M4 Pro: low benchmarks, why?

[Melbourne Park]

I bought my wife an Ultra i7 HP Elitebook with 32 GB RAM and 1 TB drive, a 1000 nit screen (claimed) and a wonderful keyboard on it. But it feels so so so so slllllooowwww. I'm going have to check whether it's a repackaged i5 from 2018. And I am having issues getting the data and email accounts from the old notebook. I'm typing this from what feels like an exceptionally quick in comparison 2017 2.9 GHz Quad-Core Intel Core i7 in Ventura. I've had a new Max in a box for three days now and I still haven't been able to unpack it because of my wife's PC notebook refusing to accept my various attempts to export Outlook archives and then re-import them into a Classic version on Outlook on Windoze 11 - which sure as all heck sucks. And with Apple, I'd call them and if they couldn't help me over the phone, they'd take over the Mac and arrow me around and basically do it for me. But I reckon the only issues I'll have with that Max will be getting some passwords across from my browsers. Who cares about speed when the OS on Windoze is so terrible it makes the whole thing feel like its been pasted over with glue.

 

[throAU]

The only issues I had with my Max (first time not importing direct from Time Machine in 10 years) was that it defaulted to turning on screen time and it took me a while to figure out how to stop getting locked out of stuff at 10pm

 

[streetfunk]

OK, I got a little concerned about performance variations when I first read this thread, but happy to see my new MBP 16" M4 Max clocked in 4057 singe-core and 26181 multi-core in Geekbench. This was on battery (not plugged in) and with the performance mode on Automatic. Good enough...

did you had it in a freezer first ?
just asking for the protocol,...and for the stupids here in this thread to make something clear.

i for my part "hoped" to come close to something like 3950 SC - GB6

right now, it effectivly looks like my M4 is a plume.
I´ll do a own thread for that, and will tell the story of what happened there.
...ordered a a M4base in the meantime.

Who cares about speed when the OS on Windoze is so terrible

you are completly offtopic. my quote of you shows your main message no ?
you are in this thread not welcome with that ! (while it might be a valuable topic in it´s own)
The topic is another one. Please open your own thread for that !

i´m the topic starter, please respect that.
I´m so so tired of all the people who can´t keep rules going.
Old man who have turned into internet addictives, and lose control, behaving like year olds. Leave out Gen-Z and the likes. (No, this is not pointed to any specific person. Just a general observation vs. forums)

the Topic is clear !
How much you can take it offtopic should be clear too, at least to any intelligent matured beeing.
That way can we have an internet that works.
The other way around do we create an interent that just sucks all the time -and live- out of us. Just for nothing.
THIS IS NOLONGER WANTED !


and from my side nolonger accepted !
A short to the point topic serves everybodys interested in specific topics.
everything else turns the internet in an never ending blahblablahblahblah sort of world.


/ Rant ( i´m the topic starter | the rules here have been phrased out in my OG post | please post on topic ! Thanks)
( No, this, my "forced" offtopic is NOT a discussion point either)

and: Thanks for any valuable participation to everybody !

 

[Melbourne Park]

Kettle calls pot black !!!

Fact is your unit is what I'd expect. Well no ... the multi is much more than I'd have expected.

Your results single core is 15% faster than M3 Pro single and 39% than M3 Pro Multi.

Your expected score for single don't match the results I have seen either. Your single compared to Max does seem low though.

If you're unhappy, buy a 40 core Max. But Max owners don't care about single core which are not much faster than the Pro if at all. And they don't care anyway because they bought a Max.

Geek etc means little its how well it operates your app compared to before. If not enough, return and try another. Or wait for the next generation.

Curiously I did a test before answering you, and my Max results are better than the reported ones. But it all means nothing of importance to me. On the results I looked at too, the M4 Pro results were no longer listed. Give it time I reckon. You're upset though, so return it and why see if they'll let you try one in the store. Or get your refund and try another. No point feeling jilted. Got to say the M4 feels a lot better than the M3. Everything about it. Not sure why.

 

[throAU]

Be aware that any browser extensions you have installed in safari will quite likely massively influence browser benchmark results.

I've got a few and my max was WAY below what it should be. But hey, I have ad free and un-enenshirtified internet as a result of the plugins and it still feels fast so... win?

These machines basically crush any workload you throw at them. worry less about benchmarks and enjoy!

 

[Macintosh IIcx]

did you had it in a freezer first ?
just asking for the protocol,...and for the stupids here in this thread to make something clear.

i for my part "hoped" to come close to something like 3950 SC - GB6

No, just standing on table as usual, no power plugged in.

 

[Melbourne Park]

I did my first test yesterday when replying to streetfunk's rant about being off topic.

My test was after I had been doing some video resolution upgrades. I was unsure about the speed differentials to the M3 Max. But I realised I could do lots of other things and have that app in the background, without any slowing down of any apps including the hard going one. I then browsed this site and did the speed test, and I got:

3945 26039

That was running on battery with about 20% left, in a quite warm room. There is an AGA stove and it radiates heat, the Mac was on the kitchen bench opposite. Most of the room was 25C, so I think the Mac was fairly warm. I was going to take a pic of the keyboard but my thermal camera that goes on the lightning connector has failed ...

I did one in the morning, and restarted, and plugged the power cord in (battery was under 20%). I used the provided power supply and the magsafe power input.

4086 26632 Single up 2.3% and multi 3.6%. I guess that shows the units are pretty heat sensitive. And yesterday's test I did not hear a fan, but then I wasn't listening.

I then checked the mode and it was auto. I doubt auto makes any difference unless one is pushing the computer and is on battery but I don't know.

I also viewed some reviews on these machines, and while they are faster, they suck the power much more. But since they are quicker, overall its a gain in endurance/processing time. If that matters though ... and Apple didn't increase the battery size. The Pro would stress the batteries less and I suspect in real single processor mode over sustained work it would be a much better computer than a Max. Also on battery with multi core while it would take longer than Max I do wonder if it would get more work done on battery than a Max. Less heat means more efficiency afterall.

I did notice in the iFixit below that the Max heat sink is bigger than the M3 version. I wonder if the M4 Pro's has also increased? And I wonder if the heat sink is not properly adhered, whether that would cause the M4 Pro's performance to drop in a test as brief as the Geektest one is?

I bought this unit because I wanted Thunderbolt 5 for the long term expansion, as I hate Apple for their outrageous internal drive prices, and their soldering and also the complexity of their internal drives. And now many of their internals while most seem plug ins, unless acknowledged somehow by Apple they may not work.

But the speed tests for a user who bought a T-5 Ultra OWC 2 TB drive, were quite disappointing. He got basically 3,000 MB/s. Although the T-5 drive was faster than his T-3 drives when being speed test on his M4 Max.

So it seems to me that Thunderbolt 5 might take some time to optimise. The issue is always where the bottlenecks are. The reviewer was moving 4 large files, and while Blackmagic showed initial 6,200 writes and 5100 reads, it soon fell back to around 3,000'ish. Dosdude would be happy about those slower scores than I expected!

So the performance of Macs with speed tests doesn't mean much to me. I may now buy an exterior PCI case for my PCIe RAID card, which probably isn't worth much used. I reckon I will get around 3,000 MB/s for that.

I do wonder about inconsistencies with Apple speeds. I know in notebook game PCs some people say put in a top thermal paste for the CPU. Hugely difficult to do with a Mac Pro M (although the battery looks straightforward to replace:

.

If thermal paste is not perfectly installed, maybe that is a reason why performances are not much the same when doing an artificial test?

 

[Mike Boreham]

did you had it in a freezer first ?
just asking for the protocol,...and for the stupids here in this thread to make something clear.

i for my part "hoped" to come close to something like 3950 SC - GB6

right now, it effectivly looks like my M4 is a plume.
I´ll do a own thread for that, and will tell the story of what happened there.
...ordered a a M4base in the meantime.

you are completly offtopic. my quote of you shows your main message no ?
you are in this thread not welcome with that ! (while it might be a valuable topic in it´s own)
The topic is another one. Please open your own thread for that !

i´m the topic starter, please respect that.
I´m so so tired of all the people who can´t keep rules going.
Old man who have turned into internet addictives, and lose control, behaving like year olds. Leave out Gen-Z and the likes. (No, this is not pointed to any specific person. Just a general observation vs. forums)

the Topic is clear !
How much you can take it offtopic should be clear too, at least to any intelligent matured beeing.
That way can we have an internet that works.
The other way around do we create an interent that just sucks all the time -and live- out of us. Just for nothing.
THIS IS NOLONGER WANTED !


and from my side nolonger accepted !
A short to the point topic serves everybodys interested in specific topics.
everything else turns the internet in an never ending blahblablahblahblah sort of world.


/ Rant ( i´m the topic starter | the rules here have been phrased out in my OG post | please post on topic ! Thanks)
( No, this, my "forced" offtopic is NOT a discussion point either)

and: Thanks for any valuable participation to everybody !

Threads take on a life of their own. At the start, replies will be more on topic, directly relevant to you, but as time goes on many more people will chip in, especially with a title like this one, which hints at an Apple problem with the latest and greatest M4 chips.

You can’t expect to control a thread just because you started it.

 

[streetfunk]

You can’t expect to control a thread just because you started it.

i don´t expect anything.



JFYI:
my M4pro was the other day at a CPU temp of 102°C.
on all 10 P-cores, give or take 2°.

What was i doing ?
FF was open with something like 11 tabs.
I had something like 5 tabs open from my internet provider. A major company in that biz here in europe.
They have something vERY fishy going on for sure. (since quasi same happened two days in a row. which also means: it´s reproducable)

Nevertheless, around 100° on all 10 P-cores based on just having FF open is wild.
i was at the time for something like 30 minutes on the phone with them.
Not doing anything. Just switching the visuial focus between these tabs for 2-3 times during these ~30 minutes.

so, right now i am under the impression that something IS wrong with my machine.


the point has been made in that other, the "noise problem" thread:
The guess is on the table that some machines got a bad thermal paste job donne. It´s just a guess, but seems a valid one. in regards to my just mentioned case: i might open a own thread for that. Please don´t pick up on it here.

 

[Melbourne Park]

Are you too late to return it? And have Apple acknowledged that it's not performing as it should have? I'd be wanting a replacement. And I'd be requesting from them a full refund, which is more difficult for them than a simple replacement.

With the paste: gamer notebooks say as a matter of course to replace the cheap thermal paste that PC makers use. Which is surprising but cost accountants infect manufacturing a lot these days. I don't know how makers differentiate PC assemblies, from a Dell i5 to a top draw Intel i914900HX chip, that i5 would not be getting an expensive paste at all; who knows if a line making a Razor got better paste?

With new models there are always issues, although from a manufacturing point of view, these units are very similar to the previous ones. But in the Max (I'm not sure about the Pro) there is a different heat sink. And then of course there is the CPU itself, and maybe somehow its QC did not pick up it as faulty when it was in fact faulty.

But whatever, just demand a refund and settle on a replacement. And if its dinged or scratched somehow, that's not relevant because its not performing as you expected or were promised by the advertising, publications and Apple's marketing. And while it turns on, it's really a DOD - dead on arrival - because the CPU has never functioned as it is supposed to. A DoD is a replacement without any service or repair. And it's Apple's business model to often replace rather than repair.

I presume you have turned off the mac and immediately restarted while holding the D key? That brings up some type of diagnostic process. Also: Applications > Utilities > Console for looking at any recurring system errors. Also iStat Menus, I'm not sure if its free though, reports things. I guess too check the drive via disk utility.

Oh delete and reform the mac, do a clean install, and don't install any of your software or links until you run some more tests. I imagine the first thing at a genius bar would be for them to run a test, then format and re-install and then run another test. I presume you have already done all that, but if not, give it a go.

It would also be interesting to buy a mac mini Pro and install your software via a time machine migration setup, and run your software on the mac mini pro and compare - they should operate at around the same speed. But I think you already know your's is not right.

And Oh! If you don't get what you want, ask to speak to a supervisor. And at all time be polite and measured. And with the supervisor, I mean actually speak, not have the person to whom you've requested to peak to a supervisor come back to you claiming that after having just spoken to a supervisor, the supervisor has corroborated what Apple's policy is. And ask for the supervisor's name (they will probably not give you their full name but any "name" should suffice). And tell them you have extensive notes on what has happened. If they ask for notes you can tell them that you will not provide because unless they replace the computer you will be using your extensive notes for taking things to a different dispute arena.

 

[throAU]

Nevertheless, around 100° on all 10 P-cores based on just having FF open is wild.

There's something else going on with your machine, I've never seen all my P cores doing much at all unless I've got some seriously heavy work going on in the background.

I mean normal web browsing workload on my max with safari looks like this:

I've never heard the fans activate while using Firefox. The only time I have ever heard fans (and this is a full fat max in a 14" chassis) are: gaming, blender renders, long running LLM queries. General use: never.

I'd say your machine is definitely defective, there have been some reports of machines being overly hot/loud, maybe yours didn't get enough thermal paste at the factory.

 

[ChrisA]

I

2. I am not a chip expert, but believe chips are not like machined parts, capable of being the exactly the same to a very close tolerance.

You are right to admit you are not an expert. If there is anything on this Earth that should be identical, it is the speed of two of the same computer chips when they are clocked at the same rate. OK, yes you can test them and find some that can be clocked higher, but Apple does not do this. All of them are clocked the same.

 

[ChrisA]

There's something else going on with your machine, I've never seen all my P cores doing much at all unless I've got some seriously heavy work going on in the background.

You have never seen the P-cores doing much because you are not doing anything that even mildly stresses the computer. This is exactly the way it should be.

It is like if you bought a car with a 600 horsepower engine, then somehow you measure the power while driving to work and see that it is about 25 horsepower. Where you cheated? No. It only takes 25 HP to drive on the road at traffic speeds.

 

[throAU]

You have never seen the P-cores doing much because you are not doing anything that even mildly stresses the computer. This is exactly the way it should be.

Of course. I'm aware of that.

I've seen them work, but point being I was referring to the OP who is seeing them running at 100C (i.e., near max temp) when just browsing with Firefox.

I was reinforcing that it sounds like his machine has something wrong with it.

 

[Mike Boreham]

You are right to admit you are not an expert. If there is anything on this Earth that should be identical, it is the speed of two of the same computer chips when they are clocked at the same rate. OK, yes you can test them and find some that can be clocked higher, but Apple does not do this. All of them are clocked the same.

Thanks for clarifying, I guess I was thinking about the binning process, which I thought might be about faster and slower chips, as well as separating chips with cores that don't work, and using them as lower spec chips disabling the non-working cores.

Maybe the actual chips are very consistent but their performance in different machines may vary due to installation factors like cooling. I don't know if Silicon integrated chips have a thermal paste in the heat path but this could be a source of scatter.

 

[Mike Boreham]

i don´t expect anything.

It sure sounded like it from your post I quoted!

 

[theorist9]

my M4pro was the other day at a CPU temp of 102°C.
on all 10 P-cores, give or take 2°.

What was i doing ?
FF was open with something like 11 tabs.
I had something like 5 tabs open from my internet provider. A major company in that biz here in europe.
They have something vERY fishy going on for sure. (since quasi same happened two days in a row. which also means: it´s reproducable)

Did you check Activity Monitor to see if Spotlight was reindexeing? No idea what's going on with your machine, but I'll note that Spotlight reindexing can run for a while, and uses multiple cores.

 

[Melbourne Park]

Maybe the actual chips are very consistent but their performance in different machines may vary due to installation factors like cooling. I don't know if Silicon integrated chips have a thermal paste in the heat path but this could be a source of scatter.

Thermal paste is used in all CPUs to transfer heat to heat sink which is itself cooled in these notebooks by air flowed aided by in the Pros two fans. Thermal paste allows the heat sink to be attached and also to be separated although that is unusual in notebooks. In intel Mac Pros upgrading the CPUs was a common practice and involved removing the heat sinks. It's normal in PCs too. But not in Mac Ms though.

The base 14 has one fan only. The 16 models have better cooling than the 14, including the twin fan 14 Pros. These M4 chips have a higher clock speed and they run hotter than the M3 chips do, results from tests have established this. Since they run hotter cooling is more important in these chips than in the M3s. Hence they also use more battery wattage when being highly utilised. But while they draw more wattage, they get the task done more quickly, so the stress stops earlier and hence overall they use less wattage to achieve a similar task. We know this from user comparisons on line, but so far we do not even know the transistor count of an M4 Pro or Max. We know that the M4 has 28 billion, compared to the M3's 25 billion and the M2's 20 billion.

 

[Melbourne Park]

Did you check Activity Monitor to see if Spotlight was reindexeing? No idea what's going on with your machine, but I'll note that Spotlight reindexing can run for a while, and uses multiple cores.

The issues have been for some time though.

But I'd reformat and do that speed test again. Running on the only browser factory installed, straight away after software install. Then get a replacement from Apple.

Another thread has show people with some scratches or marks on their M4 displays out of the box. Some have said its best to inspect one's screen with a torch to be sure your's doesn't have a tiny defect. I suspect these things happen with newer models - but such tests are worthwhile when unpacking. The youtube inboxes don't show that do they? Perhaps having the Nano option had disrupted the installation process somehow ... I recall one person finding a hairline crack. So it's possible that the first off the line are more likely to have QC issue than the last of the M3s would have had. Why not something CPU related? Or for that matter the motherboard or something defective attached to the motherboard.

 

[streetfunk]

Are you too late to return it?

Well, i had allready a phone call with apple in my country bevore of that "incident".

I got the machine delivered on a monday. I called them 2 weeks later, last day of the 2 week period, monday, which was by 100% within my return grace period. At that day, my apple shop account told me i have nothing i´m allowed to return. (oh well, i just understand right now, they might have counted things to the minute, and not just the calendar day as we would do in my country)

On the phone, they told me i can return it within the extended return period, so until 8.Jan.
"i think" the apple shop was stating same,=> but in reference to the day you order your machine, => and not in reference to the day of arrival.

I have the name of that apple sales person i talked to.
I ordered in the meantime a M4base. ( i have now a M4pro, 14c) ....delivery has a 2 weeks delay time right now.
So, i get my new M4 between 30 Dez and 10. of Jan.

I´m not sure how apple plays that game. As the money monger company they are.
But i will insist to be able to give my M4pro back.

on normal usages do i NOT have such temperatures. But i have higher CPU temperatures than i would expect to see.
I really would guess my machine has a bad cooling. As mentioend, a bad colling paste job might be the cause.


The 102°C degree incident was based on a fishy webpage from a 43 billion dollar company, my internet and phone provider. I have no clue whats going on in the internet. But it´s shocking...my internet and phone provider is a subbrand of "liberty global". Their black friday deals page was the cause ( 5 tabs or so open, each with one of their black friday deals pages open)

otherwise, my M4pro/14c CPU runs a 10-15°C higher temperature than my M2pro/12c would do with the same things open/parked idle. But hot on all P-cores cores ! Thats so not he case on my M2pro.

I´ll see how i come along with apple.

Thanks for feedback.



To see that normal people with normal usags have had high geekbnech ratings was important !
So, it´s definitly not just frogorated minis

Did you check Activity Monitor to see if Spotlight was reindexeing? No idea what's going on with your machine, but I'll note that Spotlight reindexing can run for a while, and uses multiple cores.

It has been in use for a week or so, bevore i made any GB tests.
and close to 3 weeks in use when the aformentioned ""FF-only/102°C" incident happened.
Not checked the activity monitor

 

[Mike Boreham]

Thermal paste is used in all CPUs to transfer heat to heat sink which is itself cooled in these notebooks by air flowed aided by in the Pros two fans. Thermal paste allows the heat sink to be attached and also to be separated although that is unusual in notebooks. In intel Mac Pros upgrading the CPUs was a common practice and involved removing the heat sinks. It's normal in PCs too. But not in Mac Ms though.

The base 14 has one fan only. The 16 models have better cooling than the 14, including the twin fan 14 Pros. These M4 chips have a higher clock speed and they run hotter than the M3 chips do, results from tests have established this. Since they run hotter cooling is more important in these chips than in the M3s. Hence they also use more battery wattage when being highly utilised. But while they draw more wattage, they get the task done more quickly, so the stress stops earlier and hence overall they use less wattage to achieve a similar task. We know this from user comparisons on line, but so far we do not even know the transistor count of an M4 Pro or Max. We know that the M4 has 28 billion, compared to the M3's 25 billion and the M2's 20 billion.

Thanks for info, but not sure how it answers my suggestion that paste could be a source of machine to machine variability in benchmarks.

 

[throAU]

Did you check Activity Monitor to see if Spotlight was reindexeing? No idea what's going on with your machine, but I'll note that Spotlight reindexing can run for a while, and uses multiple cores.

Spotlight should not be using P cores at all. It's a background process and should be limited to the E cores, that's what they're for.

On M series - Background processes do not get scheduled on P cores.

 

[Melbourne Park]

Thanks for info, but not sure how it answers my suggestion that paste could be a source of machine to machine variability in benchmarks.

It think it was myself who first hypothesised in this thread about thermal paste!

I checked though and its applied by robotic machines. Also statistical checks would be done to ensure QC. I presume all Macbooks use vapor chamber heat exchanges, and that is mature technology. Apple is a huge notebook maker and would do that R&D themselves, but perhaps their designs' manufacture are outsourced, but that would not lower the quality IMO.

I doubt cooling is an issue at all with this computer. The issue started because a test for single core speed test was particularly low. That was the original complaint along with perhaps a lower than expected multi core speed test result.

In particular the single core result was considerably slower than it should have been. I thiink 15% more than an M3 Pro. And a single core test of a minute or so doesn't stress today's multi-core CPUs. Since the CPU would not get hot in a single core test, and the single core result was the one which was very low, the poor result could not have been a cooling issue.

Later the user provided information about high temperature readings - maybe a cooling issue at that time? But I doubt it. There would be ways to check all that. But it sounded to me like the user is getting their money back. Also they seem to be buying a standard spec M processor Mac rather than a Pro model.

Good luck to the buyer for next time and I hope things go well.

 

[theorist9]

Spotlight should not be using P cores at all. It's a background process and should be limited to the E cores, that's what they're for.

On M series - Background processes do not get scheduled on P cores.

Good point, I was thinking of my Intel Mac.

 

[aibloop]

OP, just send it back… You are unhappy with it. Regardless if it is a lemon or not, or if the apple elves dropped it on its head… It doesn’t matter, you dropped a chunk of money on it and you should be getting what you expect from a new machine. Something isn’t right w your purchase in your opinion, so return it while you can.

I tested the m4 pro vs the m4 max. Granted, I could only test on laptops cause the Apple store had no mac mini m4 pro display model. I DID test the 16inch models due to the better more comparable thermals.

Based on me testing a m3 pro extensively and the m4 pro and m4 max @ the applestore, I think what makes a Pro "feel" worse in comparison to a max is likely the memory-throughput difference. It was particularly telling in Blender.



At the applestore, I ran a motion project in 4k, looped a usdz animated file and duplicated it + hit play. This shows what kind of realtime performance the machine is capable of and it shows a fps counter.


M4 pro did 34 instances of the butterfly at around 24fps and lagged the entire experience.

The m4 max did 80 butterflies @ over 70fps and felt responsive with headroom in performance.





Now, the M4 pro, and even the m4 are awesome. If you do not plan to actually use it for doing heavy editing or animate and render large 3d scenes in blender or sculpt huge 3d models. You will likely be happy w a M4 or a M4 Pro. However, do not think that the M4 Pro is anywhere near the performance of a M4 Max in actual usage with heavy files moving around, it just isn’t…

Synthetic benchmarks will show that a Mac Mini M4 Pro is awesome at a single isolated task where the memory throughput is trivial, which could be you sitting in Photoshop or lightroom, having an awesome experience. But if your file sizes go up and involve 3d… You will not be having as much fun.


I was trailing the mac mini m4 pro, but after going to 48gb/64gb ram its too close in price to a future base mac studio m4 max.. I think I would feel like OP that I got the worse deal if I bought a m4 pro chip.


So I will be either getting a 16inch m4 max macbook pro or I wait for the M4 max Mac Studio. Unfortunately.