Here's How Fast the 9-Core M4 Chip is Based on Early Benchmark Result

[MacRumors]

An alleged Geekbench 6 benchmark result for the lower-end version of the M4 chip with a 9-core CPU surfaced over the weekend, providing a first look at how it performs compared to the higher-end M4 chip with a 10-core CPU.

The result indicates that the 9-core M4 chip is around 10% slower than the 10-core variant in terms of multi-core performance. While that outcome might seem obvious at first glance, the 9-core variant of the chip has three performance cores, instead of four in the 10-core variant, so this result was not guaranteed.

Assuming that the Geekbench 6 listing is accurate, the 9-core M4 chip is still around 13% faster than the M3 chip for multi-core performance, and up to 35% faster than the M2 chip in the previous-generation iPad Pro.

Here's how the 9-core M4 chip stacks up in multi-core Geekbench 6 results:

• M4 chip (10-core CPU): ~14,600
• M4 chip (9-core CPU): ~13,000
• M3 chip (8-core CPU): ~11,600
• M2 chip (8-core CPU): ~9,600

The new 11-inch and 13-inch iPad Pro models are equipped with a 9-core M4 chip when configured with 256GB or 512GB of storage, while models with 1TB or 2TB of storage are equipped with a 10‑core M4 chip. Both variants have a 10-core GPU.

For reasons unclear, the Geekbench 6 listing shows the iPad Pro with the 9-core M4 chip as running iPadOS 17.6, whereas all results for the 10-core model have shown iPadOS 17.4 so far. The first beta of iPadOS 17.6 will likely be made available for testing later this month, following the upcoming release of iPadOS 17.5.

The new iPad Pro models launch on Wednesday.

Thanks, Michael Burkhardt!

Article Link: Here's How Fast the 9-Core M4 Chip is Based on Early Benchmark Result

 

[shadowboi]

Snapdragon will need 4 years more to get there. I mean to match M2 performance

 

[EugW]

It should be noted that M4 gets a significant boost from the Object Detection subtest due to newly added functionality in M4, but most software will not leverage this. It should also be noted that this was added in the latest Geekbench 6.3, but was not there in 6.2.

Here's my summary of the M4 9-core results, against a (better scoring than average) M3 MacBook Air:

 

[MrGimper]

So those improvements are down to additional cores and higher clocks. Shock.

In fact, although not scientific, if you divide the M3 score by 8 and multiply by 9, you get the same score as the M4. And the same for 10 cores. And despite the M4's cores being clocked 0.3 GHz higher.

 

[Realityck]

So each processor is a modest improvement over the other. Did anyone come across what amount of maximum Ram the base M4 supports? Base M2/M3 was only 24 GB, compared to the original M1 with 16 GB.

 

[julesme]

Let’s not forget M1, as many are upgrading from an M1 iPad Pro:

M1 multi-core = 8,301
M4 multi-core (9-core) = ~13,000, for a ~57% improvement.

 

[BorisDG]

Still early. More results can popup and close the gap. I'm looking for the iPad Pro M4 11'' 256GB. I don't need 16GB of RAM, neither the 10 cores. iPad OS will fly even with 1. Anyway, it will be HUGE upgrade from my Air 4. I just push my thumbs that I don't get pixel defects.

 

[DHagan4755]

Those M4 Pro 12-core results for a presumably base level 14" MacBook Pro are going to turn heads, never mind an M4 Max!

 

[mrhick01]

So those improvements are down to additional cores and higher clocks. Shock.

Additional cores and higher clocks in the same wattage and thermal footprint. That counts for something.

We know Apple as a company doesn’t run their silicon at super high wattages for “fastest right now” claims.

 

[jb310]

Playing games on this is going to be SO good... well, depending on how the devices handle heat, at least. 😅

 

[MrGimper]

Let’s not forget M1, as many are upgrading from an M1 iPad Pro:

M1 multi-core = 8,301
M4 multi-core (9-core) = ~13,000, for a ~57% improvement.

It's serious like picking between 2 supercars to drive on a UK 70 MPH motorway.

 

[djinn]

Playing games on this is going to be SO good... well, depending on how the devices handle heat, at least. 😅

That's what I'm waiting to hear on the M4 iPads. How the thermals do and possible throttling.

 

[MTShipp]

Playing games on this is going to be SO good... well, depending on how the devices handle heat, at least. 😅

Just as I am doing right now, I expect surfing MacRumors is going to be so much snappier on my 13” M4 Pro coming Wednesday!

 

[howywood]

Meaningless with a POS OS

 

[nickf]

How's base M4 performance vs M2 Pro/M3 Pro?

 

[kemal]

This chip in a low end Mac mini.

 

[socialwill]

Snapdragon will need 4 years more to get there. I mean to match M2 performance

I have to think this is a big push for Apple to show just how far ahead they are. Even the new TOPS numbers are a massive jump from the M3 series for the Neural Engine. The Pro, Max, and Ultra chips will see some big jumps later this year.

 

[mrhick01]

This chip in a low end Mac mini.

Probably the ultimate emulation box.

 

[socialwill]

How's base M4 performance vs M2 Pro/M3 Pro?

I thought Apple said the M4 can do what the M2 did with less power, so with the same power it is 50% better if I recall?

 

[socialwill]

So those improvements are down to additional cores and higher clocks. Shock.

In fact, although not scientific, if you divide the M3 score by 8 and multiply by 9, you get the same score as the M4. And the same for 10 cores. And despite the M4's cores being clocked 0.3 GHz higher.

Correct, they cores do perform slightly better with higher clocks and some refinements, but they also shrunk this down from the size of the M1/M2 series. I am not a chip person, so I am not sure if shrinking the chip by default gives you better performance as well? But it does use less power for the same performance as the M2, that is pretty impressive.

 

[vtrautia]

I am considering upgrading my iPad Air 2. Finally there is thinner iPad on sale.

 

[MrGimper]

Correct, they cores do perform slightly better with higher clocks and some refinements, but they also shrunk this down from the size of the M1/M2 series. I am not a chip person, so I am not sure if shrinking the chip by default gives you better performance as well? But it does use less power for the same performance as the M2, that is pretty impressive.

I agree, it's just the comparisons to M3 and what may be in MBPs that are going (in my opinion) over-the-top.

 

[EugW]

So those improvements are down to additional cores and higher clocks. Shock.

In fact, although not scientific, if you divide the M3 score by 8 and multiply by 9, you get the same score as the M4. And the same for 10 cores. And despite the M4's cores being clocked 0.3 GHz higher.

Additional cores and higher clocks in the same wattage and thermal footprint. That counts for something.

We know Apple as a company doesn’t run their silicon at super high wattages for “fastest right now” claims.

If you remove the boost from the Object Detection subtest in Geekbench 6.3 (since it is an outlier that skews the results in M4's favour), and compensate for M4's higher clocks, the single-core speed of M4 is still about 3% faster than single-core speed of M3. This is illustrated by running Geekbench 5.5.1 on them:

Somebody else got a similar 3% Performance Per Cycle improvement of M4 over M3 after compensating for M4's higher clocks, by running Geekbench 6.3 but with the Object Detection subtest removed from the results.

 

[endemize]

I am considering upgrading my iPad Air 2. Finally there is thinner iPad on sale.

Wow that is going to be such a massive upgrade for you! I’m coming from a 2018 11” and that seems like a huge jump in itself.

 

[Unregistered 4U]

Snapdragon will need 4 years more to get there. I mean to match M2 performance

Oh, they’ll be able to match the performance, they’ll just have to do what Intel does and pump more wattage through it