The surprise is that it’s only the second level chip (Standard, Pro, Max, Ultra) but that the M4 Pro outperforms both the M3 Max (there is no M3 Ultra at this point) and the M2 Ultra (a chip with almost twice as many cores as the M4 Pro).
Not true, obviously: M4 Max in the new Macbook Pros has been benchmarked to be faster [1].
Its not just a little faster than the #2; looking at ~15-20% uplifts in both single and multi compared to the previous kings. The previous production single core #1 was ~3100; Apple skipped over most of the 3000s right into the 4000s.
The article itself directly acknowledges this saying “of course this will be surpassed once we can benchmark the m4 max” (or something similar).
I wonder if their rules are “we only announce benchmarks we’ve done ourselves”? Coupled with “well we need some readers today and our MacBook order hasn’t arrived”
Oh I didn’t look at the linked benchmark because I assumed that the person who posted it would have ensured that the compared “already existing” benchmark was already existing when the article was posted. Derp.
So the new Mini is faster than all existing Macs (at least on multicore Geekbench). That revelation sort of throws a monkey wrench into the grand plan of seldom buying a super-duper Mac that will perform competitively for years. Better to buy the cheapest Mac possible (with a Pro CPU) and leave the Studios and Pros to rich fools eager to part with their megabucks.
> Better to buy the cheapest Mac possible (with a Pro CPU)
As long as it has enough RAM, that seems like a sound idea.
I actually got the M1 MacBook Air to replace an old Linux netbook that was dying (and also needed the walled garden for a project). I think that the M1 would still be enough for my daily tasks away from the desktop for quite a few years... except that I got the 8 GB version, which in hindsight was a pretty big mistake, assuming the rest of the hardware and software will have decent longevity.
Now I'm looking for remote development environments, something I could maybe run on my homelab and remote into, or maybe just running development containers on them.
I'm still running an Intel-based MacBook Pro from 2019, and it's trucking along just fine. It's not as fast as the M* chips, obviously, but it's still perfectly serviceable for software development. I'm tempted by the new Pro, but I don't consider it a necessity.
If I picked up a new Mini or MacBook today, I'm reasonably confident it, too, would last me five years or more.
I switched from the same model three months ago. It’s night and day. I used to have my fan spinning all day; with the new one, I only heard it once since buying it, even in perfect silence. Things that used to slow down, now run in parallel. The battery keeps up with more than a full work day programming, comfortably.
It’s a big difference, but the battery of a new 2019 MacBook also got close to a work day. The M series should last a bit longer, but after 5 years it’s a gamble with batteries…
> Better to buy the cheapest Mac possible (with a Pro CPU) and leave the Studios and Pros to rich fools eager to part with their megabucks.
I'm not sure that quite follows. This benchmark is with the fastest (of 3) CPU options for the Mini. With 64GB of memory and 1 TB storage (to make a fair comparison with the Studio M2 Ultra), you're looking at a $2500 machine. We're not talking about the base $600 config here.
Granted, that's a lot less than the $4000 starting price of an M2 Ultra machine, but it's also 18 months later, and you don't get the GPU performance of an Ultra (see below).
> So the new Mini is faster than all existing Macs (at least on multicore Geekbench)
That's another thing: This only measures CPU performance.
The difference between "Max" chips and "Pro" chips lies in the GPU, not the CPU. The CPU core counts for Pro and Max are the same, but the Max has 2x as many GPU cores. ("Ultra" chips are basically two "Max" dies on one chip)
As you might expect, M2 Max and Pro have near-identical multi-core geekbench scores, but the Geekbench Metal score for the M2 Max is 80% higher than the M2 Pro (and M2 Ultra is about 2.7x the M2 Pro).
A 10-core M4 (w/ 10 GPU cores) scores about 58,000 on Metal (https://browser.geekbench.com/v6/compute/3039631), so the Mac Mini with 14-core M4 Pro (w/ 20 GPU cores) should be expected to score around 116,000 on Metal.
That's almost exactly the same GPU performance as a Studio M2 Max (which would have cost you $2600 last year with 64GB memory and 1TB storage)... If you paid $4000 for the M2 Ultra last year, you would still have comfortably 2x the performance of today's $2500 Mac Mini, and nearly the same multi-core CPU performance.
Geekbench multi falls off a cliff after ~16 cores. E.g. the Epyc 9654 with 96 cores benches lower than the Ryzen 7950X with 16 cores of the same generation.
A 1.6x single core performance difference won't negate a 6x core count advantage in peak multi core performance. The problem above is not that the cores are slower, it's that Geekbench will literally not utilize the additional cores in the first place. This compounds with what you're saying - the few cores that do get used have high clocks on the low core count optimized part but lower clocks on the high core count optimized part.
Compare this to a multithreaded benchmark that does scale to all of the cores and you'll find the higher core count CPUs are able to push significantly higher scores despite the single thread difference e.g. https://www.cpubenchmark.net/high_end_cpus.html has them at 62,711 vs 117,317 in the opposite ranking direction. That should feel about right, otherwise AMD would only make the 16 core high frequency CPUs instead of 128 core low frequency monsters.
That's not to say the Geekbench score is bad or useless. It represents a specific type of workload... just not "peak multi core performance". It's more indicative of "mixed workload performance", where the extra 2x cores on the Ultra are more apparently going to be irrelevant.
> Having such a large difference in single-core performance, will negate the sizable difference in total core count (96 vs 16).
But why? Wouldn’t total score be approximately corecount*corescore? Of course it’s not exactly that because not all cores run full speed at the same time, but how are the cores weighted that 16 cores are better than 96 cores with half the speed each?
Geekbench 5 and earlier constructed the multi-core test as essentially running N independent copies of the single-core test. This effectively pretends that every subtest is embarrassingly parallel. Geekbench 6 switched to having the multi-core test actually operate like real multi-threaded software: a fixed-size problem is broken up to be divided among available cores, with a non-zero amount of coordination between threads and potential for less than perfectly linear scaling because Amdahl's Law isn't being ignored.
But that’s a very specific thing to adjust a benchmark for, what if I want to host 50 VMs on one server for example? Then the 100 core server would be much better than the 16 core server, even though it has a lower benchmark value.
It's not a "very specific" thing to adjust a benchmark for. It's the default case for practically all consumer workloads, and Geekbench is a consumer-focused benchmark.
I played around a bit lately with finding ways to dramatically multithread code in golang, mostly for fun. What I found was that there was a threshold where the overhead of spinning up all the threads at the start and synchroninzing them at the end overwhelmed the time savings from actually performing the work in multiple threads.
It wouldn't surprise me if PDF renderer and background blur were fast enough tasks that spinning up 96 threads to split rendering across all those cores was a waste of time compared to how fast the actual task was to complete. It was akin to trying to hammer in 50 nails by getting 50 people and handing out 50 hammers and assigning each person one nail, then telling them "okay, start!", then inspecting everyone's work afterwards; at some point, it's faster just to break it into two or three tasks.
This was a surprise for me as well. I have two EPYC 9754 in a dual socket server, so 256 cores, and the test did not perform as well as I expected it too. It didn’t even load up all the cores, which is what I was needing to do.
I ended up using something else to generate the load I needed, but I can’t remember exactly what. I think it might have been a Monero benchmarking tool?
That's one way to view it. Another is the benchmark doesn't intend to measure the "peak multi-core CPU performance" the article assumes multi-core score is meant to measure. It's really measuring something more like mixed workload performance.
As a performance metric that does seem like it would be more valuable for lots of use cases, so measuring that seems good.
Maybe they need to add an additional “cpu bound multiprocessing perf”, and make it easier for professional tech reporters to understand complex concepts like benchmark numbers :D (in fairness to the reporters it does sound like the benchmark name legitimately implies that it’s a max parallel throughput benchmark, but if this is your job you should really know what your benchmarks are actually measuring).
Honestly a benchmark I would like to - which is more of a software/kernel/os/scheduling one - is “how responsive is this machine under heavy load”.
A non zero part of wanting that as part of a benchmark is that popular benchmarks often seem to be the only way to get companies to fix “uncommon” issues.
Laptops with AMD Ryzen AI 9 HX or Intel Ultra 7 look bleak performance wise and expensive price wise than compared with M4. Snapdragon X Elite is just a flop compared with M4.
Yeah, the standard Apple line "our best _____ ever" is really tiresome. It's so tiresome, it makes me tired even when the line isn't said by Apple itself but from its sycophants and evangelists
It’s not a given that a new thing is better than an old thing. Just ask Intel.
The news here is that the Pro isn’t just better than the previous Pro, it’s better than the previous Max and the previous Ultra. That’s impressive even if it’s expected that it will be better than the previous Pro.
The actual releasing of the stuff is not what I was referring when I specifically stated "the standard Apple line".
We know this is better than what you released previously, otherwise, why would you be releasing it. It's just over the top marketing/PR fluff that is over the top over used to the point it is meaningless approaching insulting
Seems that if you are into photography you still have to wait until M4 Max based Studio is out and wait out Mac Mini Pro. 2x graphics performance will accelerate many GPU-based tasks in tools like Topaz Photo AI, Lightroom and Photoshop, Luminar, etc.
Topaz I can understand, but are you really struggling with performance in lightroom and photoshop?
I have a 50MP camera and I don’t have issues with older hardware
Actually, I don’t know how like an M3 pro or better compares with a ryzen 5600 and rtx 3060, I assume much better cpu performanc and similar GPU performance?
