Mac: M3 - *Hardware accelerated RT (Part 1)

Even if 0451 is just an AI,
there may be real people behind programming it.

1 Like

Yeah, but theyā€™re making the AI, not the SSGI, so they donā€™t count or something. IDK. TTYL.

Iā€™m fairly new to Blender and in need of a new computer (waiting until the M1X). What is your rendering workflow with Blender and M1? Do people try to stay in Blender or do you use third party rendering tools? And what are your struggles with Blender and M1?

For single frame images I use my M1 Mini, it may take a little longer than a 3090 but snack and potty breaks are a needed thing. :grin:

For animations I use Amazon servers. I wouldnā€™t try to render animations on a single 3090 even if I had one.

Others here have set up home render farms connected to their M1, Iā€™m really wanting to do something similar when I can afford it.

As for non-rendering task in Blender, I think youā€™ll really love the M1X when it comes out. My M1 Mini has been a delight to work with in Blender, when using the native ARM builds.

3 Likes

Not tomorrow apparently!

1 Like

Yeah, tomorrow is the new iPhone launch. I think fall is the timeframe for the M1X 16ā€ MBP and Mini announcement.

1 Like

Some promising infos from Jules Urbach (yes, I know this guy loves promises):

ā€žbrigade will be coming to Metal / Octane X 2022 . First time I tried Photon Tracing kernel was using Octane X 2022 on my M1 Macbook Air! Itā€™s also the first time I tested a major new feature on metal before the cuda version. I love the Macā€

ā€žI am pretty bullish on the long term potential of Apple Silicon. The OB per Watt (even without RTX) is the highest Iā€™ve testedā€

More at:

4 Likes

So probably 2024? They seem to be running about 2 years behind what they promise.

A15 55% faster graphics than A14:
https://9to5mac.com/2021/09/15/geekbench-results-show-iphone-13-pro-offers-55-better-graphics-performance-than-iphone-12-pro/
This equates to over 40% more graphics core performance (4 cores in iP12 vs 5 cores in iP13).

I didnā€™t think it was possible, but I look forward to the new Macs with even more anticipation. I collect money in the piggy bank, but I promised myself Iā€™d wait for the first in-depth reviews.

2 Likes

Thanks for that bit of info/update.
It has actually opened a rabbit hole of things to try for me, Radeon ProRender, USD Hydra and now quite interested in Octane nowā€¦ they seem to have made it quite easy for Mac users: free, standalone through the Mac AppStore, for Mac Pro buyers they get a full year the real deal, etc.

I actually use Unity a lot (way more than Blender), they have a plug-in for it, would be interesting to see what sort of previews could make with it.

I currently donā€™t have a M1, but it sure is nice to see some shops like RedShift, OTOY, AMD giving it a complete go at it. Heck, on one site they even have some iPhone 12ā€™s A14 benchmarks for Octane Bench.

1 Like

Interesting, I hadnā€™t considered a GPU power increase when talking about the new M1x (or whatever they call it).

So a 16 or 32 core GPU Mac Mini Pro could increase the GPU capability of those machines compared to the current 8-core GPUā€™s more than just by a core-number factor? This implies that effectively a 16-core GPU Mac Mini Pro with new M1x processing would be more than twice as fast as an 8-core M1 GPU Mac Mini.

Correct.
I hope for a MacBook with 32 graphic cores. There would be a chance to accelerate graphics by 5-6 times. I love my MacBook Air but letā€™s face it: it is entry level cute machine. Much more power is highly desirable.

1 Like

If I get that right, M1 is A14 tech while iPhone 13 Max has a newer A15.
Normally M1X would still be A14 tech, but more cores.
A15 tech should be called M2 then (?)

If Apple jumps instead of usual A14 for M1X directly to A15 tech,
which could be with chip shortages and delays,
I think it will be called M2 + M2X.
But more likely the MBPs will get M1X, so no GPU jump beside
core multiplication and M2 for next Mini and M2X for large iMac
early next year (?)

1 Like

yes the M2 will be based on the A15 and from the first benchmarks I am guessing this.

As fas as single core performance goes.

A14 + 10% = M1 = A15

so M2 = A15 + 10%?

Ah yes, right.

Didnā€™t know ā€œBA Crewā€ stood for ā€œBash America Crewā€ :sweat_smile:

1 Like

Maybe that is why they keep depth map video work only to the iP13. That is a very interesting feature.

I use Focos a lot to adjust the DoF afterwards and shoot my photos with Halide.

I never did this before but I might trade in my iP12 for the iP13.

I nearly wanted too but I think I will wait for 14.

The new macro feature looks awesome too.

Iā€™ve got a request for you M1 owning folk. As you all know, Iā€™m considering getting a Mac, but Iā€™m always weighing my options, and considering all the angles. My biggest concern is rendering, and the speed thereof.

So Iā€™ve been playing around with the Blender 3.0 beta, and have been absolutely floored by the massive speed increases in Cycles X. For this one scene Iā€™m about to show off, it used to take a good 15-20 seconds for me to get a clean view of my render in window on my admittedly old, not aging too gracefully PC. With Cycles X, it takes about 3 seconds. The video is actually a bit slower than its usual performance due to the extra overhead from recording, but itā€™s still fast.

So I figured, for the sake of comparison, I could upload my scene for people to grab, and test with.

Once I clean up the file, and find a good place to host it, Iā€™ll post a link. In the meanwhile, hereā€™s my video.

4 Likes

Iā€™d also be very interested to see the M1 results. Rendering is also an important factor to me. I donā€™t expect NVIDIA RTX 3080 speed on an M1 CPU, but a fair bit of speed would be nice.

1 Like

I might have to do some searching for a good site that can host a 130MB file. I tried Blend-Exchange, but itā€™s been processing the upload for the last 10 minutes.

Anyone have any recommendations?