Mac mini mini - will it run Flame?

[paul_round]

I love the fact its on a website called Vintage Television!

[Slabrie]

Nice thread

We did not receive our Mac Mini yet but for macOS computers, we only provide minimum requirements so if you have a config that matches that then, as Jon has seen, you should be good to go. The new mac mini is really like a powerful laptop without a screen, thanks to how now we can compress technologies these days!

[finnjaeger]

FWIW

after 1 week of using the M4 mini on a project the user complained that it wasnt on par with the M2 Ultra we had him use before.

I cant verify anything this is just his experience but i wanted to share this.

is giving some render issues - think ok for smaller jobs but perhaps bigger comps with finer detail might not be ideal - just a heads up. we got Broken frames - not rendering the comp as it should be - flashing frames etc..

adjusting for finer details - subtle changes is tricky - small changes on say cc slider from my side - it jumps - makes it harder to comp for proper detail

there is obviosly a lot of “it feels slower” and stuff involved and he is using it remotely so macOS version + parsec might be another thing to look at - so take this all as anecdotal at best. will continue giving it to other people to see whats what.

[AdamArcher]

Thanks for sharing this experience @finnjaeger

I’m wondering if this is RAM related.

[pixelmixer]

@paul_round

I remember, back in the day, when the Bosch FGS400 arrived at MPC, it was so big and heavy, (think 4 full size Onyx1’s stuck together), they had to put all sorts of monitoring in to make sure the floor could take the weight.

Made a living on the FGS4000 for a few years. Brings back memories…

[ytf]

I have vague memories of a teapot demo in '81 or '82. It was many years later I learned of the teapot trope.

[finnjaeger]

I will switch artists around next week and then see about nuke maybe too.

just needs some actual testing, 64G of ram should have been plenty for what was beign done to these comps

[finnjaeger]

Update, switched artists back to M2 Ultra

Artist is super happy, way more snappy, no render issues “night and day difference”

other artist now using m4 Mini is also now complaining about crashing issues on heavier comps.

[Jonhollis]

@finnjaeger

…sting for finer details - subtle changes is tricky - small changes on say cc slider from my side - it jumps - makes it harder to comp for proper detailthere is obviosly a lot of “it feels slower” and stuff involved and he is using it remotely so macOS version + parsec might be another thing to look at - so take this all as anecdotal at best. will continue giving it to other people to see whats what.

thats quote odd. I have had the opposite experience. faster for almost everything no crashes.

[finnjaeger]

i dont know why.

Used it today for timeline work, and its sloooooow , i tried to do a few motion timewarps on the timeline and keep running out of memory (64G) … on HD timelines…

so yea no this thing is a full-no from me

[cnoellert]

The memory thing is annoying as all hell… but in fairness I can easily trigger it in Linux too. But you combine that with the slowness of the model rendering in time warps and the Mac basically becomes unusable unless you have a burn node to render time warps.

Lame.

[hBomb42]

@cnoellert

But yo combine that with the slowness of the model rendering in time warps and the Mac basically becomes unusable unless you have a burn

How is the new ML stuff on a Studio? Usable or nah? I tested the older ML semantic keyers on one maybe a year ago and they were plenty snappy.

[finnjaeger]

I never seen this before ever personally, it sa 1/4 res motionvector timeline timewarp , with motionblur…

anyhow yea lame.

[finnjaeger]

if you have time to wait a long time its ok. …

super slow, infuriatingly slow.

[cnoellert]

Ah, I meant the ML timewarps. I haven’t had issues with the motion-estimation.

[cnoellert]

@hBomb42

How is the new ML stuff on a Studio? Usable or nah? I tested the older ML semantic keyers on one maybe a year ago and they were plenty snappy.

Like @finnjaeger said. It works but dear god it’s slow…

[finnjaeger]

me neither, but just had this on the m4 mini, no problem on the m2 ultra .

everything is slower on the m4 mini, hence my confusion why john says they are equal

[hBomb42]

@hBomb42

How is the new ML stuff on a Studio? Usable or nah? I tested the older ML semantic keyers on one maybe a year ago and they were plenty snappy.

When I tested Talosh’s ML tw a year ago it was like a minute per UHD frame. Is the new adsk ml tw also that slow on the Mac Studio?

[finnjaeger]

Its not as slow , but its not fast, you can use it, its there it does its thing, but you really want to sort of hard commit that or use it in batch or something.

In the timeline its infuriating.

[Jonhollis]

yes still finding it to be ok. I’m only doing around 4K stuff. and mainly batch not timeline. shots for movies and shortish commercials. I dont touch the ML stuff on flame as it’s so bad. the ML timewarp\ is so slow. but is on the m4 and the M2 Ultra studio. so maybe not the same use case as yourself. just ordering up a M3 ultra Mac Studio with 512GB ram so will move on to that. one thing to bear in mind I dont only use flame. and the advances and speed of the other bits of software are really good. I think Autodesk have a lot of optimising yet to do. also it hasn’t crashed on me once. and hasn’t slowed to the point of not being able to use. all seems fine to me. I think the thunderbolt 5 drives I’m using help a lot.