Hi,

I am still before ARM transition and still doing motion design stuff on relatively new Intel based Macs (six-core i7 and six-core Xeon to be exact).

Considering my future upgrade to M1 or M2, MacBookPro or Mac Studio -- and trying to set a reasonable path through available configurations -- I wonder about possible benefits. Especially, following question is bugging me a lot - whether adding an extra GPU cores will make my Studio Artist fly faster or no? Plus: what about additional CPU cores on M1 Ultra -- shall I consider them?

So, what's your experience with Studio Artist on M2 / M1 / M1 Max / M1 Ultra, my Fellow Artists? Does anyone of you working on M1 Max or M1 Ultra and wanna share an opinion 'bout it?

You need to be a member of Studio Artist to add comments!

Join Studio Artist

Email me when people reply –

Replies

  • The extra GPU cores are not going to make a big difference for most of the existing functionality in V5.5.  Most of the V5.5 processing involves cpu cores.  Extra cpu cores help for anything that thread splits.  Not everything in Studio Artist thread splits, so it depends on what you are doing.  The paint synthesizer for example runs in one cpu thread, since it has to live in the interface thread.   Depending on what you are doing inside of the paint synthesizer, some operations may split off and use additional cores for their piece of a larger puzzle being run to generate the specific paint effect they are part of.

    Something like interactive warp will use all of the cpu threads it can grab to speed things up.  Something like the vectorizer will initially split off to a new thread, and then specific operations inside of the vectorizer might take advantage of all of the cpu cores to speed up their piece of a longer more complex processing chain that drops back to a singe thread.

    Vector rendering (for things like vector painting) does take advantage of the gpu.  But that runs so fast in most situations having a ton of extra gpus isn't really going to make a difference to it.

     

    The big win with the intel to arm transition on the mac is the lower power consumption of the arm architecture.

    Also, how apple structured the cache memory on the M1 and M2 chips was well thought out, and acts to speed things up.

     

    To be honest, i'm somewhat disappointed about apple gpu performance so far.  On paper it seems like this great thing, here's an ultra and you can get all of this additional gpu power with it.  But if you are going to spend that much money, would you be better off getting a high end nvidia card on a windows machine instead?

    A lot of my current thinking on this comes from looking at the differences running pytorch code on nvidia on windows vs trying to run it on a mac.  Assuming you even can run it on a mac right now, since the pytorch implantation for running on the mac gpu was only available starting a few months ago, it has a ton of issues and bugs, it's very unclear if apple really cares about it or if it is all external folks trying to make it work, and the speeds are slower than what you would get using nvidia cuda if you can get it to run.  So that leads to a certain disappointment on my part compared to my initial reaction when the ultras were announced.

    Maybe in a year that will improve?  Or not?

    I could say the same thing about Vulkan support on apple gpus, does anyone at apple really care, or is it all external people trying to make that cross platform ship fly on the mac side.

    So my personal view of using the gpu on the mac from my particular developer perspective is that it's kind of a mess right now.  Which has really colored my viewpoint of the Ultra machines.  That and the ridiculous overpriced external monitor apple sells for them.

     

    As we progress into the future, we would obviously like to take advantage of all of the parallel processing power that the gpu side brings.  We currently exploring various options.  How that is going to play out is currently a work in progress.

    If you are using 3d or other programs that were rewritten or have internal libraries that use Metal internally, then your take on the extra gpu cores might be very different than my personal take above.

    • Thanks a lot. These are the exact context I was looking for. Beside Studio Artist (which is, frankly, quite new addition to my repertoire), I also sit well on Adobe and FCP X (animation and motion design centered). Appreciate your insight, a lot!

      Somehow the current state of things (as described by you), reminds me story of "Mac Pro 2013 and its unemployed double GPUs" -- played in different key, but remained melody. It's both software and hardware that matters, singing as a duett.

      I got my fingers crossed for the better.

This reply was deleted.