That’s not the point. The M1 is an incredible chip, offering unheard of efficiency, no one doubts that. That 4500$ graphics card though was made for purposes which aren’t even tested in those benchmarks. If they were tested, the M1 couldn’t even score there, since it lacks that functionality in the first place. But that’s ok, they never were meant to compete with each other anyways. I dislike such comparisons for the pure sake of fanboyism.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Cinebench R23 with support for Apple M1 SoC is now out. Geekbench scores up too.
- Thread starter EugW
- Start date
- Sort by reaction score
That’s not the point. The M1 is an incredible chip, offering unheard of efficiency, no one doubts that. That 4500$ graphics card though was made for purposes which aren’t even tested in those benchmarks. If they were tested, the M1 couldn’t even score there, since it lacks that functionality in the first place. But that’s ok, they never were meant to compete with each other anyways. I dislike such comparisons for the pure sake of fanboyism.
Ah, yes, that you are 100% right about. Even running a gaming benchmark on a Tesla M10 is silly.
Last edited:
Or in other words, continue to use a MBA for the tasks people normally do on a MBA, and it’s still going to be the same experience, now with faster performance and longer battery life.
I find it amusing that people are complaining about the entry level MBA not being able to run triple-A title games or perform some other demanding tasks normally reserved for more powerful workstations, as though you could do them on an equivalent windows ultrabook either way.
Heck, that we are even having such a conversation shows that the M1 chip is going to punch above its weight and there is nothing the competition can do about it.
I’m telling people to have realistic expectations, otherwise we get endless moaning when reality hits. That GFXBench benchmark is ancient by modern standards of gaming graphics and so are Apple Arcade games. So keep it real and you’ll be happy with the new machines.
In graphic (not compute) tasks, the M1 could be on par with the Radeon Pro 570X (and way ahead of the 560X). The TBDR architecture of the M1 (for which Metal has been tailored) benefits graphics more than it benefits compute. Also, Apple GPUs can use 16-bit AND 32-bit numbers in shaders, for precision and to boost efficiency, which PC GPUs can't.
That's great! I suspected that since it can render more pixels/s:
M1 41 GPixel/s, 82 GTexel/s
Pro 560X 16.06 GPixel/s, 64.26 GTexel/s
Pro 570X 35.36 GPixels/s, 123.8 GTexel/s
Pro 580X 38.4 GPixels/s, 172.8 GTexel/s
Pro 5300 52.8 GPixels/s, 132 GTexel/s
Pro 5500 XT 56.22 GPixels/s, 168.7 GTexel/s
Pro 5700 86.4 GPixels/s, 194.4 GTexel/s
Pro 5700 XT 95.94 GPixels/s, 239.8 GTexel/s
That's great! I suspected that since it can render more pixels/s:
M1 41 GPixel/s, 82 GTexel/s
Pro 560X 16.06 GPixel/s, 64.26 GTexel/s
Pro 570X 35.36 GPixels/s, 123.8 GTexel/s
Pro 580X 38.4 GPixels/s, 172.8 GTexel/s
Pro 5300 52.8 GPixels/s, 132 GTexel/s
Pro 5500 XT 56.22 GPixels/s, 168.7 GTexel/s
Pro 5700 86.4 GPixels/s, 194.4 GTexel/s
Pro 5700 XT 95.94 GPixels/s, 239.8 GTexel/s
Last edited:
I was always confident in their CPUs and nothing I’ve seen performance wise has surprised me, but if this is true then wow. I was always much more skeptical in their GPUs.
Just few quick comments
Yes they can. Modern Nvidia, AMD and Intel GPUs fully support both single and half precision in the shaders, where half precision runs at double rate.
This figures are meaningless. GPU makers stopped focusing on thee figures long time ago, for a good reason.
Yes they can. Modern Nvidia, AMD and Intel GPUs fully support both single and half precision in the shaders, where half precision runs at double rate.
This figures are meaningless. GPU makers stopped focusing on thee figures long time ago, for a good reason.
They literally took the same scores from the A12Z DTK as linked in the original post. Didn't really have the time to watch that stuff, but if they claim that this is the M1, they are lying.
I think this post nails the MacBook Air.
Take two steps back, and look at it. It is one hell of a sweet little computer. It's not really designed to push the envelope, it's designed to fit into it! In all respects. Utterly silent as there is no fan to make noise, no moving parts at all except the keys and the hinge and a screen and keyboard that is really nice for text. Battery life that largely frees you from anxiety and having to carry around chargers, and allows conservative charging that greatly increases the longevity of the battery, as does simply needing fewer charge cycles.
This while still comfortably having performance margins for personal photo and video editing et cetera.
It's a great personal computer.
Switching away from x86 for personal computing is momentous, so of course tech geeks want to analyse this first foray into a brave new world. But that shouldn't be interpreted as detraction from what the MBA does, and does really well, its just curiosity and a desire to have a better foundation for extrapolation into the future. We don't want to see how M1 performs at for instance CineBench 23 because we want it to replace render farms, but to complement our understanding of the architecture.
There is no conflict.
That is truly Pro work! And also mine. We definitely need a Mac Pro. At least!
Here is the previous fanless MacBook, my 2017 Core m3-7Y32:
The performance improvement on the coming fanless MacBook Air is going to be unreal.
The performance improvement on the coming fanless MacBook Air is going to be unreal.
nope, because some windows users says....the mac with M1 dont support up to 128 Gb rAm
I'm not an expert but was thinking of this article. It's almost two years old so maybe things have changed.
"When the PC moved to unified shaders, the industry moved to FP32 for all GPU functions. This is as oppposed to the mobile world, where power is an absolute factor for everything, Vertex shaders are typically 32bpc while Pixel and Compute shaders can often be 16bpc. We’ve seen some movement on the PC side to use half-precision GPUs for compute, but for gaming, that’s not currently the case."
Why is it meaningless for a GPU to be able to deliver higher pixel rate per second?
The article is not wrong. Mobile GPUs have a long history of using reduced precision to improve their power efficiency. But in the recent years, desktop GPUs started implementing double-rate half-precision ALUs as well. Intel has had them for a while I think, Nvidia since Pascal and AMD since Vega. The main motivation for this was actually not graphics but increased interest in machine learning, where reduced-precision calculations are frequently used. But there is nothing preventing modern desktop GPUs from using faster half-precision calculations where it makes sense (e.g. for color shading).
Because it's not a figure that says much. Not so long time ago this and other metrics like "triangles per seconds" were common marketing points and everyone was competing in pushing higher and higher numbers. But we now live in the age of GPU-driven pipelines, unified shaders, sparse textures and complex rendering algorithms... so other figures like shader throughput, memory bandwidth etc. are more important. Basically, any modern GPU is so good at doing basic tasks like primitive assembly or rasterization that boasting with these numbers does't make much sense.
Besides, how do you interpret that? Like what does 41 Gpixel/s actually means? What is a pixel? Which pixel format are we talking about? Does it mean that you can fill up a 4K frame buffer 5000 times per second? Sounds impressive, but why would you want to do that? You actually want to draw something meaningful to it, don't you? So what is left of those Gpixels in the end? One can ask similar questions about the texturing performance. Is that the filtering performance or texture bandwidth or both? Which filtering settings? Which pixel formats? What about compressed textures? And so on. Frankly, I'd rather see some benchmarks
best 4core of all? https://www.cpu-monkey.com/de/cpu_benchmark-cinebench_r23_multi_core-16
single core even better?