First M3 benchmarks show big speed improvements over M2

Posted:
in Current Mac Hardware edited January 24

At its event on Monday evening, Apple unveiled the M3, M3 Pro, and M3 Max chips, and recent benchmark tests have provided data to support Apple's claims for speed.

Apple revealed the next generation of Mac processors
Apple revealed the next generation of Mac processors



Apple revealed the new M3 chip series at its "Scary fast" event on Monday, which was centered around the Mac. Before the initial orders reach customers next week, preliminary benchmark results will offer further insights into the M3 processor's performance.

Apple states that the entry-level M3 chip has an 8-core CPU that performs up to 35% quicker than the M1 and 20% faster than the M2. Additionally, its 10-core GPU is as much as 65% speedier compared to the M1 and about 20% faster than the M2.

New figures in the Geekbench database corroborate what Apple has asserted. For example, a Mac with an Apple M3 processor shows a score of 3,030 for single-core performance and a multi-core score of 11,694.

The data shows that the M3's performance metrics are notably higher than its predecessors. Specifically, a listing for an M1 iMac shows scores of 2,334 and 8.317 in single-core and multi-core performance, respectively.

Additionally, a Mac mini with a CPU of eight cores scored 2,631 and 9,742 in core performance.

While benchmark scores may not always accurately reflect performance in everyday use, they do offer a valuable means of verifying some of Apple's statements regarding performance improvements.



Read on AppleInsider

Alex1NFileMakerFeller
«1

Comments

  • Reply 1 of 32
    5nM/3nM = 1.6 recurring, suggesting a move from the 5nM process to the 3nM process would yield a 67% improvement in speed/power ratio. We are not seeing that.

    If you delve into the TMSC public documentation on their timelines, we see it is far more complicated than the headline figures suggest. The current "3nM" process as advertised by Apple for both their A17 series and M3 series is merely a stepping stone from the previous "5nM" process, which also is not what its name says. There are probably two more generations of SOCs, As and Ms, before we actually arrive at something that can truly be called 3nM. By then, the talk will be about "2nM". 

    I am happy with my M1 MBA 16GB/2TB until at least the M5 comes along. 
    williamlondonradarthekat
  • Reply 2 of 32
    chadbagchadbag Posts: 2,013member
    timmillea said:
    5nM/3nM = 1.6 recurring, suggesting a move from the 5nM process to the 3nM process would yield a 67% improvement in speed/power ratio. We are not seeing that.

    If you delve into the TMSC public documentation on their timelines, we see it is far more complicated than the headline figures suggest. The current "3nM" process as advertised by Apple for both their A17 series and M3 series is merely a stepping stone from the previous "5nM" process, which also is not what its name says. There are probably two more generations of SOCs, As and Ms, before we actually arrive at something that can truly be called 3nM. By then, the talk will be about "2nM". 

    I am happy with my M1 MBA 16GB/2TB until at least the M5 comes along. 
    Not sure why you think gang speed or performance is exactly a match to the size.  It’s probably a much more complicated picture. 
    mknelsonwilliamlondonmike1gregoriusmnubuschasmAlex1Ntechconcrezwitsdanox
  • Reply 3 of 32
    canukstormcanukstorm Posts: 2,729member
    By the looks of it, the M3 is faster than the base M1 Pro model, and about just as fast as the M2 Pro base model in multicore.  That's impressive
    williamlondonthtchasmAlex1Ntmayd_2radarthekatargonautFileMakerFeller
  • Reply 4 of 32
    timmillea said:
    5nM/3nM = 1.6 recurring, suggesting a move from the 5nM process to the 3nM process would yield a 67% improvement in speed/power ratio. We are not seeing that.

    Not accounting for other changes and just looking at the size reduction ratio alone, I think the math would suggest that 3nM is a 40% reduction of 5nM, so maybe a 40% increase in speed?
    chasmAlex1NFileMakerFeller
  • Reply 6 of 32
    canukstormcanukstorm Posts: 2,729member
    higher clock speeds but consuming the same amount of power all thanks to 3nm technology
    williamlondonCookItOffAlex1NdanoxtmayradarthekatnetroxFileMakerFeller
  • Reply 7 of 32
    programmerprogrammer Posts: 3,461member
    timmillea said:
    5nM/3nM = 1.6 recurring, suggesting a move from the 5nM process to the 3nM process would yield a 67% improvement in speed/power ratio. We are not seeing that.

     
    LOL... that's not how this works.  Never has, never will.  For starters, the process number represents the linear dimension of the smallest feature that the process can create.  It does not apply to everything on the chip, plus it is a single dimension whereas chips are 2-dimensional.  In theory that means that this shrink ought to allow 2.8x as many devices in the chip (i.e. the number of transistors that is often quoted).  But chips are far more than just transistors, and indeed Apple's numbers mention "only" a 37% increase in transistor count (M2Max -> M3Max).  And the number of transistors does not linearly relate to performance either -- the reality is far more complex and nuanced.  Furthermore, performance is vastly more complex than just one number -- there are a mind blowing number of factors, and greatly depends on what software you need to run.  A benchmark gives only a vague snapshot of a computer's capability, unless what you plan to use it for is running that specific benchmark algorithm (which is virtually never the case).  Performance is a vast and complex topic, so thinking you can related it to the process number is simply naive.

    As for waiting for a particular process tech, that doesn't make much sense.  The continual steady onward march of process tech ended over a decade ago, and now transitions happen with more fits and starts.  They are enormously expensive, and bring diminishing returns or additional problems.  Predicting what is going happen next year is difficult enough, further projections are worthless at this point.

    Your M1-based Mac ought to do you well for years.  When it makes sense to upgrade should depend on when it stops doing what you need, or when Apple starts shipping a machine which has a new capability that you need.  This has very little to do with the process technologies being used to create it.
    tmaywilliamlondontenthousandthingschasmAlex1Nkeithwkkeerezwitsdanoxd_2
  • Reply 8 of 32
    rob53rob53 Posts: 3,267member
    higher clock speeds but consuming the same amount of power all thanks to 3nm technology
    Which is why Apple couldn't wait to get to the 3nm process. They didn't have to change the processors that much, they simply increased the clock speed because they could do that without burning the SoC up.  

    Editor: it would be nice if you created an actual chart that showed the Geekbench scores all together. Finding them in the article is a mess.

    Alex1Nd_2radarthekatoldenboomBart Y
  • Reply 9 of 32
    nubusnubus Posts: 477member
    tskwara said:
    Not accounting for other changes and just looking at the size reduction ratio alone, I think the math would suggest that 3nM is a 40% reduction of 5nM, so maybe a 40% increase in speed?
    3nm does allow more transistors in the same space, but Apple didn't do that. For M3 Pro the number of transitors got reduced compared to M2 Pro. As a result M3 Pro isn't faster than M2 Pro. M3 base is stuck on Thunderbolt 3 that started shipping in 2015 and Apple decided to limit it to 1 external display. The original iMac got USB before PowerMac, but these days Apple is fully nerfing the iMac.

    iPhone 15 / 15 Pro both got more memory at no extra cost. Not so with Mac. And the Apple Tax is now at +300% for a memory upgrade.
    williamlondonAlex1N
  • Reply 10 of 32
    sbdudesbdude Posts: 276member
    I don't think it matters that the performance increase comes down to clock speed versus architectural improvements. That is, until we get to the next fabrication process. We're getting to point of diminishing returns on node reduction, and TSMC has already said its gate all around transistors are more difficult to fabricate. They may have won this round, but future rounds are going to be hard fought.
    Alex1NkkeetmaytenthousandthingsFileMakerFeller
  • Reply 11 of 32
    rob53 said:

    Editor: it would be nice if you created an actual chart that showed the Geekbench scores all together. Finding them in the article is a mess.

    AMEN!
  • Reply 12 of 32
    As if there was a performance problem. I never had performance issues with any Intel/M powered mac. Even my 8 year old Macbook Pro still handles any workflow. It reminds me of another Apple obsession making devices thinner. Apple should focus on more usefull features like extended battery life, satellite internet, cloud backup, touch screen macbooks etc
    dewme
  • Reply 13 of 32
    tskwara said:
    timmillea said:
    5nM/3nM = 1.6 recurring, suggesting a move from the 5nM process to the 3nM process would yield a 67% improvement in speed/power ratio. We are not seeing that.

    Not accounting for other changes and just looking at the size reduction ratio alone, I think the math would suggest that 3nM is a 40% reduction of 5nM, so maybe a 40% increase in speed?
    You guys are referring to a concept called "Denard Scaling", but that hasn't actually held up since the late 1990s, and it took a nosedive in the 2000s after 28nm.  Today's denser processor nodes lead to only single digit percent increases in performance.  Some of this is because voltage scaling slowed down.  Some parts, like caches, haven't shrunk at the same rate.  The capacitance of wires and devices is going up.  I remember NVIDIA was freaking out after the 28 nm node, because following that, even the cost per transistor wasn't going down.
    Alex1NtmayrezwitsradarthekatFileMakerFeller
  • Reply 14 of 32
    I am relieved at seeing the new chips, since I just purchased a Mac Studio M2 Ultra a couple of weeks ago. I wanted the high speed double-bandwidth memory and highest speed SSD only available in the M2 Ultra, wow it's fast. I figured there would be no M3 Ultra at first, maybe not for a full year, since the Mac Studio was only recently updated to the M2.
    For once, I predicted correctly. I usually tell my friends, watch and see when I buy a computer, because that's always the wrong time to buy a new computer.
    Alex1Ndanoxd_2Bart Y
  • Reply 15 of 32
    chasmchasm Posts: 3,411member
    The point of this article — which a number of commenters here appear to have missed outright — is “does Apple’s speed claims hold up under independent testing?”

    The answer, for the base M3 at least is “yes it does.”

    That said, some of the comments are informative regarding the 3nm process, which is appreciated.
    Alex1Nmobirdwilliamlondond_2dewmemuthuk_vanalingammichelb76FileMakerFellerBart Y
  • Reply 16 of 32
    auxioauxio Posts: 2,746member
    dutchlord said:
    As if there was a performance problem. I never had performance issues with any Intel/M powered mac. Even my 8 year old Macbook Pro still handles any workflow. It reminds me of another Apple obsession making devices thinner. Apple should focus on more usefull features like extended battery life, satellite internet, cloud backup, touch screen MacBooks etc
    The ole, "everyone is the same as me" argument.

    Are you building machine learning models? Making movies in 8K? Building high resolution 3D models? Then did you consider that perhaps you don't actually need to buy a MacBook Pro? The word "Pro" is unfortunately just a status symbol for a lot of people who don't truly understand what it means.
    canukstormAlex1Nrezwitskkeewilliamlondonroundaboutnowmichelb76h4y3ssphericprogrammer
  • Reply 17 of 32
    XedXed Posts: 2,703member
    auxio said:
    dutchlord said:
    As if there was a performance problem. I never had performance issues with any Intel/M powered mac. Even my 8 year old Macbook Pro still handles any workflow. It reminds me of another Apple obsession making devices thinner. Apple should focus on more usefull features like extended battery life, satellite internet, cloud backup, touch screen MacBooks etc
    The ole, "everyone is the same as me" argument.

    Are you building machine learning models? Making movies in 8K? Building high resolution 3D models? Then did you consider that perhaps you don't actually need to buy a MacBook Pro? The word "Pro" is unfortunately just a status symbol for a lot of people who don't truly understand what it means.
    He even backs up his claim with "Even my 8 year old Macbook Pro still handles any workflow" instead of "Even my 8 year old Macbook Pro still handles any workflow I have" or "Even my 8 year old Macbook Pro still handles my workflow."

    And let's not forget the benefits of the M-series chips for performance-per-watt... but I guess that's just another weird Apple obsession. LOL
    roundaboutnowradarthekatwilliamlondonauxiosphericFileMakerFellerBart Y
  • Reply 18 of 32
    At this stage, I find 20% a speed increase meaningless and abstract.

    To me what matters is: what does it enable me to do what I couldn’t do before?

    That answer on a generation A to generation B basis usually leads to: not much.
    On a multi-generational skip that starts to answer that question: a much better battery life, exponentially faster video editing and exporting.

    Even then there’s the question of when you really go ‘pedal to the metal’ and you find out these are usually short sprints.
    At this stage the baseline CPU usually delivers the best cost to performance ratio.
    radarthekatwilliamlondon9secondkox2
  • Reply 19 of 32
    Looking forward to working on one of these new machines. I miss my M1 Max (previous work machine).
    radarthekatBart Y
  • Reply 20 of 32
    bluefire1bluefire1 Posts: 1,304member
    When is enough speed enough speed? For most people except maybe gamers, the M2 chip is more then enough.
    edited November 2023
Sign In or Register to comment.