M3 Pro benchmarks vs real-world usage

[Golf7-R]

When I look at my M1 Max cores (work computer), the 2 efficiency cores are being used all the time while the 8 performance cores are mostly idling. I am running docker containers and compiling rust code in the background and yet still my efficiency cores are doing most of the work.

So I guess for my usage (coding) I would really benefit from having more efficiency cores like in the M3 Pro (which also have improved by 20%), especially for all the small services and apps running in the background (chat apps, email etc) during a normal work day.

Yes the benchmark scores are the same as M2 Pro, but I think having 2 more and improved efficiency cores has a greater impact on my kind of day to day work. But maybe I am wrong?!

All the negative comments around the M3 Pro almost pushed me towards a M3 Max for my private MacBook purchase. Would be great to have some more tests from the M3 Pro...

 

[Mcdevidr]

People need to stop with this already. The benchmarks are not the same as the m2pro. It’s a better chip no matter how you cut it. It looks like the binned chip has the same level of increase of regular m as the others on the series. In any case we do know the single core is much better and that alone is worth it to some.

 

[Johnny Steps]

I think people just need to stop running around in circles about upgrading and trying to justify it on the internet to a bunch of strangers. Upgrade if you want; only you know your own financial standing and whether or not you’re going into debt for a purchase. Plus if you want it just to want it, then that’s fine too.

 

[Golf7-R]

I am not justifying an upgrade. I was just looking to buy my private MacBook (don’t have one atm) and looked at the cpu utilisation from the MacBook I got from my company to check which m3 makes more sense instead of just looking at benchmarks.

 

[Aenean144]

When I look at my M1 Max cores (work computer), the 2 efficiency cores are being used all the time while the 8 performance cores are mostly idling. I am running docker containers and compiling rust code in the background and yet still my efficiency cores are doing most of the work.

So I guess for my usage (coding) I would really benefit from having more efficiency cores like in the M3 Pro (which also have improved by 20%), especially for all the small services and apps running in the background (chat apps, email etc) during a normal work day.

Yes the benchmark scores are the same as M2 Pro, but I think having 2 more and improved efficiency cores has a greater impact on my kind of day to day work. But maybe I am wrong?!

All the negative comments around the M3 Pro almost pushed me towards a M3 Max for my private MacBook purchase. Would be great to have some more tests from the M3 Pro...

Just wait for the benchmarks. You have to internalize that most of the dynamic discussion in fan forums are driven by lies and misinformation.

The M3 Pro will outperform the M2 Pro at every price point by the stated 10% to 15% by Apple. And, depending on GPU op, it will outperform the GPU in your M1 Max as well as all CPU ops. As usual, you buy more than what you think you need for RAM, then storage, then you can think about compute performance. So go, for 36 GB RAM, probably 1 TB storage, and you are probably good to go unless you are running a lot of VMs, docker containers, and doing big compile jobs, and in which case you have to go up the price tier.

 

[jonnyr78]

I posted my 16" M3 Pro Benchmark scores here, the performance seems good to me, but I think I put the thread in the wrong place https://forums.macrumors.com/threads/my-macbook-pro-16-m3-pro-benchmark-scores.2410184/

 

[TechnoMonk]

I am not justifying an upgrade. I was just looking to buy my private MacBook (don’t have one atm) and looked at the cpu utilisation from the MacBook I got from my company to check which m3 makes more sense instead of just looking at benchmarks.

That’s a good approach, get M3 Pro and run the tasks you need on both, see which one is better. M3 will also get extra support compared to M2.

 

[Mcdevidr]

Just got mine in the mail today. Real world usage is all my financial stuff and blah blah. The real benchmark I was waiting for was running Europa Universalis 4 as ive tested it for years on many different machines. Happily this is quite a bit faster on month ticks than m2 and is closer to the Ryzen 5800x3d I had which is pleasantly surprising. It is about a 30% uptick in speed over the M2 in this particular game.

 

[dgdosen]

A couple of things I don't get about the M3 Pro: if it has fewer transistors, more of those great E Cores, etc - why isn't battery life better? Apple could have made that a compelling selling point. Your "all day E-core Pro Powerhouse".

I also don't get those reviewers urging you to spec out a M3 Pro over an M3 - like the M3 Pro is so much better than the M3... I don't think it is. I'd think an M3 (or the most recent M) would be compelling to most users. If you need that 30ishGB of memory or are bottlenecked by your processor - then go straight from M3 to M3 Max. M3 Pro is like the Oldsmobile of the lineup. I don't think it's really compelling to anybody. Especially now that the P Cores are clipped.

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

 

[dgdosen]

A couple of things I don't get about the M3 Pro: if it has fewer transistors, more of those great E Cores, etc - why isn't battery life better? Apple could have made that a compelling selling point. Your "all day E-core Pro Powerhouse".

I also don't get those reviewers urging you to spec out a M3 Pro over an M3 - like the M3 Pro is so much better than the M3... I don't think it is. I'd think an M3 (or the most recent M) would be compelling to most users. If you need that 30ishGB of memory or are bottlenecked by your processor - then go straight from M3 to M3 Max. M3 Pro is like the Oldsmobile of the lineup. I don't think it's really compelling to anybody. Especially now that the P Cores are clipped.

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

One of two thoughts will occur to an M3 Pro buyer:

- I should have just bought an M3.
- I should have spent a few more dollars for the M3 Max.

 

[Aenean144]

A couple of things I don't get about the M3 Pro: if it has fewer transistors, more of those great E Cores, etc - why isn't battery life better? Apple could have made that a compelling selling point. Your "all day E-core Pro Powerhouse".

I also don't get those reviewers urging you to spec out a M3 Pro over an M3 - like the M3 Pro is so much better than the M3... I don't think it is. I'd think an M3 (or the most recent M) would be compelling to most users. If you need that 30ishGB of memory or are bottlenecked by your processor - then go straight from M3 to M3 Max. M3 Pro is like the Oldsmobile of the lineup. I don't think it's really compelling to anybody. Especially now that the P Cores are clipped.

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

Apple's advertised battery life numbers, like the WiFi web browsing one, don't use all the e-cores let alone the p-cores cores while the WiFi video ones spends most of its time with the video decode blocks that are common between SoC options. The power consumption of decode block is so low such that the test is really about how efficient the display is, not the chip.

If you actually are running all the p-cores or all the GPU cores, battery life will be inversely proportional to how many cores are being used. The Max chip will run through the battery twice as fast as the Pro chip on compute intensive processes, and while the Pro chip will take longer to complete a process, it will use less energy because of the 6 e-cores. Proportionately less energy to complete a process over the Max chip, and maybe even the regular M3.

That you think potential buyers will just jump from the M3 to Max option is not a sign of poor planning by Apple. That's definitely a sign that the product marketing team did a great job to get people to pay more. Upsell. It's also turtles all the way down. Buyers get what they think they need. Eg, a potential buyer may need 2 TB storage over the Max chip. No brainer on which to choose there for that buyer.

 

[JouniS]

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

If you upgrade the CPU/GPU cores and RAM, the next tier is only a little bit more expensive. The good deals are always closer to the lower end of the range. For example:

M3 with 16 GB / 512 GB: $1799
M3 Pro with 18 GB / 512 GB: $1999

 

[LoopsOfFury]

I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

The fan noise on my 16" M1 Max was far louder (and more frequent) than the fan noise on my 14" M1/2 Pros. Based on the reviews I've seen, the fan noise for the M2 (and especially now the M3) Max is even louder (53+ dBA). The base M3 on the other hand is actually louder than the M3 Pro (due to having only one fan instead of two), can only handle one external monitor, and has maybe half the performance in gaming.

 

[Mcdevidr]

A couple of things I don't get about the M3 Pro: if it has fewer transistors, more of those great E Cores, etc - why isn't battery life better? Apple could have made that a compelling selling point. Your "all day E-core Pro Powerhouse".

I also don't get those reviewers urging you to spec out a M3 Pro over an M3 - like the M3 Pro is so much better than the M3... I don't think it is. I'd think an M3 (or the most recent M) would be compelling to most users. If you need that 30ishGB of memory or are bottlenecked by your processor - then go straight from M3 to M3 Max. M3 Pro is like the Oldsmobile of the lineup. I don't think it's really compelling to anybody. Especially now that the P Cores are clipped.

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

Jeez yea you're right. I don't see how anyone can justify spending $1999 when they could just spend $2799 or hell $3199. Goodness. Yea you're right since I would like 16GB ram, screw it may as well go to 96 right, but then if I'm there why not go all the way to 128. Hell since we talking about it why would I only spend that much when I could pay someone to create some kinda battery hook up and just to go an M3 Ultra Studio and make it portable by paying what I don't know 4k for some type of custom battery setup.

 

[MayaUser]

That's $400 for 4 more P-Cores?

so $100 per P-core

 

[jdb8167]

A couple of things I don't get about the M3 Pro: if it has fewer transistors, more of those great E Cores, etc - why isn't battery life better? Apple could have made that a compelling selling point. Your "all day E-core Pro Powerhouse".

I also don't get those reviewers urging you to spec out a M3 Pro over an M3 - like the M3 Pro is so much better than the M3... I don't think it is. I'd think an M3 (or the most recent M) would be compelling to most users. If you need that 30ishGB of memory or are bottlenecked by your processor - then go straight from M3 to M3 Max. M3 Pro is like the Oldsmobile of the lineup. I don't think it's really compelling to anybody. Especially now that the P Cores are clipped.

This illustrates my point. Here are two specs I was looking at:

2,799 for an M3 Pro 14 with 36GB/1TB
vs
3,199 for an M3 Max 14 with 36GB/1TB

That's $400 for 4 more P-Cores? Of course! I just don't see how anyone who would justify bumping up from the M3 wouldn't also bump to the Max. It's a no brainer.

I think it was PCMag that measured 30 hours in their battery test.

Yes it was: Apple MacBook Pro 14-Inch (2023, M3 Pro) Review

There are several things that make the M3 Pro better than the M3. You get two external monitors for example. The performance is a significant improvement as well. Memory starts at 18 GB instead of 8 GB. You might not find these things compelling but a lot of people will.

 

[dgdosen]

I think it was PCMag that measured 30 hours in their battery test.

Yes it was: Apple MacBook Pro 14-Inch (2023, M3 Pro) Review

There are several things that make the M3 Pro better than the M3. You get two external monitors for example. The performance is a significant improvement as well. Memory starts at 18 GB instead of 8 GB. You might not find these things compelling but a lot of people will.

I don't disagree with that. Here's what I'm looking at (and I know everyone has different needs):

2,199 for an M3 14 with 24GB/1TB/4P/4E/10G
2,799 for an M3 Pro 14 with 36GB/1TB/6P/6E/18G
3,199 for an M3 Max 14 with 36GB/1TB/10P/4E/30G

It all comes down to prioritizing the biggest bottleneck (Theory of Constraints, anyone?). I want to first target my biggest time killer. If/when I'm compiling something large, that number of P Cores is definitely the biggest bottleneck. I might compare that total time to how often you face delays in other things that might cause 'memory bottlenecks': excessive paging, laggy behavior, beach balls, etc - and that can happen all the time and add up quickly, especially giving something like docker a chunk of ram.

I know from using my current Apple Silicon laptop, 16GB just won't cut it. 18? I doubt it... 24? Maybe.. 32? Yep. I think I want to try the former with 24GB.

If my biggest bottleneck are the P cores - I'd think getting 10! on a Max would be the best bet.

 

[flybass]

When I look at my M1 Max cores (work computer), the 2 efficiency cores are being used all the time while the 8 performance cores are mostly idling. I am running docker containers and compiling rust code in the background and yet still my efficiency cores are doing most of the work.

So I guess for my usage (coding) I would really benefit from having more efficiency cores like in the M3 Pro (which also have improved by 20%), especially for all the small services and apps running in the background (chat apps, email etc) during a normal work day.

Yes the benchmark scores are the same as M2 Pro, but I think having 2 more and improved efficiency cores has a greater impact on my kind of day to day work. But maybe I am wrong?!

All the negative comments around the M3 Pro almost pushed me towards a M3 Max for my private MacBook purchase. Would be great to have some more tests from the M3 Pro...

It is surprising you aren’t using more cores if compiling code. For most of my projects there are enough dependencies that can be compiled in parallel that generally all cores are used (at least building without caches or after a few hours of development)

 

[Ma2k5]

so $100 per P-core

And what about the 12 gpu cores! Also faster memory bandwidth.

 

[dgdosen]

Revisiting this thread. I've been watching activity monitor and, as with the OP, noticing that - most of the time - my work involves taxing the e-cores while the p-cores idle. So that has me thinking - maybe the Pro is a good option. Especially on a 14".

So the follow up question for those early adopters - does anyone regret getting an M3 Pro vs an M3 Max? Or vice versa?

 

[dgdosen]

One of two thoughts will occur to an M3 Pro buyer:

- I should have just bought an M3.
- I should have spent a few more dollars for the M3 Max.

Revisiting this thread again to take back these words. I know it pales in comparison to the power of the M3 Max, but I'm using an M3 Pro - and it is a great experience (for me).

My workflow revolves around many small services with a few large frequently updated images - and the M3 Pro is proving to be a great fit. It's very fast and efficient for this work. It handles short bursts of heavy activity (much shorter than for audio/video 'stuff') with ease. It feels like a great tool for the job.

I'll also share this analysis from notebookcheck regarding the M3 Pro and M3 Max. It compares both the power of the Max chip and the efficiency (which resonates to me) of the Pro chip: https://www.notebookcheck.net/Apple...nificantly-upgraded-its-Max-CPU.782974.0.html.

 

[wyrdness]

Are you running on battery or power adapter? More efficiency cores will only make a difference if you're running on battery.

If it's choosing to just use the efficiency cores, that indicates that your workload is very light and that the performance cores aren't required. In that case, upgrading to a faster machine isn't going to benefit you in any way.

How many parallel jobs are you compiling, because that makes a huge difference?

 

[bcortens]

Just wait for the benchmarks. You have to internalize that most of the dynamic discussion in fan forums are driven by lies and misinformation.

The M3 Pro will outperform the M2 Pro at every price point by the stated 10% to 15% by Apple. And, depending on GPU op, it will outperform the GPU in your M1 Max as well as all CPU ops. As usual, you buy more than what you think you need for RAM, then storage, then you can think about compute performance. So go, for 36 GB RAM, probably 1 TB storage, and you are probably good to go unless you are running a lot of VMs, docker containers, and doing big compile jobs, and in which case you have to go up the price tier.

Ummm... only in single core for the 10-15%

In multicore the M3 Pro is only sometimes faster than the M2 Pro and in code compilation the M3 Pro is no faster than the M2 Pro. M2 Pro vs M3 Pro

Lets go to GPU:
The M3 Pro is, on average, slower in metal compute than M2 Pro M2 Pro vs M3 Pro GPU Compute

The M3 Pro suffers a lack of transistors, the idea that they were going to pull off a magic trick and be able to remove 3 billion transistors and somehow improve performance across the board is just fantasy.

 

[Aenean144]

Ummm... only in single core for the 10-15%

In multicore the M3 Pro is only sometimes faster than the M2 Pro and in code compilation the M3 Pro is no faster than the M2 Pro. M2 Pro vs M3 Pro

Lets go to GPU:
The M3 Pro is, on average, slower in metal compute than M2 Pro M2 Pro vs M3 Pro GPU Compute

The M3 Pro suffers a lack of transistors, the idea that they were going to pull off a magic trick and be able to remove 3 billion transistors and somehow improve performance across the board is just fantasy.

Remember that I said "M3 Pro will outperform the M2 Pro at every price point by the stated 10% to 15% by Apple".

At $2000, a 11c M3 Pro does indeed outperform a 10c M2 Pro in multi-core by about 10% to 15%. This is the M3 Pro 5+6+14 configuration. The upgrade to the full 6+6+18 configuration is $200. I can't find the pricing for the M2 Pro MBP, but for the Mac Mini, the upgrade from the 6+4+16 M2 Pro config to the 8+4+19 M2 Pro config is $300, and I'm pretty sure Apple would have the same upgrade cost for the MBP models.

So, at every price point, yup, multi-core performance is going to be about 10% to 15% better than the M2 Pro.

For GPU performance, it depends on how you count for the hardware-based dynamic memory allocation, mesh shaders and raytracing. The M3 Pro GPU does indeed does not perform as well as the M2 Pro in Geekbench GPU compute benchmarks, but once there is an op that takes advantage of those new features, then there could be 1.1x to 4x improvements in GPU performance. How's that is priced is a nice marketing exercise. So for this, the buyer needs to know how well their workflow works with the GPU, which isn't all that difference from the past.

 

[jdb8167]

Lets go to GPU:
The M3 Pro is, on average, slower in metal compute than M2 Pro M2 Pro vs M3 Pro GPU Compute

The M3 Pro suffers a lack of transistors, the idea that they were going to pull off a magic trick and be able to remove 3 billion transistors and somehow improve performance across the board is just fantasy.

Why openCL instead of Metal? Not sure if an openCL benchmark has much relevance.