Thursday, October 21st 2021
Apple M1 Max Beats GeForce RTX 3080 Laptop GPU in GFXBench 5.0, but Doesn't Shine in Geekbench
This should be taken with a fair helping of salt, considering GFXBench 5.0 is mobile device focused benchmark, even though the company behind claims it's a platform independent benchmark. Regardless of that, it looks like the new 32 core GPU in Apple's M1 Max SoC offers some pretty competitive performance, as it manages GeForce RTX 3080 Laptop GPU in said test.
However, this is a median score for the GeForce RTX 3080 Laptop GPU and many of the tests that make up GFXBench 5.0 aren't using DirectX, which is one likely reason for Apple's M1 Max GPU beating the Nvidia card. On the other hand, all tests seem to support Metal, which is Apple's 3D API, whereas the Nvidia card has to fall back to using OpenGL which tends to offer lower performance than DirectX in games. In most of the tests we're looking at an average performance advantage of less than 10 percent in favour of Apple, but it's nonetheless impressive considering that Apple hasn't been in the GPU business for very long.On the other hand, in a Geekbench OpenCL test, the M1 Max is losing out against a GeForce RTX 3060 Laptop GPU by quite some margin depending on the test. Neither is of course a real world scenario and we're going to wait a little while longer to see how well Apple's new SoCs really performs and more importantly, what the limitations are in terms of software compatibility.
Sources:
3DCenter, Geekbench
However, this is a median score for the GeForce RTX 3080 Laptop GPU and many of the tests that make up GFXBench 5.0 aren't using DirectX, which is one likely reason for Apple's M1 Max GPU beating the Nvidia card. On the other hand, all tests seem to support Metal, which is Apple's 3D API, whereas the Nvidia card has to fall back to using OpenGL which tends to offer lower performance than DirectX in games. In most of the tests we're looking at an average performance advantage of less than 10 percent in favour of Apple, but it's nonetheless impressive considering that Apple hasn't been in the GPU business for very long.On the other hand, in a Geekbench OpenCL test, the M1 Max is losing out against a GeForce RTX 3060 Laptop GPU by quite some margin depending on the test. Neither is of course a real world scenario and we're going to wait a little while longer to see how well Apple's new SoCs really performs and more importantly, what the limitations are in terms of software compatibility.
60 Comments on Apple M1 Max Beats GeForce RTX 3080 Laptop GPU in GFXBench 5.0, but Doesn't Shine in Geekbench
gaming pcs might be powerfull, but they are bricks, not energy friendly, and to work well they need a powerplug attached
Honestly, though, no one is buying a Mac primarily for games. They can get a $329 iPad for that, especially for the current state of gaming on Mac. Why people will buy it is for the applications that it will quite likely excel at. Like video work, which is quite popular these days. Or photography. An M1 Mac running DxO PureRAW can process a 20MP file in about 20 seconds. An M1 Max should significantly reduce that time, probably down to 5 seconds maybe? 20MP RAW is small potatoes these days, 50MP is high end now. If you are a pro photographer that shoots in the hundred or even thousands, the time savings is worth the cost. I bet you’d need a pretty stout GPU in a Windows machine to get the same result. I‘m curious to see just how well these chips can do with such tasks.
Competition from a company selling it's M1 bazinga laptop for $1400, haha, ok.
Perhaps it'd be different IF that APU was socketed and it could be dropped into future motherboards that allowed a separate dGPU, AND the 40CUs could still be used for separate compute, or rendering, or video encoding for streaming while your dGPU solely runs a game, AND/OR the 8GB of HBM2 could be used by the CPU entirely if the CUs weren't doing anything.... That could be a whole different animal, but that'd take a lot of work.
I dont have any experience with any kind of MAC, I do own their stock and eat at least 1-2 apples per day.
This stuff wont swing console gamers but it will convince quite a few people to buy an apple tablet that can game that over a PC / Android tablet that can't.
If this is anywhere near true, like say it gets the performance of a 3060 and not a 3080, well that's near true enough. It will mean the age of the iGPU has, after 25 years, finally arrived. It'll destroy the dGPU space and in particular nVidia. There may be many other side effects, like the end of full size tower PCs and excessively sized PSUs. High end GPUs are made for bragging rights and benchmark wins, without the bread and butter market of 3050 / 3060s the upper end will starve for lack of funds.
Edit: Very skeptical. A 3060 would be 4-6x faster than the fastest iGPU in the x86 world, in some cases more than that.
News websites: HURRRR M1 BETTER THAN 3090 durrRRAWRrRRRR
It's going to be something along the lines of what we see with H264/H265/HEVC hardware accelerated encoding and decoding: They made it fast for certain tasks, and it will BE fast at those tasks.
However, anything else? Well actually we'll never know because apple wont let people run programs that dont use their approved code ;)
Most apps are produced not for Apple but for smartphones at large. Good games and good devs are attracted to the largest target market. Windows and gaming on it, is still standing because of exactly that, and its also the reason Linux for gaming has still not taken off, despite strong Valve effort. Entertainment is expensive and needs large audiences to show its progress. Bigger, better, newer, more innovative is what drives gaming.
Look at how the M1 works right now and its clear this is not what Apple wants to do wrt gaming. Are they gonna code everything for Metal? How's that Vulkan support coming along by now...? And how are you going to pull those DX titles in? All their gaming efforts so far have made gaming on Mac, well... useless.
We can dream. Sure, but the real question is, is Apple the best caretaker of gaming on its platform... it has a worse history than Microsoft and a policy that totally doesn't fit, all gaming related efforts never really took off (Xbox the exception, but that's a console). Now look at Sweeney or G. Newell and their policies wrt platform management/distribution and how to capture publishers. They're everything Apple is not - they're even actively fighting Apple policy in court.
Also... regarding this failure of a gaming benchmark... the fact it falls apart in Geekbench and can't even catch a totally crippled Nvidia GPU not built for the workload tells us exactly nothing.
Nice soundbite for the dimwits who still see Jobs in their dreams. Gaming is a different beast and Apple is not the best safari for it. That I can get into. Sure it will, and with that it will never evolve beyond casual (browser/smartphone-type) gaming.
On a side note Apple has done everything to destroy macgaming...
1) Metal API.
There are two major APIs out there: DX12 and Vulkan.
They are both low-level API so they much better utilize hardware than the old and abandoned OpenGL and the not so abandoned but dying out DX11.
DX12 works only on Windows and it's proprietary., so it's not really an option for any OS other than Windows
Vulkan supports all major OS and is cross-platform. And it supports all major features of DX12 if not more.
Thanx to Vulkan Linux users can now use Steam Proton or Wine+dxvk and play 80+% of all native Windows games on Linux with little to no performance loss. And Valve is pushing hard for Linux (or rather Linux+Proton) support especially now with their Steam Deck.
Apple decided not to adopt Vulkan and invented their own proprietary API- Metal.
It's worth noting though that Metal was released several years earlier than Vulkan's first release. However, later on nothing prevented Apple from embracing Vulkan when it got much better than Metal especially after seeing what wonders it did for Linux gaming, but Apple just didn't..
Keeping going Metal route instead of Vulkan was a huge mistake.
I'll let myself quote one guy from Reddit:
Metal is missing much of the newer functionality available in Vulkan and DX12. Vulkan isnt the easiest to kick-off in but it's one of the best APIs out there for multicore + added Ray tracing support now. Now mind you MoltenVK doesn't support any of these RT extensions as you can't really translate BVH intersections from one api to another on the fly (it's an intense process anyways). So could Apple put some skin in the game and add vulkan support ? Maybe. Will they do it to kill off their own Metal? eff no. So unless they add hardware Ray tracing accelerators to their Silicon, there is no effin way software ray tracing will render 40-60 fps on M1, M2, whatever in the next few years.
2) They decided to make a drastic move and change architecture from x86 to ARM.
Metal API was issue enough but now ARM? Developers of console\PC games just don't want to bother with this. The tiny market of Macbook fans who can afford a $4-5k laptop and even tinier % of these people who wanna play AAA games on these $4-5k laptops just isn't worth the hassle..
A: To sell on Apple's platform.
What are they fighting for?
A: To be able to sell through Apple's platform without restrictions.
Why?
A: Apple has a giant market that's lucrative to them. Steam and Epic games want to sell to apple and tap into that market (and future growth) without having to pay the apple tax. They MAINLY SELL GAMES. So clearly they disagree with the "never evolve beyond casual (browser/smartphone-type) gaming" assessment.
Microsoft, a walled garden console makes up a huge portion of gaming market, Sony, another exclusive, walled-garden platform makes up a huge portion of gaming. These aren't dreams lol -- Sweeney and Newell are great but they are a minority. Apple isn't going to ignore a $70bn market that they already have a captive audience in on Macs/ipads/appletv. AAA titles on their hardware are for sure in their sights.
They can easily hit all 3 of those markets and are already a huge player in the Mobile. There's another 70bn on the table for them to port content to Metal / or support Vulcan - there will at the very least be an attempt.
Doing a fair benchmark across different platforms is tough, which is why technical details of benchmarks is crucial. If e.g. a benchmark is comparing something well written for API #1 vs. something poorly written for API #2, then it's really benchmarking the implementation rather than the hardware or the APIs.
Realistically what we are seeing is platforms for digital distribution are looking to become platform agnostic - Apple is the odd one out here because of the way they design hardware.
And when they produce the software too, we know we are looking at a horribly unfair marketplace. Since the EGS lawsuits devs/publishers have that confirmation too. Let's get practical about those numbers and we can distill what's really the desire.
Notice the marketplace that drives mobile revenue - 63.6 Billion of the 70 we're speaking of, is made in smartphone games. By definition as casual as you can get, even just because of the device limitations and use cases / typical audience that cares to play/spend a lot of time on a 5-7 inch mobile screen. Its clear EGS and Steam want a part of that massive potential, sure. But they're not porting the real (PC/console) games to smartphones. They're going to make mobile versions of them, simplified, MTX'ed, etc. This whole affair is about making easy money by using your IP and existing assets to turn them into smartphone games, a'la Diablo. Fortnite characters in what is otherwise a time grinder, ka-ching. You can look at one of the thousands of Chinese spinoffs on mobile, that more often than not just copy over libraries of assets and add a Chinglish looking title to it. If you start looking at them, you just start recognizing more and more assets you just 'simply know from somewhere'.
So in that market, these real game IPs have an immense market value. People who never got into those 'real' games can now taste the 'real' thing, the characters they heard about and the worlds they never got into because it was out of reach. In their minds, all of that 'is coming to smartphones'. And you have to consider the generational push too. Young generations that know only smartphone and some arcane world of PC gaming for the boomers. The timing is no accident, just like all those remasters and reboots aren't.
So that's really what the push is all about, but for Apple, (as you could see with their intent (marketing) with Arcade...) they want to bring the quality games to their store and 'launch real gaming', much like how Epic is a bit more stringent on curation (and how quickly that's falling apart... another thing these two tech companies can't seem to escape :D). And for that 'real gaming'... the market isn't 63.6 Billion worth of potential. Its that PC/console number you need to look at.
Oh wait, I can't.
Which customer did AMD lose?
And in case you missed it, AMD has been supplying Apple with GPUs for pretty much a decade. Now that Apple is moving away from x86, they are also moving away from AMD. Hence, AMD lost a customer in Apple.
I see.
Contrasting $1400 notebook by Apple with $200 chromebooks is a weird way to see things.
Even in US, Apple accounts for about 16% of PC's sold (units).
Much less so worldwide.
And while there is another extreme, chromebooks, vast majority of notebooks sold are neither Apple, nor chromebooks.
And, while yes, AMD has also lost a customer (it was mostly Intel), only a fraction of Apple's computers were equipped with AMD GPU anyhow.
And, oh, the irony, with current market trends, AMD might be happier NOT selling GPUs to Apple.... :D