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

[ZombiePhysicist]

Time Photographer Diana Walker's Favorite Shots of Steve Jobs

Longtime Time Magazine photographer Diana Walker has shared a few of her favorite photographs of Steve Jobs, and briefly sharing her experiences...

www.macrumors.com

That would have been the PowerMac G5 in those photos as they were taken in 2004.

I dont think the 30" display was released in the G5 days.

 

[MacHeritage]

I dont think the 30" display was released in the G5 days.

Yes, it certainly was released in the G5 days. June 28, 2004 to be exact. Other then my memory of the release and Steve showing it off with G5 power, here is the press release from Apple: https://www.apple.com/newsroom/2004/06/28Apple-Unveils-30-Cinema-HD-Flat-Display/

 

[mattspace]

View attachment 2157381

There is so much insight to be had with that photo. The expandable Mac. That 30" display would work with other Macs at the time. The snaggle of external devices. Even look at the back of the display, looks like some USB dongle in there. And the absolute mess of a personal working space (which I am all for...shows he wants lots of random access to his stuff).

Yea, truly cool find. Thanks for sharing it.

He likes the cheesegrater so much, he has a scale model of it on the desk, propping up a photo *lol*.

 

[ZombiePhysicist]

Yes, it certainly was released in the G5 days. June 28, 2004 to be exact. Other then my memory of the release and Steve showing it off with G5 power, here is the press release from Apple: https://www.apple.com/newsroom/2004/06/28Apple-Unveils-30-Cinema-HD-Flat-Display/

I sit corrected! Thanks for the check!

 

[avkills]

Anyone thinking Apple may do a Super Bowl ad during the halftime show to promote the new Mac Pro? Doubtful, as they would probably hype Apple Music instead. But wouldn't that be cool; if they did something as good as the iconic "1984" ad.

 

[mode11]

I'm just hoping we hear something by WWDC.

 

[innerproduct]

They teased the macpro at the release of the “studio “ last year. Surely we won’t have to wait until wwdc?

 

[mode11]

Just trying to temper expectations. I find when it comes to Mac Pro news, the best approach is to take your worst possible estimate, then multiply it by two. Then you'll only be slightly disappointed.

I remember last summer thinking we should have details of the new Mac Pro by about mid-October (2022).

 

[MacPoulet]

They teased the macpro at the release of the “studio “ last year. Surely we won’t have to wait until wwdc?

NAB is around the corner in April...

 

[StellarVixen]

View attachment 2157381

There is so much insight to be had with that photo. The expandable Mac. That 30" display would work with other Macs at the time. The snaggle of external devices. Even look at the back of the display, looks like some USB dongle in there. And the absolute mess of a personal working space (which I am all for...shows he wants lots of random access to his stuff).

Yea, truly cool find. Thanks for sharing it.

Love it!

Messy office = busy office.

 

[ZombiePhysicist]

View attachment 2157381

There is so much insight to be had with that photo. The expandable Mac. That 30" display would work with other Macs at the time. The snaggle of external devices. Even look at the back of the display, looks like some USB dongle in there. And the absolute mess of a personal working space (which I am all for...shows he wants lots of random access to his stuff).

Yea, truly cool find. Thanks for sharing it.

Weirdly tripped over this video today. This guy was just the GOAT. First, check out the gorgeous color next in the background. Sadly, in many ways, those NeXT systems are still better than macOS today, not to mention the way macOS has degenerated into s*** the last 10 years or so...

But more importantly, listen to how accurate this f***ers prediction about TODAY was in 1991. CRAZY. This guy didnt just see where the puck was going, he saw decades of games in advance in minutia. Whenever I think maybe this guy was a mere mortal, you see something like this, and you realize, he was the GOAT of this industry, by far. Relevant part starts at 1:39 marker, but entire interview is great.

 

[MacHeritage]

I sit corrected! Thanks for the check!

You are very welcome. I remember it very well and I couldn't wait to one day own one. Well the family purchased one of the very last 30-inch Displays (second to last?) available from Apple from their refurb store in 2010, long after "new" ones were gone. Took me 6 years before we purchased one but it has lasted all this time for everything I do computer wise. The 30" Cinema Display was used with the 2007 MacBook Pro until we purchased the Mac Pro 2010, also refurb. It is still my main display since 2011 to this very day.

Well worth every penny. And the design hasn't aged a bit. Would look great beside a 2019 Mac Pro, just like my 2010.

 

[MacHeritage]

Anyone thinking Apple may do a Super Bowl ad during the halftime show to promote the new Mac Pro? Doubtful, as they would probably hype Apple Music instead. But wouldn't that be cool; if they did something as good as the iconic "1984" ad.

I wish! I don't think that Apple Inc. exists today. When they didn't have the iPhone and iPod, yes they might have. But the Apple of 2023, no. They are just promoting services where you can rent your life from them and own nothing, that I could care less about.

 

[ZombiePhysicist]

You are very welcome. I remember it very well and I couldn't wait to one day own one. Well the family purchased one of the very last 30-inch Displays (second to last?) available from Apple from their refurb store in 2010, long after "new" ones were gone. Took me 6 years before we purchased one but it has lasted all this time for everything I do computer wise. The 30" Cinema Display was used with the 2007 MacBook Pro until we purchased the Mac Pro 2010, also refurb. It is still my main display since 2011 to this very day.

Well worth every penny. And the design hasn't aged a bit. Would look great beside a 2019 Mac Pro, just like my 2010.

You're preaching to the choir. I think I have 8 of them. 6 were for active use and bought a couple of spares. Now I actively only use 3, and replaced 3 with a really nice cheese grater (it actually has aluminum cheese grater bezel) Samsung 8k 85" screen. I've had some of mine I think since 2006 or so. So far only one is somewhat dying. It just will dim during the day, and you increase the brightness and then over the course of the day it will dim down again. Rest still seem to be working well. Not since most of them have been on 24hrs a day for I guess around 16 years straight!

 

[innerproduct]

Been away from the forum a few days. Maybe I gained some perspective 😂
Just checked out current mac benchmarks on the redshift forum and it is basically like this:
Mp2019 with one 5700x : about 10:30 min
Mp2019 with two 5700x : about 6 min
Mp2019 with 6800duo: about 3:30
Mp2019 with two 6800duo: about 2 min
M1max 32core: 9:30
M1ultra 64core: 6:30
M2max 30core: 9
M2max 38core: 7:30

So extrapolating we get
M2ultra76core: 3:45-5min

That is at best as you all clearly can see a massive downgrade for those who use macs for 3d rendering.
This is not reasonable development in any way and simply not what will happen with the mac pro.
Let’s instead think of what kind of system that would entice a current mac pro owner to upgrade? And obviously at least 50% perf gain. Probably more.
Let’s go with more: 2x.
That would mean the the benchmark should be rendered in 1 min.
Let’s now take that as the starting point for what the mac pro needs to be.
Assuming 70% efficiency when scaling to high core counts, a dual m2ultra might break 2:30 but not more, so in the end, not even the mythical “extreme” would be good enough. What about if it were clocked higher? Well, that would mean linear scaling of perf but worse thermal scaling. Let’s say 25% max gains and we will still be to slow to beat the 2019 mac pro fully loaded.
Clearly we have a few options at this point: new intel machine with 7900duos as a stopgap until ASi is mature enough and maybe release a “low end”
ASi mac pro m2 ultra parallel to this machine.
Or: there will be supported for accelerator cards on the ASi mac as well.

What is most plausible? From a business perspective? Marketing?

 

[Boil]

Been away from the forum a few days. Maybe I gained some perspective 😂
Just checked out current mac benchmarks on the redshift forum and it is basically like this:
Mp2019 with one 5700x : about 10:30 min
Mp2019 with two 5700x : about 6 min
Mp2019 with 6800duo: about 3:30
Mp2019 with two 6800duo: about 2 min
M1max 32core: 9:30
M1ultra 64core: 6:30
M2max 30core: 9
M2max 38core: 7:30

So extrapolating we get
M2ultra76core: 3:45-5min

That is at best as you all clearly can see a massive downgrade for those who use macs for 3d rendering.
This is not reasonable development in any way and simply not what will happen with the mac pro.
Let’s instead think of what kind of system that would entice a current mac pro owner to upgrade? And obviously at least 50% perf gain. Probably more.
Let’s go with more: 2x.
That would mean the the benchmark should be rendered in 1 min.
Let’s now take that as the starting point for what the mac pro needs to be.
Assuming 70% efficiency when scaling to high core counts, a dual m2ultra might break 2:30 but not more, so in the end, not even the mythical “extreme” would be good enough. What about if it were clocked higher? Well, that would mean linear scaling of perf but worse thermal scaling. Let’s say 25% max gains and we will still be to slow to beat the 2019 mac pro fully loaded.
Clearly we have a few options at this point: new intel machine with 7900duos as a stopgap until ASi is mature enough and maybe release a “low end”
ASi mac pro m2 ultra parallel to this machine.
Or: there will be supported for accelerator cards on the ASi mac as well.

What is most plausible? From a business perspective? Marketing?

Not really a fair comparison, a SFF desktop (with GPU performance of a NON-Duo 6800) versus a GPU-wise fully-loaded (so four 6800 chips) 2019 Intel Mac Pro?

Of course the 7,1 Mac Pro is going to win that particular render speed test; but what about the actual workflow, which machine will give (assuming properly ASi-optimized DCC software) the faster viewport response?

One way to improve render speeds would be ASi GPGPU add-in card(s); keep working with the iGPU and send render jobs to the GPGPU(s).

A way to improve the iGPU may be thru asymmetrical SoCs, pairing a "standard" M2 Max (12C/38G) SoC with a "GPU-specific" (0C/57G) SoC, giving one a 12-core (8P/4E) CPU & 95-core GPU. A M2 Extreme variant would have a 24-core (16P/8E) CPU & a 190-core GPU.

More GPU horsepower while working in the DCC software of choice, without having a bunch of CPU cores one may not really need?

As for the ASi GPGPUs, these would utilize the "GPU-specific" SoCs, with options of two, four, or eight of these special SoCs per card, and the appropriate RAM to match.

The "Pro" move would be to offer a rackmount chassis for a number of these ASi GPGPUs, allowing the DCC artist to work on a Mn Ultra (the 12C/95G variant, of course) Mac Studio or a Mn Extreme (24C/190G) Mac Pro Cube, whilst sending render jobs to a renderfarm consisting of one or one hundred or more of these ASi GPGPU render boxes!

Maybe we even see a 4-way SoC with one "regular" M2 Max chip & three "GPU-specific" chips, 12-core (8P/4E) CPU & 209-core GPU?

Apple HAS TO show us SOMETHING by WWDC 2023; right, RIGHT? ;^p

 

[deconstruct60]

Clearly we have a few options at this point: new intel machine with 7900duos as a stopgap until ASi is mature enough and maybe release a “low end”

DUO 7900XT probably really isn't much of an option. First, it is questionable that it has a GPU-to-GPU InfiniFabric connection feature (that may only be a 'Pro' compute GPU package only feature going forward.). AMD already has a two die in a single package solution on the Pro compute side; MI210 (and more coming in MI300 series). If they needed something with two dies on the consumer side in the next year or two , then pretty likely it would be 3D packaged into one very expensive unit; not linked packages. Second, It blows the the TDP budget (the 7900XT is 315W ) 2 * 315 = 630W. Actually in part why would want to 3D package them together to dramatically lower the power consumption of IF linking them together since so far over 'normal' budget on power anyway. [ Even if Apple dials back the 7900XT to 275W , that would still be 50W over budget. ]

Note: there was no 6900 Duo either. ( even though the solo 6900 cost more than a 6800 duo. Price wise nobody is going to buy twice that again. )


In the current MP 2019 the two MPX bays collectively have about 1000W of power to hand out. Each bay maxes out about 500W. Pretty doubtful Apple is going to sign up for card that blow far past 500W per card/module. The MI210 is around 300W. Apple's 500W 'line in the sand' is probably a boundary they are not going to move higher. It unlikely Apple is going to want to crank the aggregate system dGPU power budget so high that they get cramped on what kind of top end SoC they can pack into the remaining 300W or so of total system power budget. If anything, long term, they'd be looking to lower the GPU power budget under 1000W; not over.



Maybe a small chance get a 7950 XTX where they crank up the cache and clocks to run the board power up near 400W and just get one of those. But not a 'Duo' implementation, but just one higher power consuming single.

ASi mac pro m2 ultra parallel to this machine.
Or: there will be supported for accelerator cards on the ASi mac as well.

An accelerator card that could be deployed any Mac would be more plausible ( take load/unload time hit running it through TBv4 to most Mac but faster when in a Mac Pro. same drivers for all of the systems). Trying to shink the unit sales absolutely as small as possible isn't going to help bring this to market at all.

 

[innerproduct]

@Boil and @deconstruct60 , in my opinion, you are going way to much into details and missing the main topic of what I wrote.
The point is, even with a m2max “extreme” with overclocking, we would be looking at a macpro with less compute power than the current version.
It is not a matter of “fairness” or minutiae about amd components.
Do you really think it would be a good idea from a business perspective to launch a mac pro that at best is on par with a 4 year old machine when it comes to compute ? It just doesn’t make sense.
So, my thinking is that either apple couldn’t 5 years ago calculate the number of ASi gpu cores it would take to reach a certain target. (I find that highly unlikely since it is easily done even for forum members. ) Or, they decided to drop the mac pro line. (Also highly unlikely considering all the hype for that machine and that they still mentioned that it was upcoming when they launched the studio)
So, going with the narrative that apple are able to design chips, have more info than common people at forums, and that they want to impress old macpro buyers enough to buy a new machine, I can only arrive at:
- apple will release a macpro that has compute power at least at a maxed out mp2019, but cheaper
- apple will also release lower specced versions
- compute will not be solved simply by using 4 m2 max chips
- if there has been major problems with the mac pro ASi tech, maybe there will be a staggered launch where a low/mid range ASi solution lives paralell to a slightly updated intel version

To me, a staggered launch would make most sense. What do you think?

 

[AndreeOnline]

It is not a matter of “fairness” or minutiae about amd components.

You are totally right. It doesn't matter why a certain design would fall short, just the fact that it does.

Do you really think it would be a good idea from a business perspective to launch a mac pro that at best is on par with a 4 year old machine when it comes to compute ? It just doesn’t make sense.

Let's get back to the business case shortly.

So, my thinking is that either apple couldn’t calculate... or, they decided to drop the mac pro line.

I'm sure you'll be the first to admit that those two (condensed) scenarios are far from exhaustive.

So, going with the narrative that Apple wants to impress old Mac Pro buyers enough to buy a new machine,

First quick question: do you think it's necessary for Apple to impress old Mac Pro owners? Would they be OK with replacing all of them with 2x the number 'new to the Mac Pro' customers? What about impressing Mac Pro users that stepped in via the eBay Mac Pro market and pimped 10+ year-old machines with NewEgg parts—is that important too?

I can only arrive at:
-What do you think?

I think it will be hard to replace a GPU-loaded 2019 Mac Pro with an Apple Silicon one.
At the same time, we are in a somewhat similar situation now to when some computers offered hardware HEVC encoding and decoding and some didn't. You could have a MacBook Air running circles playing back and exporting some video files compared to a dual GPU Mac Pro that "people felt" should be faster.

Nvidia's Optix works great for them when raytracing and can spit out a frame in 4 seconds whereas a "powerful" Mac might need 30 seconds. That single use case doesn't reflect the quality of the GPUs necessarily, but we're circling back to 'it doesn't matter why'. It's the 'have' vs 'have not'.

Sooner rather than later, it would be a good idea if Apple would show some technology that addresses major GPU weaknesses. If not, they will be giving away that market for sure.

 

[innerproduct]

You are totally right. It doesn't matter why a certain design would fall short, just the fact that it does.


Let's get back to the business case shortly.


I'm sure you'll be the first to admit that those two (condensed) scenarios are far from exhaustive.


First quick question: do you think it's necessary for Apple to impress old Mac Pro owners? Would they be OK with replacing all of them with 2x the number 'new to the Mac Pro' customers? What about impressing Mac Pro users that stepped in via the eBay Mac Pro market and pimped 10+ year-old machines with NewEgg parts—is that important too?


I think it will be hard to replace a GPU-loaded 2019 Mac Pro with an Apple Silicon one.
At the same time, we are in a somewhat similar situation now to when some computers offered hardware HEVC encoding and decoding and some didn't. You could have a MacBook Air running circles playing back and exporting some video files compared to a dual GPU Mac Pro that "people felt" should be faster.

Nvidia's Optix works great for them when raytracing and can spit out a frame in 4 seconds whereas a "powerful" Mac might need 30 seconds. That single use case doesn't reflect the quality of the GPUs necessarily, but we're circling back to 'it doesn't matter why'. It's the 'have' vs 'have not'.

Sooner rather than later, it would be a good idea if Apple would show some technology that addresses major GPU weaknesses. If not, they will be giving away that market for sure.

In short, sure, agree with a lot of what you say. Basically I believe I say that I think Apple have a solid understanding of what the bottlenecks are and they probably have a solution in order to make ASi macs attractive for more use cases. Maybe some of their planning didn’t pan out and they have to slightly change the launch. Will they have RT hw to solve some really power hungry tasks? Probably in the long term, but since it was not available (or publicly enabled?!) that leaves us to speculate.
For some rendering tasks though, RT hw helps very little (advanced shaders) and there is also all tasks that would rather have general purpose hw. Truth be told, a really fast CPU is supernice for rendering since it is so responsive. A 48 core CPU coupled with a cloud rendering backend is a very clean solution.

Hehe, in the end I guess we’ll see.

 

[goMac]

First quick question: do you think it's necessary for Apple to impress old Mac Pro owners? Would they be OK with replacing all of them with 2x the number 'new to the Mac Pro' customers?

If they're not competitive with Windows PCs, they're not going to be getting many 'new to the Mac Pro' customers.

A substantial price cut maybe would get them there. If they could get back into the $2500-$3000 starting price range. But I have doubts.

 

[AndreeOnline]

If they're not competitive with Windows PCs, they're not going to be getting many 'new to the Mac Pro' customers.

OK.

I don't share that opinion. Once I stopped building PCs and went over to Mac, I don't compare anymore. In my mind Macs and PCs don't compete, since I'm after macOS and I only get that with Macs.
But I can totally see how the platforms can go head-to-head if the OS is secondary and stays in the background. It's pretty common among "1 app pros" where you spend all your time in a single application. Once you're in, you're in.

I don't think Apple will win over any "single use case PC builders" that rely heavily on Nvidia, for example.

Regardless, I feel price–performance is better with Mac, overall. Pretty substantially actually. I don't think we've seen the full effects of the switch to Apple Silicon yet, in terms of hardware and software integration.

 

[smckenzie]

I can get in to the 1m30s on the redshift benchmark if I chuck a 6900xt into slot 5 along with the 2 duos.

So any new Mac Pro would need to best that otherwise what’s the point? Some with 4090’s are getting 40 secs. Crazy.

And for kicks I got a nvidia a4000 offf eBay that I’m gonna put in for use in bootcamp. Doesn’t sound like any new MacPro is going to offer me any of this flexibility which is half the reason I got one.

 

[mode11]

So extrapolating we get
M2ultra76core: 3:45-5min

Some with 4090’s are getting 40 secs. Crazy.

Are you referring to the results here: https://www.cgdirector.com/redshift-benchmark-results/?

If so, a single RTX4080 will get you 1:47, a single RTX4090 1:16, and dual RTX4090's 47 secs. At scan.co.uk currently, RTX4080's are about £1250; RTX4090's £1750. So Apple will have their work cut out to approach the value proposition of a PC for 3D rendering performance. Using the cost of a fully-loaded 2019 Mac Pro for the comparison is soft-balling Apple somewhat.

Personally, I think I'm heading in the direction of a Studio Max, though will wait for the Mac Pro announcement in order to understand Apple's strategy for high-end ASi desktops. I'd also prefer to get an M2 Studio; apart from it generally being a good idea to skip first-gen Apple products, it will also confirm Apple's commitment to the model after a string of one-off wonders - including rumours now of the Studio itself being a holdover model until the Mac Pro's release.

 

[majus]

Personally, I think I'm heading in the direction of a Studio Max, though will wait for the Mac Pro announcement in order to understand Apple's strategy for high-end ASi desktops.

It won't be much longer. Mac Pro and other good things on March 7. Just my two cents.