I also upgraded my RTX 2070 for an AMD RX 6800.
Just for the sake of having a better non proprietary “open source” card for Linux.
OK, finally had much more problems than benefits for linux though.
(lost 2 of 3 Linux installations that way)
And not seen any performance advantages anywhere.
But hey, I have 16 GB VRAM instead of 8 GB.
Should fit any of my geometries for a few years now.
No Nvidia cards would have no disadvantage; quite the opposite. But as content creators do we need 400 FPS when building our environments?
AS native build of UN engine is currently in the works. You can actually build a version from GitHub if so inclined to test; things are still janky though. Hopefully we see something come 5.2
If you haven’t started learning Unreal, stop what you’re doing at this exact moment and start searching tutorials, seriously I’m not playing 
I already feel a little late to the show. But better late then never, right? This mainly applies for massive scenes and realistic film and animation production.
Even if you loved how Eevee looks in Blender… Unreal 5.1 is going to blow your mind.
Sorry for another share from the same content creator but he’s explaining the advantages for 5.1 really well. At 4:00 when he starts showing the sun light dynamically shining through the car windows while adjusting the intensity of the reflections is so choice.
From time to time I tried to get used to UE.
But for me it is still a monster of a Software, not really intended for our purposes.
I do not understand what it does, where it installs these Terabytes of data for
what purpose, for each slight update version, if you don’t actively stop it from
doing so …
I think I have some control of Twinmotion.
not my preferred UI for a (pro) Visualization Software but it works at the end.
I see ll the bells and whistles in UE first quite some time but not really accessible
for me. At least 80% of UE’s GUI presentation is irrelevant for me and just in the
way occluding features I may need.
But even If I would be a real game designer, I would likely not love that App’s
way of thinking. UE is an App that seems too think it is the only App worth or needed
to boat your whole system.
I’m right there with you. I’m still in the process of learning the ins and outs of the software while searching things like “how to link assets to another project without migration” kind of google prompts.
It is a different backend workflow (file location, file linking etc.) kind of beast than Blender, but it’s virtually the same when it comes to making the art; geometry is geometry, materials are materials, lighting is still lighting with light sources, and world coordinates are world coordinates.
All in all, it’s still not Houdini rake your face off type of different though 
It was probably their best move ever. They are now in control of their own CPU after having to rely on Intel for too many years. And the performance is great, especially under battery with no fan noise. I’m a developer and compile times are way better than my old Intel Mac Pro.
Now, 3D rendering isn’t nearly as good as Nvidia, but I would doubt that it was a priority.
I can imagine that it will become one and they will add RT cores in a future iteration. Considering they are spending resources on making Metal versions of Blender, I wouldn’t be surprised if they cater more to this niche market.
Just note, Nanite and Lumen are not Unreal only technologies. Octane and Redshift are using it too in new versions. And Eevee next is going to support some of it too. So Unreal is just the less user friendly version of it,
I am happy that Blender does not computer shaders for half an hour
after each little point update. And then again computes the same
2798 shaders each time I open a file
But maybe that is just a UE on M1 Ventura thing …
People, stay calm. Apple fully supports Blender. This is a wonder of the world! Apple wants the cup and it will take a while until the strategy works out
Of course, Nvidia is extremely fast here, but not for long.
Ue 5 did that on Monterey too I think.
Have not tried it since as that bugged me.
Well, UE is primarily a Game Engine.
It is just a shame that a small part of UE includes a complete feature set
of top notch (architectural) real time visualization tools needed.
But hidden under a monstrous non logically hierarchical UI,
using non-(our) industry terms and naming, inside a gigantic non-App/OS-
standards behaving bloatware.
Thank god Epic bought Twinmotion.
So 2-3 years later these features will find their way into Twinmotion, which
makes them easy accessible for mere humans.
Although I would rather prefer the UE original if it was available in a slimmed
installer with a GUI where you can hide all what you don’t need.
(but I asked for such a hide option for Modo too and failed )
I noticed in 3.5 alpha that Cycles renders about 15% faster on average than 3.3.1 LTS using my m1 max 32 core macbook pro. That’s nothing to sneeze at!
Technically they haven’t even really implemented the latest optimizations that Apple submitted since they haven’t gotten around to reviewing them yet!
Here is something:
Slightly even more overclocked M2 versions on cost of efficiency (?)
I still don’t get that barely noticeable M2 IPC gain over M1,
although a newer architecture AND slightly optimized process.
I mostly only see more cores, higher clocking by more energy
consumption for M2.
OK, more memory offers.
And exactly the same with M2 Pro and Max,
something I expected to never happen because being too embarrassing.
That is no disadvantage for anyone who planned to buy their first Apple ARM
MBPs in a few months.
But beside that there is no plausible reason for anyone to upgrade from its
M1 MBP - who would prefer a M2 MBP over price dropping M1 MBP offers ?
My 2012 starting Apple depression curve got finally a quite noticeable negative
upwards bump with M1 presentation in 2020. But with end of 2022 I am already
again on bottom Apple crisis indication level.
Maybe the Apple-leaving middle upper management felt similar.
But hey, they could still employ Elon Musk to accelerate everything …
in one or other direction.
Haha, less managers, more coders!
It’s my understanding that M2 never was meant to be a big jump in performance. I think it may be just a stop gap while they work on M3. No evidence though.
Maybe.
I think M2 was always planned as a big step as usual.
There were long time rumors about TSMC’s 3 nm process coming this ! year.
Apple having preordered enough capacity. Intel gave back their preorders
because they could not design fast enough for 3 nm.
So most expected that Apple is already heavy into 3 nm preproduction.
But with some difficulties by Corona, War, Chip crisis, … they may have
pulled the plug as it may have been regarded to risky to manage
3 nm process in time.
And delayed a rumored, already completed Mac Pro.
But decided to call it M2 nevertheless.
These bad news came in mid 2022.
Let’s hope Apple gets sorted again soon.
My concern is that the ARM Architecture Apple develops for their A chips first,
has also dropped in progress. AFAIK M1 was based on A14, while iOS had
already A15, which should be in M2 now and was not impressing.
And this year the A16, again not impressing, so much that they even put
it only in iPhone Pros (for other, marketing reasons).
But thanks to stagnancy, A15 only no disadvantage for normal iPhone 14.
So the holy M3 will also at max get the - again - mediocre A16 Architecture.
AFAIK still ARM 8. Even their own SoC architecture, as far as I understand
ARM 9 adoption would help nevertheless.
So beside any process miniaturization effects, I do not expect much single
core speed improvements. For me it looks Apple falls back behind x86 soon.
So it may be only about the GPUs that can increase in numbers and complexity.
Yes it is slowly but surely getting there.
Blender bench from 3.1 to now 3.4 is 20 % faster (same for the 32 core which is not mine).
My 14 core is still slower then even a GTX 1080 however.
The 32 core is close to a AMD RX 6650M now (if anyone is curious).
So if the M2 Max is ideally 50% faster I will be happy.
Getting it close to the AMD RX 6700 XT
And I would call that decent for a notebook.
yeah i just re-did bmw (i know terrible benchmark but i’m still curious) and got 35s on M1 Max 16", which is the fastest i’ve yet seen on this system.
an interesting note is this was the 3.5 beta, and i was initially getting over 50s for the same render until i disabled light trees, since i knew that was a recent addition. the results on bmw were identical with it on/off, so that’ll be something to consider on newer versions of blender. if your scene doesn’t benefit from many-light sampling, it can incur a substantial performance hit, and is by default enabled.