Jump to content

Mac Silicon OS and VW


Recommended Posts

2 hours ago, line-weight said:

 

Sure. But they don't necessarily "do the math" for what's best for VW-focussed use. For example VW seems quite memory hungry but often wastes processing power. One of the issues with the first M1 mac minis was the maximum of 16gb RAM.

 

VW 2023 is significantly more optimized for ARM and RAM than 2021 and 2022, so that helps. Other apps? It really depends, and they're sharing the same resources as VW.

 

The new M2 Mini offers double the RAM upgrade, so in that regard, the maxed out M2 Mini is significantly better than the maxed out M1 Mini. So that's kind of a double win for memory in 2023. BTW we're on SP3 now, so at least VW is stable (can't speak for Ventura).

 

VW 2023 will also take full advantage of the GPU improvements in Shaded mode.

 

404256694_ScreenShot2023-01-26at12_02_14.png.3c1330898f5f195341c371931eb0d3ba.png

MacBook Pro's are currently filling in the midrange configurations.

 

BTW here are the apps on my machine that are still running on Rosetta 2:

 

2007792091_ScreenShot2023-01-26at11_57_51.thumb.png.0a9ce32a94991d25d615563e54bf445e.png

 

Edited by Mark Aceto
Link to comment
32 minutes ago, Gilbert Osmond said:

Not sure whether someone's later reply concerning a Mac Mini M2 possibly throttling (due to heat) on a long render vs. a Studio is on point. Has anyone tested this head-to-head? Seems to me the M2 architecture in general is vastly more efficient than the prior bloated Intel junk, so that heat buildup on long renders may not be an issue? 

Apple would not have put the active cooling fans they did into the Studio (and the MacBook Pro) unless they needed them.

 

I have seen test results (can't find them right now) of an M1 MacBook Air (no cooling fan) running the same test as a MacBook Pro (with cooling fans). They ran almost identically for about 10 minutes. Then the Air fell off a cliff as it hit the thermal limit and backed off the clock speed to keep it cool.  The MacBook Pro ran at full speed for the duration of the test (~60 minutes).

 

So yes, there are still thermal events that need to be considered.

 

  • Like 1
Link to comment

I just priced out the cheapest 14" MacBook Pro that I could configure with >32gb shared memory, and I was SHOCKED at how overpriced it is.

 

I completely understand the choice to pay $2500 for a Mac Mini vs $3500 for a MBP. That pricing is CRIMINAL.

 

I honestly thought there were some mid grade options between $2500 and $3500 but, if I'm not mistaken, they're all capped at 32gb shared memory, so they force you into paying more for that "trim" option. Elon must have given Tim some pointers on how to bleed customers when he visited the spaceship.

Link to comment
3 hours ago, Pat Stanford said:

Studio has more memory and far more cooling. So something like renders that run for and extended period of time (more than say 10 minutes) may cause a Mini to throttle due to the thermal envelope where the Studio can probably run flat out continuously.

 

A Mini with 32 (or even 16) GB is probably sufficient for 90% of the Vectorworks users. The people you see on this board (you @line-weight, @Christiaan, @Mark Aceto, @jeff prince, etc.) are high end power users and are going to want more power. But even for these users the Mini would probably be fine for 90% of their use. If you offload the renders to the VW Cloud instead of running them locally, then the Mini would probably be acceptable for 95%+ of even the power users needs.

 

$0.02.

Have to say, I don't think of myself as a "power user" (always find that term a bit cringe).

 

I get by fine with an M1 mac mini with 16GB RAM for now.

 

Almost everything that slows me down or annoys me is stuff to do with VW rather than the mac mini. I'm not in a hurry to think about moving to an M2 anything at this point.

  • Like 2
Link to comment
27 minutes ago, line-weight said:

Almost everything that slows me down or annoys me is stuff to do with VW rather than the mac mini.

By my definition, that makes you a "power user" you are pushing the limits of what VW does. A large percentage of the things that annoy you are things that 90% of the users would never come across because they are not using the software to the extent you are.  QED. You are a Power User 😉

 

Link to comment

I'm not going to pretend to know how 90% of VW users use VW. Nor am I going to define or label who is or isn't a power user. However, as someone who's been using a Mac since at least 1987 (Mac SE) with the exception of 1999–2003, here's what I objectively know from experience, and professionally recommended to my friends and colleagues here in the forum:

  • Single core performance - From top to bottom, this is a level playing field, so it's not even a consideration. The cheapest M2 machine is faster than the most expensive M1 machine (and the $50,000 Intel Mac Pro because it's a stupid Xeon instead of a 13900) and always will be. This will not be the bottleneck.
  • Multicore performance - Beyond 12 cores will mostly be utilized by Renderworks (the Cinema engine will use every core you throw at it). However, most everything else in VW on either a design or sheet layer won't really benefit from more than 12 cores. This is another no-brainer for most users; don't waste your money here. This will not be the bottleneck.
  • GPU - GPU's are the new CPU's. However, in the context of the Mac Mini, it's not a factor. You'll get what you'll get, and you'll like it. This will not be the bottleneck.
  • Shared Memory (RAM + VRAM) - Let's start with the painful truth: even though a Mini is a desktop, it cannot be upgraded. Therefore, you want to get more than you think you'll need a few years from now. Give yourself some headroom for having an Adobe app open at the same time as VW. If you're backing into a budget, this is the upgrade choice in 2023 that will determine everything else. This could be the bottleneck, so the recommendation is 32gb.
  • Storage - Desktops, laptops, iPads, iPhones... this is where Apple really stick it to their customers, it also cannot be upgraded, and we're always filling it up. Cloud storage and external hard drives can certainly help with savings here.
  • Thermals - Fortunately, this is no longer the bottleneck in 2023 but it's worth mentioning after everything we suffered through during Sir Jony Ive's thin and light reign. There's a very good reason the new MacBook Pro's are thiccer than a sticker.
  • As computers get bigger and badder hardware, developers' apps become more resource hungry. Everything is relative.
  • Developers are still optimizing for ARM. When I hop on my 2019 and even 2014 MBP, they just feel like rock solid machines. They were the most refined iterations of their generations (RIP Nvidia GPU's and drivers).
  • We don't draw in a vacuum. We have other apps open. Some of them, like the free Twinmotion with an Easy-Bake Oven interface that a toddler could master, require a minimum of 38gb combined memory (76gb is recommended)
  • Some of those that cut corners are the same that complain loudest... about the way things "should be"
  • "All of this has happened before, and all of this will happen again." See you all when the M3 is announced... 

 

Edited by Mark Aceto
  • Like 2
Link to comment
3 hours ago, Don Seidel said:

context fellow VW users, context.

 

If you're a power user, you're debating the wrong topic . The VALUE (cost to performance) of the M2Pro Mini 32GB ram for $2000 is an excellent buy and well suited for the average user.

 

 


Is it a value when compared to the last intel iMac made with 96 gb aftermarket Ram and the AMD video card it came with?  That was about $2200 this past summer IIRC.   I’m really wondering because site model updates don’t seem to happen any faster on the mini and TwinMotion does not appear to be an option there either.  I don’t know why, it just seems to be the case.

 

We just picked up an i9-10900x based machine with an RTXA4000.  I’m expecting the same results with site model updates, but interested to see how the rendering performance compares against my iMac.


I think some voodoo love is needed from Vectorworks to extract better performance with site models, regardless of the hardware used.

Link to comment
12 hours ago, Pat Stanford said:

By my definition, that makes you a "power user" you are pushing the limits of what VW does. A large percentage of the things that annoy you are things that 90% of the users would never come across because they are not using the software to the extent you are.  QED. You are a Power User 😉

 

Hm, I'm not sure I agree, unless it's the case that, for example, 90% of users don't really attempt to use VW in a 3d-based workflow. Although, it wouldn't entirely surprise me if that was the case.

  • Like 1
Link to comment
15 hours ago, Mark Aceto said:

 

VW 2023 is significantly more optimized for ARM and RAM than 2021 and 2022, so that helps.

 

 

 

Interested to know where you see this producing real-world use benefits. For example my comparison between 2021 and 2023 so far has shown:

- Updating section viewports in 2023 takes about 60-70% of the time it does in 2021. That's a significant difference, although still a long way from the "real-time update" that everyone would really like.

- Renderworks renders I see a marginal but measureable improvement

- Viewing heavy (ie a lot of complex viewports) sheet layers no improvement

- Memory-hogging (ie filling up RAM and then not releasing it after completing processes) no improvement.

 

Unfortunately for me it's the last two things that would make most difference to my day-to-day use.

 

What I can't test myself is whether things would be different if I had more RAM available.

Edited by line-weight
  • Like 1
Link to comment
2 hours ago, line-weight said:

Interested to know where you see this producing real-world use benefits. For example my comparison between 2021 and 2023 so far has shown:

- Updating section viewports in 2023 takes about 60-70% of the time it does in 2021. That's a significant difference, although still a long way from the "real-time update" that everyone would really like.


Not leaking / holding onto as much memory.

 

2 hours ago, line-weight said:

Renderworks renders I see a marginal but measureable improvement


Not aware of any RAM improvements here in VW (RW is typically a CPU bottleneck and barely touches the GPU / VRAM).

 

 

2 hours ago, line-weight said:

Viewing heavy (ie a lot of complex viewports) sheet layers no improvement

 

I wish more RAM or VRAM would solve this problem. Sadly, this has never been a hardware bottleneck. They’re interested in finding a solution though.


 

2 hours ago, line-weight said:

Memory-hogging (ie filling up RAM and then not releasing it after completing processes) no improvement.

 

@Christiaan you wanna take this one?

 

 

2 hours ago, line-weight said:

What I can't test myself is whether things would be different if I had more RAM available.

 

 

I can summarize my experience. Once I hit a certain threshold, memory was no longer the bottleneck. Today on M2, that would be about 32gb (shared). At the moment, on M1 Ultra 128gb, I have more memory than I’ve ever needed. It’s never been pegged (aside from UE/TM). I’m regularly hovering in the 25-50% range with a few exceptions, so that would be 32-64gb (shared). That’s a roundabout way of saying the sweet spot for shared memory on ARM is 32-64gb (for me). Less than 32gb (shared), and I’d be hating life. More than 64gb (shared), there’s not much of a return. Over 96gb (shared), I don’t even notice it.

 

Btw I think Apple improved thermal cooling in the M2 Mini, right? That would be another benefit compared to the M1.

Link to comment

I reported on some of my memory-hogging experiments here.

 

My impression is that having more RAM might not really help (it simply would take a bit longer for it to fill up and everything collapse). I think that was based on some comments from other users, for example this one here by @zeno describing memory on a (I think) 64GB machine getting eaten up, and going off up to 140 or 190GB.

  • Like 1
Link to comment

What really grates my gears is that Apple doesn’t offer a 48gb or 64gb shared memory upgrade option for the M2 Mini because they want to force users into a Studio or MBP.

 

For context, the 5-year old Intel Mini offered up to 64gb RAM + 1.5gb VRAM (65.5gb total) and was eGPU compatible. But, you know, then they’d sell a lot less Studios and MBP’s… 

 

I’ll use a Mac as my daily driver until the bitter end but post-Jobs Apple is pure greed and subscription “services.”

 

839096640_ScreenShot2023-01-27at08_14_48.png.bf181b800447306ea2822c026a3cefd9.png

 

Edited by Mark Aceto
  • Like 3
Link to comment
1 hour ago, line-weight said:

I reported on some of my memory-hogging experiments here.

 

My impression is that having more RAM might not really help (it simply would take a bit longer for it to fill up and everything collapse). I think that was based on some comments from other users, for example this one here by @zeno describing memory on a (I think) 64GB machine getting eaten up, and going off up to 140 or 190GB.

 

A lot of contextual relativity to unpack here... 

 

First of all, I've never run out of memory with 128gb, so yes, more memory helps. If @Christiaanor @zenowould like to share one of their mega drawing sets with hundreds of section viewports, I'd be happy to test if I could peg this machine. It's defeatist to say more memory won't help but there is a diminishing ROI at some threshold for every user.

 

BTW their feedback directly helped the memory management improvements in VW 2023.

 

That said, a lot of the initial feedback was regarding VW 2021 and 2022. Keep in mind, the M1 Mini was released over two years ago at the end of 2020 (shortly after VW 2021 was released). And VW 2021 ran in Rosetta 2, so VW 2022 was the first version that ran natively on ARM, so of course it wasn't as optimized as it is now (or will continue to be in the future). Whenever Apple change architecture, we all go all along for the ride (growing pains).

 

140 to 190gb on a 64gb machine? Those are some Activity Monitor funny numbers... 

 

If I could summarize everything:

In other words, the user experience of running VW 2023 on a maxed out M2 Mini should be noticably better than running VW 2021 or 2022 on a maxed out M2 Mini.

 

*** For the heavy RW users, these are the Cinebench scores:

  • 75,671  AMD Threadripper 3990X
  • 41,012  Intel i9 13900K
  • 21,740  Apple M1 Ultra
  • (M2 Pro is presumably 2/3 the M1 Ultra)

So, it seems Apple have painted themselves into another corner with ARM.

 

Edited by Mark Aceto
Link to comment
14 minutes ago, Mark Aceto said:

 

 

 

140 to 190gb on a 64gb machine? Those are some Activity Monitor funny numbers...

 

 

My (limited) understanding is that once the xGB of RAM is used up, stuff that would otherwise be put into RAM is put on the hard disk instead, of course slowing everything down.

 

On my 16GB machine Activity Monitor has registered up to 80GB before things seize up.

 

This is why I wonder how much difference having 32GB rather than 16GB would make - because if VW is happy to shovel in 80GB of stuff on a 16GB machine then surely it would do the same on a 32GB one.

 

(And ultimately - why does it do this anyway - surely it shouldn't have to. It does this when I batch-render a bunch of RW VPs, as if it can't do one, release memory, do the next one, etc. Maybe this behaviour has been fixed for section VPs but not RW VPs?)

 

Edited by line-weight
Link to comment
Just now, line-weight said:

This is why I wonder how much difference having 32GB rather than 16GB would make - because if VW is happy to shovel in 80GB of stuff on a 16GB machine then surely it would do the same on a 32GB one.

 

Doubling the low end from 16 to 32 would make a big difference.

 

Adding 16 to the high end would not be noticeable for most users, myself included.

 

I feel like a broken record but:

  • Look at the Intel MBP in my signature: 32 + 8 (40 total; not 32)
    • So 32 is really more like 24 + 8 if we're comparing ARM's to Intels
  • I've never run out of memory with 128gb, and I'm usually sitting somewhere between 25 and 50% (32 to 64gb)

If 32gb shared memory is in one's budget, it's a no-brainer for 2023 and beyond.

 

If a power user is concerned about running out of memory, and has the budget, I can vouch for the Studio not disappointing. I complain a lot. I've never complained about running out of memory.

  • Like 1
Link to comment
14 minutes ago, Mark Aceto said:

 

Doubling the low end from 16 to 32 would make a big difference.

 

Adding 16 to the high end would not be noticeable for most users, myself included.

 

I feel like a broken record but:

  • Look at the Intel MBP in my signature: 32 + 8 (40 total; not 32)
    • So 32 is really more like 24 + 8 if we're comparing ARM's to Intels
  • I've never run out of memory with 128gb, and I'm usually sitting somewhere between 25 and 50% (32 to 64gb)

If 32gb shared memory is in one's budget, it's a no-brainer for 2023 and beyond.

 

If a power user is concerned about running out of memory, and has the budget, I can vouch for the Studio not disappointing. I complain a lot. I've never complained about running out of memory.

 

I'm not disputing that more memory is better - and sure, given the option buying a new mac today I'd take 32 over 16.

 

However - I believe there is a problem with the way VW deals with memory, that has not been entirely fixed in 2023, which means that it just keeps wanting more and more, until everything stops working. Perhaps only in certain situations, which not all users will experience.

 

That's based on my own experience, and what I've seen other users noting. In addition to @zeno describing it happening on a 64GB machine, for example here is @Tom W. describing having to restart VW2023 due to running out of memory on an intel machine with 48GB (if his signature is correct).

 

Again, I'm not disputing that 32 will generally be better than 16 - but I'm not sure it would solve this particular problem, other than that it would perhaps not occur until a little further into a session.

  • Like 1
Link to comment
10 minutes ago, line-weight said:

 

I'm not disputing that more memory is better - and sure, given the option buying a new mac today I'd take 32 over 16.

 

However - I believe there is a problem with the way VW deals with memory, that has not been entirely fixed in 2023, which means that it just keeps wanting more and more, until everything stops working. Perhaps only in certain situations, which not all users will experience.

 

That's based on my own experience, and what I've seen other users noting. In addition to @zeno describing it happening on a 64GB machine, for example here is @Tom W. describing having to restart VW2023 due to running out of memory on an intel machine with 48GB (if his signature is correct).

 

Again, I'm not disputing that 32 will generally be better than 16 - but I'm not sure it would solve this particular problem, other than that it would perhaps not occur until a little further into a session.

 

I think I got threads mixed up because there's a Venn diagram of the same handful of people complaining about the same handful of things in a handful of slightly differently places. Just spend your life savings on a Mac, and everything will work great.

 

It's like we're all sitting around the monolith at The Dawn of Mac:

 

image.png.e34be46447c0f47c250d5601fa5ca995.png

 

(And we know what happened to HAL's memory.)

 

Moving on... 

Link to comment
20 minutes ago, line-weight said:

for example here is @Tom W. describing having to restart VW2023 due to running out of memory on an intel machine with 48GB (if his signature is correct).

 

Yes my signature is correct. Yesterday I was updating some VPs in a VW2022 file + was having big memory increases. So for example:

  • Open VW + open the file + RAM stands at 8.4 GB.
  • Update a Shaded section VP + the RAM increases to 30 GB.
  • Update a second Shaded section VP + RAM ends up at 60 GB + things get very unresponsive/slow + time to close the file asap...

Today I updated this file to VW2023 + performed the exact same steps. This time the RAM levels were 7.2 GB, 35 GB + 46.4 GB respectively, so contrary to what I said in the other thread, VW2023 actually performs better although the underlying problem is the same. And I agree with @line-weight that having more memory wouldn't 'solve' anything, only delay the point at which I'd need to shut things down + start again (which wouldn't be a bad thing: I was aiming for 64GB not 48GB but my machine wouldn't accept all of the memory modules when I inserted them so I ended up with 8/8/16/16=48...)

 

I was sure there was a point after one of the VW2022 SPs when this memory problem suddenly got much better but perhaps it's more to do with the type of file I'm working on at the time (e.g. very detailed).

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...