M1 Pro 8/14 core vs M2?

[mactinkerlover]

Okay so this is what I'm really wondering. I'm surprised no one is really discussing this. We obviously know that the 10/16 core m1 pro will be significantly more powerful than the m2, but what about the binned m1 pro? How do y'all think the 8/14 m1 pro will compare to the 8/10 m2? Obviously, m2 isn't benchmarked yet so we don't know for real the difference, but I want y'alls speculation, and we can come back and see if we were right once the machines are in the wild.

My speculation is that the graphics on the 14 core m1 pro will be noticeably more powerful than the m2, but the cpu performance of m2 could be very close to the 8 core m1 pro.

Have fun discussing y'all!

 

[theorist9]

I assume you're asking about multi-core CPU performance, since the M2 is going to be faster than any M1 variant in single-core (SC perf. for M1/Pro/Max/Ultra are the same). So let's use GeekBench 5 scores as a crude measure:

M1 (4P+4E) = 7,400
M2 (4P+4E) = 7,400 x 1.18 = 8,700 (Apple says it's 18% faster in CPU multicore than the M1)
M1 Pro (non-binned, 8P+2E) = 12,100
M1 Pro (binned, 6P+2E) = 12,100 x 0.8 = 9,700 (M1 Pro binned is ~20% slower than non-binned).

So it looks like the binned M1 Pro should be about 10% faster than the M2 for CPU multi-core. If you're using the M2 Air, that figure could go higher for an extended workload if it causes the Air's M2 to throttle.

The equivalent GPU calculation is left as an exercise for the OP.

 

[Chancha]

The M1 Pro binned still has 6 performance cores vs the 4 on the M2. It has double the potential memory bandwidth (200 vs 100GB/s). Active cooling (if compared to the Air). Not sure about the media engines or ML cores but those don't impact normal use. Altogether we will likely see a mixed bag of results with one better than the other interchangeably on a list of various benchmaking tasks.

In reality the bigger deciding factor in choosing 14" over the M2 Air will be the screen and ports vs portability. These are more dramatic differences than what we will see in performances.

 

[JPack]

The M1 Pro binned still has 6 performance cores vs the 4 on the M2. It has double the potential memory bandwidth (200 vs 100GB/s). Active cooling (if compared to the Air). Not sure about the media engines or ML cores but those don't impact normal use. Altogether we will likely see a mixed bag of results with one better than the other interchangeably on a list of various benchmaking tasks.

In reality the bigger deciding factor in choosing 14" over the M2 Air will be the screen and ports vs portability. These are more dramatic differences than what we will see in performances.

Agreed, I think for the vast majority of uses cases, the lowest binned M1P wins because greater number of P-cores, significantly larger cache and higher memory bandwidth.

 

[UBS28]

The binned M1 Pro has 6 high performance cores while the M2 only has 4.

I don’t see how having 50% more high performance cores will lead to “close” performance.

 

[theorist9]

The binned M1 Pro has 6 high performance cores while the M2 only has 4.

I don’t see how having 50% more high performance cores will lead to “close” performance.

Did you not read post#2?

 

[IJBrekke]

Echoing others, my speculation is that peak performance will be pretty similar in real-world usage, but that thermal throttling on the Air M2 will ultimately be the difference. The 13” M2 MBP vs. the binned 14” MBP will be closer, probably close enough to recognize that you’re primarily paying for other things besides increased performance in choosing the 14”.

 

[quarkysg]

The 13” M2 MBP vs. the binned 14” MBP will be closer, probably close enough to recognize that you’re primarily paying for other things besides increased performance in choosing the 14”.

The 14" MBP has twice the memory bandwidth so I would think the 14" will still come out ahead in terms of CPU MT workloads, as least for workload that uses large amount of memory. M2 definitely will win the ST workload shootout.

 

[senttoschool]

The 14" MBP has twice the memory bandwidth so I would think the 14" will still come out ahead in terms of CPU MT workloads, as least for workload that uses large amount of memory. M2 definitely will win the ST workload shootout.

It's very easy to calculate.

M2 MT: 18% improvement over M1. 7,434 * 1.18 = 8,772 Geekbench5.
M1 Pro 8 cores: 9,511 Geekbench5.

M1 Pro w/8 cores has 8.4% faster MT performance than M2.

Of course, 14" MBP can sustain the MT for longer because it has a fan and larger chassis.

All in all, I would rather have the M2 CPU performance over M1 Pro w/ 8 cores. This is because ST is more important to me and the MT is close enough.

 

[quarkysg]

It's very easy to calculate.

M2 MT: 18% improvement over M1. 7,434 * 1.18 = 8,772 Geekbench5.
M1 Pro 8 cores: 9,511 Geekbench5.

M1 Pro w/8 cores has 8.4% faster MT performance than M2.

Of course, 14" MBP can sustain the MT for longer because it has a fan and larger chassis.

All in all, I would rather have the M2 CPU performance over M1 Pro w/ 8 cores. This is because ST is more important to me and the MT is close enough.

Not sure tho. if GB5's runs depends more on CPU cache or memory? I somehow suspect they will come pretty close in performance and the M1 Pro will still come out ahead in real work tasks, even for ST.

I could be wrong tho. Have to wait for benchmarks for the M2 to assess. Hopefully Anandtech folks will do a SPEC benchmark for the M2.

 

[gpat]

The M2 will run much hotter than an M1.
The 14" Pro has a fan, it would be invaluable in this comparison.
This will be an issue in real life if you plan to squeeze any kind of power out of the M2.
But if you don't, even the M1 Air would be enough, so the M2 Air wouldn't be the better choice either way.

 

[playtech1]

Do we have any sense of what improvements the M2 brings on single threaded performance?

I took from Anandtech's article that the big change is likely to be that the efficiency cores are beefed up, improving MT performance, but I think ST performance (or really just the P core performance) still has relevance in the general 'feel' of speed.

Of course M series chips are a bit unusual in that there is little point analysing the chip in a vacuum, since it will always be in an Apple device and may be configured and perform differently in each one.

 

[IJBrekke]

The 14" MBP has twice the memory bandwidth so I would think the 14" will still come out ahead in terms of CPU MT workloads, as least for workload that uses large amount of memory. M2 definitely will win the ST workload shootout.

Ahead yes, but my general point is that the difference will be at the top end and negligible for most users. The other attributes of the 14” will feel more different than the speed of the machines, including the memory bandwidth gap.

 

[jav6454]

I assume you're asking about multi-core CPU performance, since the M2 is going to be faster than any M1 variant in single-core (SC perf. for M1/Pro/Max/Ultra are the same). So let's use GeekBench 5 scores as a crude measure:

M1 (4P+4E) = 7,400
M2 (4P+4E) = 7,400 x 1.18 = 8,700 (Apple says it's 18% faster in CPU multicore than the M1)
M1 Pro (non-binned, 8P+2E) = 12,100
M1 Pro (binned, 6P+2E) = 12,100 x 0.8 = 9,700 (M1 Pro binned is ~20% slower than non-binned).

So it looks like the binned M1 Pro should be about 10% faster than the M2 for CPU multi-core. If you're using the M2 Air, that figure could go higher for an extended workload if it causes the Air's M2 to throttle.

The equivalent GPU calculation is left as an exercise for the OP.

It doesn't work that way, but the idea behind it stands. In the end M1 Pros are still faster than M2.

 

[theorist9]

It doesn't work that way, but the idea behind it stands. In the end M1 Pros are still faster than M2.

My calculation works under the assumptions that I presented. It's a prediction, but a prediction is all you can do. When you claim "it doesn't work that way", I take that not to mean that you disagree with my approach, but that you didn't understand it.

 

[jav6454]

My calculation works under the assumptions that I presented. It's a prediction, but a prediction is all you can do. When you claim "it doesn't work that way", I take that not to mean that you disagree with my approach, but that you didn't understand it.

I understood it both that it's a prediction and that it's an assumption. Just pointing out it's not liner. That said, the entire conclusion based off it holds.

 

[theorist9]

I understood it both that it's a prediction and that it's an assumption. Just pointing out it's not liner. That said, the entire conclusion based off it holds.

It should have been obvious I was making a first-order back-of-the envelope calculation, which means one of the key simplifications is assuming linearity. To tell someone who's doing it that 'they're doing it wrong' (your exact words were "it doesn't work that way") reflects a lack of understanding of how these simple calculations are done.

I even explicitly called it "crude" so people wouldn't think it was meant to be otherwise. But I guess not everyone gets the message.

Plus, to the extent my calculation gets it wrong, it wouldn't be for the reason you think. We're talking about small changes here (~20%), so the non-linearity wouldn't have much effect. Thus the biggest potential source of error isn't non-linearity, it's that the benchmark Apple used to get the 18% increase might have been very different from Geekbench.

 

[UBS28]

You can get a 14" MBP M1 Pro cheaper than the M2 MacBook Air if both are similarly speced from retailers. The M2 is too expensive.

 

[theorist9]

Do we have any sense of what improvements the M2 brings on single threaded performance?

I took from Anandtech's article that the big change is likely to be that the efficiency cores are beefed up, improving MT performance, but I think ST performance (or really just the P core performance) still has relevance in the general 'feel' of speed.

Of course M series chips are a bit unusual in that there is little point analysing the chip in a vacuum, since it will always be in an Apple device and may be configured and perform differently in each one.

I don't. As you said, the expectation is that some of the 18% increase in MT performance is from big improvements to the efficiency cores (such that Ecore/Pcore performance is higher in the M2 than the M1), in which case the SC speed of the performance cores will go up by less. My completely speculative guess is that the SC speed will go up by ~10%.

 

[theorist9]

You can get a 14" MBP M1 Pro cheaper than the M2 MacBook Air if both are similarly speced from retailers. The M2 is too expensive.

I'm very suspicious of this claim. Where can you get a new 14" MBP for less than a new M2 Air, similarly-spec'd, and both from Apple-authorized retailers?

A min-spec binned 14" MBP has 16 GB RAM and a 512 GB SSD, and is $2000 at full price and $1800 with an educational discount plus the current $50 off promo.

A 16 GB/512 GB base Air is $1600 at full price, and $1480 with an educational discount (the $50 off is unavailable because this would be BTO). [Edit: I'd originally said $1700 and $1580, but that was incorrect.]

I do agree that, for the ~$300 difference, the 14" M1 Pro is a better value. It has better cooling, slightly higher MT performance, significantly higher GPU performance, more ports, and a larger and better-quality display. Plus it can drive 3 external monitors compared with only one for the Air. All the M2 Air offers is slightly higher SC performance plus reduced size and weight.

 

[UBS28]

I'm very suspicious of this claim. Where can you get a new 14" MBP for less than a new M2 Air, similarly-spec'd, and both from Apple-authorized retailers?

A min-spec binned 14" MBP has 16 GB RAM and a 512 GB SSD, and is $2000 at full price and $1800 with an educational discount plus the current $50 off promo.

A 16 GB/512 GB Air is $1700 at full price, and $1580 with an educational discount (the $50 off is unavailable because this would be BTO).

I do agree that, for the ~$200 difference, the 14" M1 Pro is a better value. It has a better display, better cooling, and more ports. Plus it can drive 3 external monitors compared with only one for the Air.

Like I said, you can find such prices from retailers currently. I saw it yesterday when I was shopping.

 

[theorist9]

Like I said, you can find such prices from retailers currently. I saw it yesterday when I was shopping.

Then please provide a link. E.g., it's possible you didn't realize the low price you saw was for a refurb. Or maybe it wasn't from an authorized retailer. There's no way to know without a link.

 

[UBS28]

I challenge you to provide a link.

I'm not going to give away where I live on the internet. The fact is, I can buy a 14" M1 MacBook Pro cheaper than a M2 MacBook Air 16 GB RAM 512 SSD from retailers.

You may cry that you don't believe it, it doesn't change the fact that is true.

 

[theorist9]

I'm not going to give away where I live on the internet. The fact is, I can buy a 14" M1 MacBook Pro cheaper than a M2 MacBook Air 16 GB RAM 512 SSD from retailers.

You may cry that you don't believe it, it doesn't change the fact that is true.

I'm not crying, and you need to grow up. You know how the internet works. Random guys make all sorts of BS claims. That's why it's always reasonable to ask for evidence when anyone makes a suspicious claim. If you can't handle someone asking for evidence, you don't belong here.

 

[UBS28]

I'm not crying, and you need to grow up. You know how the internet works. Random guys make all sorts of BS claims. That's why it's always reasonable to ask for evidence when anyone makes a suspicious claim. If you can't handle someone asking for evidence, you don't belong here.

Cry more man. I really don't care if you don't believe it. And you don't decide who belongs here. Who the hell do you think you are?

There are some good deals here on the 14" M1 Pro MacBook Pro. Live with it that such deals exist.

Btw, welcome to my ignore list.