Early reports say between 2-4% on the lower end of fps.. So it's not improving anything on the higher end, just making fps more consistant.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Major gaming news at WWDC: Metal and New games
- Thread starter Homy
- Start date
- Sort by reaction score
The issue is that on Windows you can basically draw whenever, and the default behaviour is basically just to YOLO as many frames at the screen as possible in an infinite loop, which is why Windows games will end up rendering like 1000 fps on loading screens etc.
MacOS's standard way of doing things is more of a "don't call us, we'll call you" affair where the game says "Hi, I'd like to render at 60 fps please" and then 60 times a second macOS says to your game "Hey game, draw your frame now"
The problem is that before game mode that callback was pretty weird and janky. Hopefully now it's much less janky.
Fun fact: Back like 10+ years ago when John Carmack ported Doom to the iPhone he essentially fixed Apple's janky screen presentation code, and most games went over to using John Carmack's custom loop instead of Apple's because it had much better frame pacing.
MacOS's standard way of doing things is more of a "don't call us, we'll call you" affair where the game says "Hi, I'd like to render at 60 fps please" and then 60 times a second macOS says to your game "Hey game, draw your frame now"
The problem is that before game mode that callback was pretty weird and janky. Hopefully now it's much less janky.
Fun fact: Back like 10+ years ago when John Carmack ported Doom to the iPhone he essentially fixed Apple's janky screen presentation code, and most games went over to using John Carmack's custom loop instead of Apple's because it had much better frame pacing.
As usual there are different variants of GPU cards on PC. There are 4 faster 4080s on that list.
But can it run Cyberpunk? 
Weird that Tsai gets 27 fps at 1440p high on M1 Max but LTT gets 12 fps at 1080p medium on M1 Ultra 64c. Maybe they should have waited for the shaders to complete like when Elden Ring went from 15 fps to 37. FF7 runs very well btw.
Weird that Tsai gets 27 fps at 1440p high on M1 Max but LTT gets 12 fps at 1080p medium on M1 Ultra 64c. Maybe they should have waited for the shaders to complete like when Elden Ring went from 15 fps to 37. FF7 runs very well btw.
Last edited:
No one ever listens to Time Traveling Space Wizard and our only insurance policy against the upcoming alien incursion John Carmack. This is the man who revolutionized 3D graphics and helped make consumer VR possible. Apple and Meta are fools to not listen to him.
But can it run Cyberpunk?
Sometimes I wonder whether LTT is running completely different settings from everyone else. I consistently get 40-50 FPS on high settings with Cyberpunk, which isn't bad given that it's running what's basically under double translation (x86 to ARM and DX12 to Metal).
Lmao? Did they forget to check the "64 bit binaries" button in Steamworks?
RISC vs. CISC, perchance. The only benefit to using x86 is less memory usage, hence smaller binaries, as has been long known by semiconductor engineers. CISC was a byproduct of the days when memory was very expensive. RISC processors require more memory to store the additional instructions needed to perform complex tasks. It's the same reason that macOS releases take up less space on Intel when compared to Apple Silicon. For instance, the recent Ventura 13.4 update was 2.7GB for Apple Silicon, while the Intel build came in at 1.55GB. (Not all of that is necessarily because of architecture differences, but it's certainly part of the equation.)
I'm just putting that forth as a possible explanation, the actual reason could be less mundane, but it's what I first thought of.
I think most Mac users would take the tradeoff of a few extra gigabytes on their SSD in exchange for the considerable benefits of Apple Silicon. (Excluding those who need the trappings of x86 PCs, such as Boot Camp for Windows, or GPU expandability, but that that ship has sailed and finally sunk at WWDC.)
Andrew Tsai made a reply to Linus' video
Essentially GPT is broken on M1 Ultra and M2 Ultra, while on the Pro and Max chips it works fine. It's clear the video was rushed and ill informed and even LMG admitted it themselves and are working on a follow up video that covers GPT in better detail
LTT is the temple at which PC partisans worship. I don't know why we should give them the time of day. Linus and his cronies have repeatedly shown that they are biased against the Mac, and use it as clickbait, to stir up their PC gamer audience that hates Apple. It's a great business model for Linus, but not helpful for those who want to use the Mac for something other than target practice with rotten tomatoes.
Loving the moves that developers are making. The Stray game looks good and it's a great starting game to diverge my current genre(simulation and base building games) and I may have to shell a money for the release on Steam or MAS but I am a little bit hesitant to buy it since it does not have Linux support. I just wish it has Linux support besides Proton. That said, I'm 70%-80% on buying it.
Umm...you know they've warmed up to Macs ever since the transition to Apple Silicon right? They even made an Apple focused channel that is arguably the best Apple Youtube channel ever.
Lately it seems that Linus just loves to hate on Apple with little to no research.
Any other scenario (not macs) we would see graphs of 5 different GPUs being tested. But they didn’t bother to test on any other Mac here? Why is that?
Doesn't change the fact that as a big tech channel they didn't bother to do proper research and did a lousier job than any average user on Reddit in their video about GPTK. That should have never happened when you have over 15 million subs and Emily Young (Anthony) is supposed to be their Mac expert. I agree with @Colstan and @Ethosik because it's not the first time they first do such a video to please their PC followers only to do another video later to clear some things up. It's all clickbating.
Last edited:
Um, because it's a proof of concept video and not a GPU review? There's lots of similar videos where they don't test on a range of hardware (e.g. the retro PC emulator video) because what would be the point? Honestly why would somebody expect the the fastest GPU available at the time of making to perform worse in the first place?
The point of the video wasn't how well does X AS SoC run games using Apple Game Porting Toolkit (Can we really not use GPT as an acronym here), it was AGPT runs Windows games!
Because if you do any research at all (which Linus didn’t) you will know that since M1 Ultra’s existence it is suffering greatly with scaling issues. Even leading to issues just like this. It’s a know issue with Ultra. So just one more Mac test with Max or Pro would have shown this to them.
And actually it was about how well the macs could run it as they were going on about the Ultra getting so low fps.
Last edited:
I wouldn't blame him for not anticipating issues with the M1 ultra, but Anthony is apparently the only Mac geek that wasn't aware of the good performance of the GPT. He got poor results and didn't even compare with reports that were all over the internet. Had he done so, it would have noticed that something was wrong.