Obviously those are not seconds because it would mean M4 Max is 20% slower than M3 Max.
Last edited:
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.
3D Rendering on Apple Silicon, CPU&GPU
- Thread starter vel0city
- Start date
- Sort by reaction score
Obviously those are not seconds because it would mean M4 Max is 20% slower than M3 Max.
The Blender benchmark Score is a measure of how quickly Cycles can render path tracing samples on one CPU or GPU device. The higher the number, the better. In particular it's the estimated number of samples per minute, summed for all benchmark scenes.
opendata.blender.org
Blender - Open Data
Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
The Blender benchmark Score is a measure of how quickly Cycles can render path tracing samples on one CPU or GPU device. The higher the number, the better. In particular it's the estimated number of samples per minute, summed for all benchmark scenes.
Blender - Open Data
Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
That was interesting because people always talk about rendering time in seconds and minutes but here higher numbers mean faster rendering.
Uh huh, counterintuitive.
The render times for Octane and Redshift scenes will feel more helpful when we get them.
View topic - NEW: Octane X - is finally here! [MEGA FAQ THREAD]
View topic - NEW: Octane X - is finally here! [MEGA FAQ THREAD]
render.otoy.com
But like you said, the Blender Benchmark sums for all the scenes, and the score database on their site doesn't list the individual scene scores. The fact that Apple isn't sharing the full result so that one can easily compare to the PC scores in the result browser speaks volumes, I would assume.The Blender benchmark Score is a measure of how quickly Cycles can render path tracing samples on one CPU or GPU device. The higher the number, the better. In particular it's the estimated number of samples per minute, summed for all benchmark scenes.
Blender - Open Data
Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
EDIT: I just ran the Blender Benchmark to see the individual scene scores on my 4090...just as I expected.
Monster:
M4 Max: 237
Nvidia 4090: 5,393
Junk Shop:
M4 Max: 152
Nvidia 4090: 2376
Classroom:
M4 Max: 102
Nvidia 4090: 2621
That is actually shockingly bad performance on the M4 Max in comparison to the Nvidia hardware, TBH. The M4 Max, according to that benchmark, is only eeking out ~4% of the performance of a 4090...
Last edited:
Do they have redshift benchmarks also? (I don’t have an active account at the moment so can’t search), thanks for posting this is the first time I’ve seen these – no youtuber really tests this stuff and I’m so sick of seeing useless geekbench numbers that mean nothing to me for specific work.Uh huh, counterintuitive.
The render times for Octane and Redshift scenes will feel more helpful when we get them.
View topic - NEW: Octane X - is finally here! [MEGA FAQ THREAD]
View topic - NEW: Octane X - is finally here! [MEGA FAQ THREAD]render.otoy.com
View attachment 2448324
View attachment 2448321
edit: thanks, I totally missed the fine print in that shot
Last edited:
Yeah the Redshift benchmark scene above was rendered on the M2 Ultra (76 core GPU) in 3 mins and 25 secs.
Just check opendata for blender scores and there you have all you need the know about the best possible case. We are talking about 20% better perf. Nice, but changes nothing. This is at the cost of higher temps and noise so instead of silent running like the m1max, we are moving into pc territory with jet engines blasting…. And still not beating a 4070 even though that is on way older nodes. Tbh, if a m4ultra comes out, even with linear scaling, it would look quite tame compared to what will ba available in desktops at the same time. 5090 will be the single consumer card to compare to at that time. And that is rumored to beat out the 4090 significantly. So the mac/pc situation is still the same. Apple makes awesome laptops but doesn’t have anything powerful on desktop. And like i said before, that is fine for me at least. Now, If there was a mac desktop with about 2x the perf of a m4max i would take a look at it, but bying one would be more of a fanboy decision than a business one.
The scores in that original screenshot from the youtube video have to be totally wrong or something. There's no way it's getting 4% performance in individual scenes but somehow getting a 5000 point overall score (or the database median score for that single M4 Max result is a scam)
I wonder if they forgot a digit at the end of all the scores in the youtube video-- maybe 237 is actually supposed to be 2,370...
That would add up
EDIT: Just saw the graphs you posted from the other guy--yeah, i think Tech Chap failed extraordinarily at making their graphics accurate
Yeah, it makes sense. I asked him anyway. 2519 is about half your performance, just like Opendata.
Your results do not align with the official benchmarks or third-party benchmarks. Are you certain you are using the same settings?
In the official Blender benchmark database the M4 Max performs similar to the desktop RTX 4070 or mobile RTX 4080. The desktop 4090 is obviously significantly faster, but it is not 20x faster as your results suggest.
| Device Name | Median Score | Number of Benchmarks |
|---|---|---|
| NVIDIA GeForce RTX 4090 | 10885.57 | 635 |
| NVIDIA GeForce RTX 3090 | 5333.05 | 284 |
| NVIDIA GeForce RTX 4070 | 5126.74 | 281 |
| Apple M4 Max (GPU - 40 cores) | 5083.33 | 1 |
| Apple M3 Max (GPU - 40 cores) | 4257.46 | 86 |
| NVIDIA GeForce RTX 4070 Laptop GPU | 3457.44 | 260 |
| Apple M4 Pro (GPU - 20 cores) | 2530.47 | 6 |
| Apple M3 Pro (GPU - 18 cores) | 1768.64 | 43 |
Blender - Open Data
Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
The red numbers are from The Tech Chap that I posted, not the poster’s. Tech Chap lacks a zero in his numbers.
Slightly OT, but might be of interest to a few people on this thread, as they've used it in the past; looks like Modo is being discontinued: Foundry announces strategic decision to wind down development of Modo
I think the last version I used was 501 or 601; was really great for modeling and rendering, although lost it's way after the foundry purchased it, and Brad Peebler left (still wonder what he's up to at Apple).
I think the last version I used was 501 or 601; was really great for modeling and rendering, although lost it's way after the foundry purchased it, and Brad Peebler left (still wonder what he's up to at Apple).
Those are not seconds. Blender Benchmark App shows the sample rate; therefore, the higher number indicated better performance.
My M3 Max performs roughly on par with RTX 4070 laptop GPU (OptiX enabled). It renders Scanlands scene in 01:08 whereas RTX 4070 Laptop takes 01:05 with OptiX and 02:52 if you only utilize CUDA instead of enabling OptiX.
Similarly, RTX 4090 Laptop completes the scene in 00:42 with OptiX and 01:47 with CUDA.
What's even more interesting is the render times on battery:
M3 Max : 01:08
RTX 4070: 01:15
RTX 4090: 01:02
I read about upcoming Apple Neural Engine support (would have a similar effect to Nvidia Tensor cores); however, last I read, there is no confirmed timeline for the implementation.
What would that be used for? Denoising? Are tensor cores currently used for denoising?
NVIDIA OptiX™ AI-Accelerated Denoiser
developer.nvidia.com