lots of non-demanding jobs, but Apple has changed it to mine. It's mine.

That is a 48% increase over its predecessor, the M3 Pro. The M3 Pro made some very different choices. It was the 6-6 arrangement, 6 efficiency cores, 6 power cores. So it's bound to not do as well in multi-core. The M3 Pro was just plain aiming for like, I guess... Better low power performance for lots of non-demanding jobs, but Apple has changed it to mine. It's mine.

The M4 Pro is really cranking it up. The M4 Pro's multi-core performance is 8% faster than the M3 Max. So now it is better than the previous Max chip. Oh, and by the way, the M4 Pro's multi-core performance in this one really hurts. 6% faster than the M2 Ultra.

The M4 Pro is really cranking it up. The M4 Pro's multi-core performance is 8% faster than the M3 Max. So now it is better than the previous Max chip. Oh, and by the way, the M4 Pro's multi-core performance in this one really hurts. 6% faster than the M2 Ultra.

Remember we're talking about the M4 Pro here. This is not the best M4 chip. This is like the middle one. There's M4, then there's M4 Pro. Its multi-core performance is better than the M2 Ultra. That's bad. Wow. That's real bad.

Remember we're talking about the M4 Pro here. This is not the best M4 chip. This is like the middle one. There's M4, then there's M4 Pro. Its multi-core performance is better than the M2 Ultra. That's bad. Wow. That's real bad.