Does M3 Pro / Max Really BEAT THE BEST Intel/AMD CPUs?

[@MrSamPhoenix]

Synthetic benchmarks usually don’t tell the story. Do you have a video showing the real world performance of these chips?

[@DoItHowTo]

first video of yours I've seen I look forward to watching your channel grow!

[@ocloredmind4973]

The thing is, Apple M series of chips are built using the ARM architecture, which is very efficient but not that powerful. AMD and intel however, use the x86 architecture, which focuses on raw performance and speed at the cost of power of consumption.

[@thmo_]

A nice comparision, but useless since you're comparing just benchmarks on two completely different architectures. These benchmarks are nice to compare ARM to ARM, and generations of those architectures with each other (intel 12th vs intel 13th, etc.), but useless in this specific case. Only real world application and use cases are meaningful for this and useful for people contemplating about buying either or.

[@andysbowden]

Apple Silicon is power efficient because it is a mobile SoC design. x86-64 is not a mobile SoC, it’s a general purpose CPU/APU. This seems to be forgotten when comparing the M series with x86-64. Single core clock speed is 4.05Ghz. This doesn’t change across the family. the M3 is the same as an M3 Ultra. AS doesn’t throttle regardless of running on battery or plugged in. A 14 gen Intel and equivalent AMD will throttle on battery but eat a M* when plugged in. I’ve got a M1 Max 16”MBP. I hardly ever use it on battery. 99% of its life it’s connected to my display and powered by mains.

[@PKperformanceEU]

Who cares. You re typical introvert incel who cant make connections/speak up right front

[@xpowerchord12088]

They are desktop class chips on the laptops in terms of power, they would not fit in a phone. The IPC on the Apple Silicon destroys x86 chips. An x86 at 4.05ghz processes 2 -3 instructions per clock, on apple silicon it does double that. There is so much more to this than clock speed or cpu. Apple ssd and ram speed and how it caches information is a totally different beast.

[@frostgodx]

Have a drink of water my friend

[@seamon9732]

A comparison using synthetics is already horrible. But to use Geekbench? Is that an intentional insult to viewers intelligence?

[@HuguesBalzac]

If your main use case is running benchmarks... And I have known a few people like that tbf.

[@SpeedKillersGaming]

Need to see productivity and gaming benchmarks.

[@clubnine-h1u]

There is also the price to consider. My designer need's a new computer, lot's of CAD and 3D rendering. He wants mac, but there is a more than 1000€ difference in my country with Mac Studio M2 Max and Ryzen 9 7900 desktop with RTX A2000 12gb GPU for examp. The power consumption vs performance isn't that bad with this Ryzen build either...

[@cheungching9430]

which one you finally ended up for your designer? I am in the same situation...

[@MelodicTurtleMetal]

Not a hard choice. If he's doing professional work without needing mobility get a amd/intel workstation, you'll get vastly better render times and overall performance. If he wants to be somewhat mobile then the apple choice is much better, with decent performance without needing to be plugged in.

[@johnsolo1701d]

3:34 This is comparing 8 power cores on the M3 vs 12 power cores on the 10600.

[@TheEvox81]

No. This is comparing 8 cores on the M3 to 6 cores on the 10600K. Thanks for playing though.

[@johnsolo1701d]

@@TheEvox81 Cinebench uses threads

[@TheEvox81]

@@johnsolo1701dI know. But hyperthreading threads are not equivalent to additional cores. In the BEST case, hyperthreading adds about 20% performance over the single thread equivalent. That would make the 10600 a 7.2 core, to Apple's 8 core. The 10600 still has fewer core "equivalents" and is still faster.

[@EduardoMarchesan-z1k]

@@TheEvox81and its 4 years old processor

[@boganx1291]

That guy really invest on case

[@MrMastadox]

Not an Apple to Apple comparison. But ok. The M3 kicks ass for mobile devices. The reason Amd and Intel can't get close to it efficiency wise, is because their chips are still x86 chips. In other words. The processors are completely different, because the x86 processors are CISC. Complex instruction set computing chips. They need to be able to perform so many different instructionsets , which makes them far less efficient compared to ARM procesors like Apple is using now. For AMD and Intel to match , they would need to make completely different processors. But they can't just do that. Because their markets use software that needs x86 instructionsets

[@bendikarbogast1229]

I'm sorry to bring it to you, but CISC bad, RISC good is not the full picture and very misleading at best. This is completely disregarding any differences in micro architecture, cache design/layout, big little architectures, how sleep states are handled and processing node/manufacturer which have way larger performance and efficiency implications than the supported instruction set. There are a lot of good arguments to be made for Apples architecture, in all of the previously mentioned categories, the ISA in particular though is brought up way to often in this context.

[@cj09beira]

Not really true, its more linked to this small fact that apple is usually a full node ahead of everyone else, giving them a free efficiency boost, the rest is down to packaging and connectivity reasons, as having all the ram on board the cpu socket lowers power usage, and so does the lack of pcie interface for dedicated gpus, etc.

[@MrMastadox]

@@bendikarbogast1229 where did i say its bad? It just is different. It serves different purposes. And has to run software that uses instructions that require x86 architecture. You are stawmanning. The inefficiency comes from making sure the cous are also still compatible with old software. Intel tried to get rid of that before. Its now suggesting a 64bit only version called x86-s. Which will reduce compatibility with older 32 bit software. Anyway. You are reading things i didn't write. You ignore how i said you can't compare the 2. Not an apple to apple comparison. But that you simply can't compete on efficiency if you are running a chip that has to support so many different instructions. Its simple logic. Like mining on gpus. When mining is your goal. Having an ASIC will be far more efficient because it is designed to only have to do specific instructions.

[@MrMastadox]

@@cj09beira no. The huge difference in efficiency here is not the node. It is mostly the huge amount of instructions the x86 cpus have to support. Intel tried to go to a different architecture before. And is now trying to skim it down by suggesting a x86s architecture that is 64bit only. Cutting out the 32bit compatible instructions. The arm cpu is just far more specialized for specific compute tasks. Where the old x86 has to be able to do so many different types of instructions and be compatible with old software. It is wildly inefficient compared to an ARM cpu. Like mining crypto on gpus is inefficient compared to an ASIC cpu which has a more simple architecture that is only focused on a few instructions. ARM cpus are just more efficient because they don't have to be compatible with age old instructionsets for compatibility with old software. A newer node surely can help with efficiency. But the biggest efficiency difference comes from x86 cpus having to be compatible with a shit ton of instructions.

[@bendikarbogast1229]

@@MrMastadoxTo quote your comment word by word: "They need to be able to perform so many different instructionsets (instructions, but ok), which makes them far less efficient compared to ARM processors like Apple is using now." You clearly took that position, my argument is therefore not a strawman as you pretend it is.

[@georgi.georgiev]

Although a nice and interesting comparison, I think you underestimate a bit the computational and rendering powers of the M3's. For example, if you try Cinebench R24 - you'll get a completely different picture due to the better optimization for the M3's. Also, when M3 Ultra goes out, which is the real and only desktop grade Apple CPU - this once again will paint a totally different picture. As you said - Intels can undervolt and underclock. So, if you want a fair comparison - undervolt and underclock and Intel to the levels where it becomes at least barely usable in a laptop chassis and test again.

[@TheEvox81]

"bUt ThE eFFiCiEncY" No one cares. When I need to do work, I don't care about my processor's TDP. What I do care about is having the performance I need, when I need it. Don't get me wrong, when you don't need 100%, there are benefits to the efficiency... Less heat and better battery life, to name a couple. But make no mistake: if Apple could simply crank up the power and blow x86 away, they would.

[@therealcouchpotato9560]

Apple is getting most of their processing wins from the other stuff on the SoC, namely the media engines that carry the slack for the GPU and the CPU so you can get crazy fast rendering times on programs that are optimized to take advantage of them.

[@atnfn]

Lol, undervolt... yea cripple the intel chips so Apple can win. Just appreciate that Apple makes the most power efficient chips by far instead of being a fanboy and demand they win at everything. (btw all my computers have AMD chips, so I'm certainly no intel fanboy).

[@MelodicTurtleMetal]

The M3 isn't just a processor though. Power efficiency is really the only win it has. Much of its advantage over intel or amd only apply if those systems don't have dedicated graphics. As soon as you shove a decent gpu in those machines the M3 stops competing everywhere but performance per watt. Which is great for a mobile device - but for a workstation? At the consumer level Apple make a great product - at professional and enterprise level they're a joke

[@adam872]

Base level performance of the M series chips is seriously impressive. Multi-core performance I've learned over the last couple of years really depends on the composition of the core count. Each of Intel, AMD and Apple all do slightly different things packaging wise and their sweet spots performance wise are also a bit different. They're all pretty fast though. The discrete graphics performance of an nVidia equipped PC will be superior, however. I think everyone acknowledges that.

[@MelodicTurtleMetal]

Amazing what Apple have done with such power draw, especially considering the somewhat decent integrated graphics. It's a pity Mac's are borderline useless in most enterprise environments. Though with my requirements being lower these days I'm contemplating getting an M3 device for general use. Maybe.

[@adam872]

@@MelodicTurtleMetal the funny thing is that I've used Macs or seen them used in multiple enterprises since OSX came out a couple of decades ago. Supporting a large fleet of them in a business would definitely take more up front effort, but ongoing support in my experience has been less than with Windows. The only drawback for me would be the lack of compatibility with some nice engineering apps I've used over the years. Most workflows seem to function just fine though, certainly for your regular productivity type users.

[@MelodicTurtleMetal]

@@adam872 networking, group policy, and software rollouts are a pain in the ass too. If you're not using typical Adobe software then usually you have a compatibility problem. Though I'm seeing a huge move towards services like Azure which, while often annoying, removes the compatibility issues and it's frankly no less annoying on windows than Mac. Another 5 years and they might be able to coexist in a workplace.

[@adam872]

@@MelodicTurtleMetal it's definitely a different way of thinking if the existing environment is all Windows and running a traditional domain. That said, if we think about computers in the same way we do mobile devices for management, InTune, Jamf or Airwatch makes managing Macs (and PC's to be honest) reasonably easy. All the settings are managed, printers, software rollouts etc. For an old time Unix guy like me it's a pretty short jump. I've got several clients that have moved to Azure AD + InTune and managing Macs in that scenario isn't super hard.