MP 7,1 Mac Pro 8.1 - WHAT IF...

[AdamBuker]

Hmmmm, interesting. Sounds good...how much faster would a config like this be than the M2 Ultra Mac Studio (incoming)???

I think this would come down to how well Apple could overcome the challenges in such a design and how well software is optimized for it. Yet in my scenario even the base model would be faster as it would have 256 GB RAM, 48 core CPU, 120 core GPU, 64 NE cores with up to four PCIe slots (I kinda figured I'd need more SOCs for more slots at higher bandwidth). What I think could be interesting is that even if you priced each cluster of M2 Ultras at $6000 each, you're talking about a fully upgraded system for $24,000-$30,000 (depending on storage and other add-ons) instead of over $50,000 for the current Mac Pro.

Best of all, if Apple could keep the sockets compatible for at least 2-3 generations of SOCs, this could add incredible value to the machine.

Do I expect any of this to actually happen? I don't know, but a man can dream.

 

[AndreeOnline]

That is more so true from a narrow Apple perspective.

I think it holds up pretty universally.

Most people don't swap the CPU first chance they get. Whatever the timing was during my PC building days, it never once felt like the right move to replace just the CPU. I was probably on a 2-3 year cadence, and whenever I was ready to start planning the next build, I knew what was coming next... I waited for it, and it was always CPU, MB, RAM and GPU..... oh and drives and often a case too.

I have a very strong feeling not much has changed since then.

 

[maikerukun]

I think this would come down to how well Apple could overcome the challenges in such a design and how well software is optimized for it. Yet in my scenario even the base model would be faster as it would have 256 GB RAM, 48 core CPU, 120 core GPU, 64 NE cores with up to four PCIe slots (I kinda figured I'd need more SOCs for more slots at higher bandwidth). What I think could be interesting is that even if you priced each cluster of M2 Ultras at $6000 each, you're talking about a fully upgraded system for $24,000-$30,000 (depending on storage and other add-ons) instead of over $50,000 for the current Mac Pro.

Best of all, if Apple could keep the sockets compatible for at least 2-3 generations of SOCs, this could add incredible value to the machine.

Do I expect any of this to actually happen? I don't know, but a man can dream.

Well the problem is, I think your proposition would put it on par with a Vega II duo...it still wouldn't match even a single w6800x duo. Not even sure it would 1/1 match a 3090. Which wouldn't be enough for what we are all hoping for here...

 

[ZombiePhysicist]

Yeah, so this is the reason I tried to go out of my way to make it clear. A bit more on this further down.

A couple of people managed to chip in with their take on what drew them in and why they are staying even if interests seem to be misaligned one way or another.

Since I asked the question, I know the exact motivation behind it. I know a lot of successful companies that I don't particularly admire. I don't buy their products, and I don't spend any time at all thinking about what they might make next, or seeking out discussions about it online. They are simply 'off my radar'.

I've read quite a few posts here lately that are ripe with resentment, and I wonder why someone might feel trapped to the point that venting is all that is left.

Using the post feedback reactions, there seems to be substantial support, or agreement, for posts bashing or belittling Apple's efforts. The same support is often found for posts criticizing users that are excited about Apple's current direction.

I guess time will tell if it's just this inflamed Mac Pro issue, or if the same behavior can be seen in other threads as well, and it's simply the perception of Apple that is the culprit.

I can only be explicit and measured. How everyone reacts, in the end, is up to them. I guess we've seen a bit of both.

Same goes for Apple. It won't really matter what they say, people will read into it what they want.

What is wrong in giving that feedback. You present some kind of absolutism as reasonable. I can like most of the things a company does, but not all. For those things I do not like, I can give feedback and hopefully it's of some use. There is nothing mutually exclusive about loving the direction apple has gone with regard to its laptops and hating the direction and lack of care they paid to the Mac Pro. Being frustrated by that seems the height of rationality and reasonableness to me.

As such, the criticism of the frustration seems loaded with value judgments all its own.

Then again, and as always, YMMV.

 

[AndreeOnline]

What is wrong in giving that feedback.

Giving feedback is not only fine, it's necessary

That's what I meant when I said:

Regardless, my take on it is that it's of course fine to be irritated, frustrated, and mad about decisions that Apple makes. It doesn't matter if it's a headphone port that goes missing, a huge Mac Pro with no internal solutions for memory cards, soldered-on memory, or general architecture.

Giving feedback, in my opinion, would look something like:
"I've heard there is a risk that Apple will XXX, but it's extremely important to us who do YYY that feature ZZZ is maintained. If not, we can't NNN, which makes the platform much less useful—perhaps even impossible to use."

And this constructive feedback would ideally be sent in via the proper channels.

If someone is surprised about a missing feature to the point that emotions can't be controlled, the language might be rosey for a minute, but feedback should still be relevant, and limited, to the feature.

Also from the same post previously:

But to me, it makes a difference how feelings are expressed, or where they are coming from. If there is a deep distrust towards the company, or the people running it, I can't say what makes it worth investing so much interest in it.

I don't believe describing a company as 'word twisting' and their communications as lies as being compatible with:

"I find their Notebooks to be excellent, but they're Pro Workstations to be lacking due to XXX". It doesn't add up to me.

So, my issue is, I look at public forum venting much less as 'constructive feedback' and much more as 'human group dynamics'. People are looking to find their roles in a loosely defined, perceived hierarchy and act out their emotions for effect or to boost their forum persona. But the nature of the discussions will color the forum as a whole, so care should be taken to only represent the direction one is interested to pursue.

 

[ZombiePhysicist]

Giving feedback is not only fine, it's necessary

That's what I meant when I said:

Giving feedback, in my opinion, would look something like:
"I've heard there is a risk that Apple will XXX, but it's extremely important to us who do YYY that feature ZZZ is maintained. If not, we can't NNN, which makes the platform much less useful—perhaps even impossible to use."

And this constructive feedback would ideally be sent in via the proper channels.

If someone is surprised about a missing feature to the point that emotions can't be controlled, the language might be rosey for a minute, but feedback should still be relevant, and limited, to the feature.

Also from the same post previously:

I don't believe describing a company as 'word twisting' and their communications as lies as being compatible with:

"I find their Notebooks to be excellent, but they're Pro Workstations to be lacking due to XXX". It doesn't add up to me.

So, my issue is, I look at public forum venting much less as 'constructive feedback' and much more as 'human group dynamics'. People are looking to find their roles in a loosely defined, perceived hierarchy and act out their emotions for effect or to boost their forum persona. But the nature of the discussions will color the forum as a whole, so care should be taken to only represent the direction one is interested to pursue.

We can agree to disagree. Your thoughts on what is 'appropriate' feedback is not the only valid view. People's mileage will vary. And some entities require different intensities of feedback before it gets their attention. Apple basically will not react nor care unless there are howls of protest (e.g., you're holding it wrong) or outright abandonment of the platform (e.g., trashcan). Calm feedback seems to fall at the feet of their 'smug we know better' attitude.

That said, I see your refined sensibilities on this as being perfectly reasonable. The problem is we do not live in a perfectly reasonable world, nor does apple take feedback perfectly reasonably. That you do not see others' more intense sense of feedback of yelling "hey, you're about to walk off a cliff like you did with the trashcan, please don't" and see no apparent validity to it, well, you're entitled to your view/opinion as they are to theirs, and their view of the efficacy of your preferred modes of communication.

 

[AndreeOnline]

We can agree to disagree.

Agreed.

 

[maikerukun]

I agree, it's healthy you two agreed to agree to disagree.

 

[innerproduct]

Please apple, just give show us what you’ve got so we can either be happy or move on 😂
Have we talked about what would be the minimum system we would accept and buy? I guess we have but anyway, I think that I see these options as viable for me:
(n.b. Animation Rendering would still be done on cloud or maybe on a pc using nvidia cards. Both for perf reasons but also for freeing up main computer. 2 Macs with one as rendering box seems wasteful)
1. Dream situation: m2ultra with possibility to use regular amd gpus as well as apple mpx modules. Upgradeable ram, ssd, gpu. Will pay about 10000$ for a base system and then add dual 7900xtx. Might consider apple mpx modules if some benefit.

2. Sub 10000$ for cpu perf similar to a m1ultra and gpu simular to 4x m1 ultra. 96 gb ram. 2 tb ssd built in. This system would be ok for a few years as a designer worksation, ipr for lookdev etc but already when bought, slightly underpowered in the gpu perf.

3. Sub 6000$ for a tentative m2ultra 76 cores. Same as (2) just less gpu perf and price. Would be chaneged as quick as possible to even better system when available. Would not pay a lot extra for slots that can’t host gpus, so probably prefer an updated “studio “ in this case
This system is not what I am looking for but might get it anyway.

Other than these? I Don’t think so and everything is except (1) is not really what I was hoping for. In the unlikely event we get a xeon mac that might actually be worth it.

 

[ZombiePhysicist]

I agree, it's healthy you two agreed to agree to disagree.

Apparently that's the only valid view as agreed upon according to some.

 

[AndreeOnline]

Have we talked about what would be the minimum system we would accept and buy?

Mac Pro M2 Ultra, 64GB, 1TB: $5999

Allows for internal storage in generous TB numbers. Easy upgradable RAID SSD, drop in audio or video cards for various monitoring tasks.

Areas this computer excels:

<------------------------------------------- all of these ---------------------------------------------------------------->

Problem areas

<--- these --->

 

[AdamBuker]

Well the problem is, I think your proposition would put it on par with a Vega II duo...it still wouldn't match even a single w6800x duo. Not even sure it would 1/1 match a 3090. Which wouldn't be enough for what we are all hoping for here...

The base model might not, but I would think my maxed out version that I described earlier would for sure.

I think most people would purchase the base model, but this way those that need massive upgrades can still get them with greater value for the money.

 

[innerproduct]

Apologetics: Run this test scene from the official Otoy/Octane forum to understand the graveness of the issue at hand here when it comes to 3D and lack of horsepower in the rendering/3d department.

NEW: Octane X - is finally here! [MEGA FAQ THREAD] - Page 133 - OTOY Forums

render.otoy.com

M1Ultra: at best around 26s
Consumer/gamer PC with 1(one) 1500$ graphics card: 3s (although at slightly lower resolution, let's almost double the rendering duration as over compensation to 5s)

If you work with motion graphics using 3D (more and more common) we are not talking about a difference of a few seconds here and there. It is just not possible to compete with these machines.

It is very common to render image sequences in the 10s range at 24-30 fps. 4K is becoming standard. And you have to show your client variations quickly etc etc. If a single image takes 10 min on the best current Mac that is compatible with your software while a good but similarly priced PC can render that scene in 1 min, that becomes a difference between 5 H and 50 H. So either the animation renders out during the night or it will take a few days. So, would you buy the Mac for that task ? Of course not! You could desperately remedy the situation by paying for cloud rendering but for many that is not economically beneficial if doing this fulltime. Also, cloud have transfer issues and not always the same SW running that you want (including plugins etc).

So if the current (severely overpriced Mac Pro 2019, 12 core + dual 6800duos) at about $20000 renders the example animation in 10 H it is still usable but at a bad price/perf ratio. If that system is replaced with a system with at best 1/4 that power, don't you see the issue? No Apple spin in the world makes that new machine a viable replacement or upgrade. If this happens, Apple has effectively said: "We don't care about the 3D market and ar pulling out. There is no money for us in it, we don't care" etc. If they do this move, it's of up to them, but feels so stupid and unnecessary.

 

[JesperA]

Apologetics: Run this test scene from the official Otoy/Octane forum to understand the graveness of the issue at hand here when it comes to 3D and lack of horsepower in the rendering/3d department.

NEW: Octane X - is finally here! [MEGA FAQ THREAD] - Page 133 - OTOY Forums

render.otoy.com

M1Ultra: at best around 26s
Consumer/gamer PC with 1(one) 1500$ graphics card: 3s (although at slightly lower resolution, let's almost double the rendering duration as over compensation to 5s)

How optimized is it though? Not a long time ago they could barely get Octane X Classic to even find the GPU:s in the M1 Macs:

So have they gone from barely being competent developers enough to get their rendering software to detect the GPU:s to being fully optimized on M2 in just a couple of months?

 

[fuchsdh]

I think it holds up pretty universally.

Most people don't swap the CPU first chance they get. Whatever the timing was during my PC building days, it never once felt like the right move to replace just the CPU. I was probably on a 2-3 year cadence, and whenever I was ready to start planning the next build, I knew what was coming next... I waited for it, and it was always CPU, MB, RAM and GPU..... oh and drives and often a case too.

I have a very strong feeling not much has changed since then.

The numbers of upgraders probably bear this out. The people who just need power need configurability at purchase, fit out a system, and then mostly run it until it's paid for itself. Obviously if you have the capability to perform upgrades, some will choose to (and if they're hard to do, fewer will) but even then you're usually only talking one or two components.

By the time you get to a ship-of-theseus upgrade, you're probably into more cost-effective territory just buying a new computer, unless absolute price is your overriding concern.

But that gets to the point that there are two different groups looking to the Mac Pro, the tinkerer/enthusiasts and the professional use cases. There's some overlap, but they also are looking to very different ideas of what constitutes a hit for them.

 

[prefuse07]

very well said! ^

edit: except you forgot that, with the ship-of-theseus argument... Over time, parts generally decrease in value, so no, you are NOT better off just buying a new machine at that point, but rather upgrading gradually.

 

[mattspace]

By the time you get to a ship-of-theseus upgrade, you're probably into more cost-effective territory just buying a new computer, unless absolute price is your overriding concern.

Except that the whole time you've been upgrading, your machine has stayed at an acceptable leading edge of performance, rather than falling further, and further behind the curve.

*edit* My big problem with this "companies (or pros) don't upgrade and just buy a new computer every 4 years" argument is I know too many companies who have an inhouse IT person who does component-level upgrades, and I know too many freelancers whose work done with their personal machines is credited as being done by the studio that contracts with them, and they all do component level ugrades, because it doesn't matter if you can "write the machine off on tax" your $10000 machine at a 30% marginal tax rate is only reducing your tax bill by $3000 - you're still out of pocket by the remaining $7k.

 

[prefuse07]

Except that the whole time you've been upgrading, your machine has stayed at an acceptable leading edge of performance, rather than falling further, and further behind the curve.

which is the whole reason to upgrade in the first place.

The whole "time to upgrade? ok time to buy a new machine" mentality is completely ridiculous, especially in this day and age where everyone supposedly supports the environment.

 

[singhs.apps]

It is very common to render image sequences in the 10s range at 24-30 fps. 4K is becoming standard. And you have to show your client variations quickly etc etc. If a single image takes 10 min on the best current Mac that is compatible with your software while a good but similarly priced PC can render that scene in 1 min, that becomes a difference between 5 H and 50 H. So either the animation renders out during the night or it will take a few days. So, would you buy the Mac for that task ? Of course not! You could desperately remedy the situation by paying for cloud rendering but for many that is not economically beneficial if doing this fulltime. Also, cloud have transfer issues and not always the same SW running that you want (including plugins etc).

Right.

 

[jmho]

So have they gone from barely being competent developers enough to get their rendering software to detect the GPU:s to being fully optimized on M2 in just a couple of months?

Given that detecting GPUs correctly is about 1/1000000th of the difficulty of writing a full production grade path tracer, the most likely thing to have happened is that the Mac version was seriously under-prioritized and under-tested with all the developers using PCs because absolutely NOBODY wants to write path tracing code on a Mac Studio.

If you think waiting 25 seconds instead of 3 for renders is painful, imagine working on these things and having to wait 25 seconds instead of 3 every time you want to run your code.

 

[innerproduct]

How optimized is it though? Not a long time ago they could barely get Octane X Classic to even find the GPU:s in the M1 Macs:

View attachment 2154929

So have they gone from barely being competent developers enough to get their rendering software to detect the GPU:s to being fully optimized on M2 in just a couple of months?

Briefly: no this is not how it’s been. Octane using metal backend was released in preview during 2020. These were called PR releases and hyped massively by the company. At that time we had to pay for the privilege. Also, the standalone version is not usually used that much. People usually use octane through a plugin to their dcc. In my case Houdini but most mac folks use Cinema4d. After Preview release 14, otoy dropped support for intel macs and now octane only supports ASi. So now you cannot use the latest version of octane with your mac pro 2019 but have to stay at pr14.
The m1/m2 version is free these days, as a thank you to those of us who has been paying for many years for non working crap. The latest version is what you see some standalone variant of in the app store and that was releases quite late in the process.
Performance was actually better in pr14 than in the current version by about 10% on ASi.
So, no, it is not unoptimized. If it was, it wouldn’t help us users anyway since this is what we got.
Octane on PC is quite solid but If I had to render on Mac I would be using Redshift since it is alot more stable and supported

 

[maikerukun]

Apologetics: Run this test scene from the official Otoy/Octane forum to understand the graveness of the issue at hand here when it comes to 3D and lack of horsepower in the rendering/3d department.

NEW: Octane X - is finally here! [MEGA FAQ THREAD] - Page 133 - OTOY Forums

render.otoy.com

M1Ultra: at best around 26s
Consumer/gamer PC with 1(one) 1500$ graphics card: 3s (although at slightly lower resolution, let's almost double the rendering duration as over compensation to 5s)

If you work with motion graphics using 3D (more and more common) we are not talking about a difference of a few seconds here and there. It is just not possible to compete with these machines.

It is very common to render image sequences in the 10s range at 24-30 fps. 4K is becoming standard. And you have to show your client variations quickly etc etc. If a single image takes 10 min on the best current Mac that is compatible with your software while a good but similarly priced PC can render that scene in 1 min, that becomes a difference between 5 H and 50 H. So either the animation renders out during the night or it will take a few days. So, would you buy the Mac for that task ? Of course not! You could desperately remedy the situation by paying for cloud rendering but for many that is not economically beneficial if doing this fulltime. Also, cloud have transfer issues and not always the same SW running that you want (including plugins etc).

So if the current (severely overpriced Mac Pro 2019, 12 core + dual 6800duos) at about $20000 renders the example animation in 10 H it is still usable but at a bad price/perf ratio. If that system is replaced with a system with at best 1/4 that power, don't you see the issue? No Apple spin in the world makes that new machine a viable replacement or upgrade. If this happens, Apple has effectively said: "We don't care about the 3D market and ar pulling out. There is no money for us in it, we don't care" etc. If they do this move, it's of up to them, but feels so stupid and unnecessary.

So I'm on that message thread and posted my 2019 Mac Pro results in there and it renders the test scene in 1s flat...it is the fastest system out there when it comes to Octane for 2 GPU's or less. 3 RTX 3090's can best that by a second but that's it.

I'm only including that because you didn't post what the score was for the w6800x duo setup is...

 

[prefuse07]

So I'm on that message thread and posted my 2019 Mac Pro results in there and it renders the test scene in 1s flat...it is the fastest system out there when it comes to Octane for 2 GPU's or less. 3 RTX 3090's can best that by a second but that's it.

I'm only including that because you didn't post what the score was for the w6800x duo setup is...

That's seriously impressive!!!

 

[maikerukun]

How optimized is it though? Not a long time ago they could barely get Octane X Classic to even find the GPU:s in the M1 Macs:

View attachment 2154929

So have they gone from barely being competent developers enough to get their rendering software to detect the GPU:s to being fully optimized on M2 in just a couple of months?

Wanted you to be able to see the progress with OctaneX on the MacBooks. This is a video I just screen captured on my maxed out M1 Max MacBook Pro. It's still about 8x slower than my 2019 Mac Pro but it's definitely usable. This scene on my Mac Pro runs in realtime in OctaneX.

I just recorded on my iPhone the Mac Pro (2019) and unfortunately Octane X has been gimped in C4D 2023, and so I'm running this in C4D R23 "4 years older than the current C4D", and so the textures weren't loading for this quick example, however, when I take the time to do so, the scene runs in practically realtime...it's just so much faster and it's not even close.

So here's M1 Max MacBook Pro running Octane X Prime in C4D 2023:

And here's 2019 Mac Pro running Octane X in PR14 in C4D R23 (4 years old version)

Here's the good thing though...I can open my work files on my Mac Pro on my MacBook Pro and work on projects there...it's way slower but it's at least still usable. Which is good. This was NOT even close to possible on the Intel MacBooks, and I mean beyond unusable and moving at 1 frame per minute on the intel MacBooks so I'm not complaining about Apples MOBILE game at all...I just want the AS Mac Pro to be AT LEAST ON PAR with the 2019 Mac Pro.

 

[maikerukun]

Wanted you to be able to see the progress with OctaneX on the MacBooks. This is a video I just screen captured on my maxed out M1 Max MacBook Pro. It's still about 8x slower than my 2019 Mac Pro but it's definitely usable. This scene on my Mac Pro runs in realtime in OctaneX.

I just recorded on my iPhone the Mac Pro (2019) and unfortunately Octane X has been gimped in C4D 2023, and so I'm running this in C4D R23 "4 years older than the current C4D", and so the textures weren't loading for this quick example, however, when I take the time to do so, the scene runs in practically realtime...it's just so much faster and it's not even close.

So here's M1 Max MacBook Pro running Octane X Prime in C4D 2023:
View attachment 2155283

And here's 2019 Mac Pro running Octane X in PR14 in C4D R23 (4 years old version)
View attachment 2155284

Here's the good thing though...I can open my work files on my Mac Pro on my MacBook Pro and work on projects there...it's way slower but it's at least still usable. Which is good. This was NOT even close to possible on the Intel MacBooks, and I mean beyond unusable and moving at 1 frame per minute on the intel MacBooks so I'm not complaining about Apples MOBILE game at all...I just want the AS Mac Pro to be AT LEAST ON PAR with the 2019 Mac Pro.

When I have time, I'll definitely use a scene that has all the materials loaded on both versions and about 10 times heavier so that you can see a strong difference between the two but for now hopefully this is a comparison that makes a little sense...