Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
Another really bad sign of how apple is not even in any pro game anymore. Go to any pro gear like a pro scanner. All of the devices have windows, and somewhat incredibly, linux drivers. And absolutely nothing for apple.


Like none of the workgroup, high speed, or production scanners here have any Mac support, but almost all if not all have linux support. Let that sink in.


On that right hand side of that page is the best selling scanners. #1 Epson DS 530 II

quick check of epson page...



Windows/MacOS supported.
( in part because some aspects of scanning on epson is SilverFast ... which is keeping up with M-series changes.
https://www.silverfast.com/silverfast-9-supports-the-new-macos-13-ventura/ )


Ricoh/Fujistu fi-8170 .. macOS supported.

The Canon M260 has a VueScan driver

Kodak Alaris scanners ... how niche is that?

The stuff on that 'high speed' tab is all > $3,000 scanners. Best selling $300-1,000 . This has more to do with scanner volume sold than "Pro vs non Pro" support. The extremely expensive high speed scanners are typically hooked to servers that ingest into a repository. As long as the pro macs that are used by individual users can access and view the documents , there is no huge numbers loss here (any more than Apple giving up on relatively very expensive laser printers. ). The more expensive the scanner the more it is closer to being an embedded appliance. **




Apple is going to spend far more time and energy and money on making the Apple camera that folks already have bought be a better document scanner than in trying to push folks toward buying a $3,000 device to do exactly the same thing at the large scale user scale. Apple is going to whittle down scanners sold like they whittled down the mainstream camera market.



So apple doesnt support basic standards like U.3 drives

U.3 drives should support NVMe. NVMe drive in a MP 2023 should work just fine.

U.2 Sonnet card. ... macOS support in MP 2023.

There is no huge block to some vendor doing a EDSFF ( E1 , E3 ) SSD format add-in card that would work via NVMe with MP 2023 either. It is more a matter of enough buyers to justify the validation work that some 'show stopper' that Apple is blocking the product with.

The bulk of users though are going to cobble together multiple M.2's onto a cheaper aggregate set up for better $/TB costs though. That's the bigger 'block'.


and professional level scanners. Just more and more reasons why pros have left this platform.

New Epson 13000XL ... macOS support.

V850 Pro works also.


macOS larger footprint in photography leads to larger volume which leads to more robust support. Super high volume archving paper documents into big data storage isn't (and wasn't ) a big macOS lynchpin market. APFS isn't a 'bit rot' protection file system to super long term , high bulk storage. Even without the change to M-series ( on Intel) it didn't have huge bulk archival deep traction either. It isn't about keeping people buying the most expensive , single function gadget possible.



** P.S. Yes there is Linux Support, but on the Windows side there is Windows server certifications right alongside any Linux support. It is the same effectively cobble an appliance together with a slow-moving/slow-updates server sytsem coupled to the very expensive scanner to be used as a shared use (server) resource for a large group. Linux showing up in a server context is not as surprising as you are trying to make it out to be.
 
Last edited:

ZombiePhysicist

Suspended
Original poster
May 22, 2014
2,884
2,794
On that right hand side of that page is the best selling scanners. #1 Epson DS 530 II

quick check of epson page...



Windows/MacOS supported.
( in part because some aspects of scanning on epson is SilverFast ... which is keeping up with M-series changes.
https://www.silverfast.com/silverfast-9-supports-the-new-macos-13-ventura/ )


Ricoh/Fujistu fi-8170 .. macOS supported.

The Canon M260 has a VueScan driver

Kodak Alaris scanners ... how niche is that?

The stuff on that 'high speed' tab is all > $3,000 scanners. Best selling $300-1,000 . This has more to do with scanner volume sold than "Pro vs non Pro" support. The extremely expensive high speed scanners are typically hooked to servers that ingest into a repository. As long as the pro macs that are used by individual users can access and view the documents , there is no huge numbers loss here (any more than Apple giving up on relatively very expensive laser printers. ). The more expensive the scanner the more it is closer to being an embedded appliance. **




Apple is going to spend far more time and energy and money on making the Apple camera that folks already have bought be a better document scanner than in trying to push folks toward buying a $3,000 device to do exactly the same thing at the large scale user scale. Apple is going to whittle down scanners sold like they whittled down the mainstream camera market.





U.3 drives should support NVMe. NVMe drive in a MP 2023 should work just fine.

U.2 Sonnet card. ... macOS support in MP 2023.

There is no huge block to some vendor doing a EDSFF ( E1 , E3 ) SSD format add-in card that would work via NVMe with MP 2023 either. It is more a matter of enough buyers to justify the validation work that some 'show stopper' that Apple is blocking the product with.

The bulk of users though are going to cobble together multiple M.2's onto a cheaper aggregate set up for better $/TB costs though. That's the bigger 'block'.




New Epson 13000XL ... macOS support.

V850 Pro works also.


macOS larger footprint in photography leads to larger volume which leads to more robust support. Super high volume archving paper documents into big data storage isn't (and wasn't ) a big macOS lynchpin market. APFS isn't a 'bit rot' protection file system to super long term , high bulk storage. Even without the change to M-series ( on Intel) it didn't have huge bulk archival deep traction either. It isn't about keeping people buying the most expensive , single function gadget possible.



** P.S. Yes there is Linux Support, but on the Windows side there is Windows server certifications right alongside any Linux support. It is the same effectively cobble an appliance together with a slow-moving/slow-updates server sytsem coupled to the very expensive scanner to be used as a shared use (server) resource for a large group. Linux showing up in a server context is not as surprising as you are trying to make it out to be.
You cherry pick useless (to the point asserted about professional scanners with) 'best selling' scanners and ignore the huge swaths of workgroup/highspeed/production (ie professional) scanners I discussed and that do not work, then unwittingly conceding the point, yet delusionally purporting the opposite, by showing some 3rd party software necessary to make a scanner work since indeed the manufacturer doesnt supply it.

And if that noise wasn't enough, then ignored entire threads dedicated to just how much drives do not work, that most U.2 drives do not work and no known U.3 drives work and talking about NVMe working on a 2023.

tenor-1292521532.gif


Your post has now been deemed completely disingenuous and non-credible and given such weight moving forward with your completely cluttering nonsense 'analysis'.
 
Last edited:

ZombiePhysicist

Suspended
Original poster
May 22, 2014
2,884
2,794
Well now that we see what the A17 Pro is, I have a bad feeling about the M3. 10% boost in CPU. 20% boost in GPU? Not sure how meaningful raytracing is when so few apps/games use it.

Sure, the M series will have room for a few more GPU/CPU cores, but not sure it will be a big enough boost to matter. Without a double Ultra (Extreme) edition, I fear it will be game over for any Mac Pro variant. Hope I'm wrong and perhaps I'm not extrapolating it correctly?
 
Last edited:

AdamBuker

macrumors regular
Mar 1, 2018
121
185
Well now that we see what the A17 Pro is, I have a bad feeling about the M3. 10% boost in CPU. 20% boost in GPU? Not sure how meaningful raytracing is when so few apps/games use it.

Sure, the M series will have room for a few more GPU/CPU cores, but not sure it will be a big enough boost to matter. Without a double Ultra (Extreme) edition, I fear it will be game over for any Mac Pro variant. Hope I'm wrong and perhaps I'm not extrapolating it correctly?
Boil pointed this post from a different forum on another thread. I have no idea if this guy's math is right, but if so, it's at least a step in the right direction, other issues notwithstanding.

Here's the quote below:

Just some napkin math with blender open data (v 3.6.0): M2 Ultra scores 3447.15; naively extrapolating 20% GPU compute improvement with 4x raytracing speed results in a score of ~16546.32. RTX 4090 scores 13100.33. Even without any exciting M3 surprises, that's looking pretty good.
 

zephonic

macrumors 65816
Feb 7, 2011
1,314
709
greater L.A. area
Also I struggle to see how Apple justifies the prices for these higher end systems as anything other than "Hey we spent a lot of money on R&D and you guys have money so if you could just give it to us that would be great".

Frankly, I'm not sure how much R&D was needed? It looked like they jerry-rigged a PCIe controller onto M2U and stuck it in the 7,1 case.

$3K over an identically spec'ed Studio buys 2 extra ThB ports, 1 extra USB-A port, one extra HDMI port, one extra Ethernet port, and 6 PCIe slots. If somebody really needs those PCIe slots, well ok then...Otherwise, I just don't see it.

If enough people point out what a loser this machine is, like they did with the trashcan, they can (and have been) embarrassed. It's doable.

But this is a very different time, and a different market. And I suspect many cMP users will be serviced just fine with a MacStudio, making the already small market for the Mac Pro even smaller.

I mean, ten years ago I absolutely needed a Mac Pro for my work flow. Last year I replaced my cMP with a M1Max MBP, in favor of an M1U Mac Studio. I could use the extra power, but the convenience of the laptop was decisive.

this isn't the PPC->x86 transition at all which finished around 6 months UNDER the two year deadline. This time Apple is at least 6 months OVER the deadline. And in the first year of the transition Apple sold millions more x86 systems than they sold PPC ones in that transition year. The inertia is way bigger this time and Apple screwed it up. If Apple finished quicker than projected , then they would have a mildly creditable excuse to pull in termination. However, when they screw up AND they want to leave early.... and have already pre-charged folks for upgrades taking off the table ... that has got class action suit written all over it. Bigger class and documented company screw up by 'greedy trillion dollar' company ... good luck with that.
...
Every single step of this x86->M series transition as looked less and less like the x86 transition . In 2006 Apple was doing three different , divergent ports all at the same time. ( PPC , Arm , x86). This is a convergence , not divergence. They threw 32-bit x86 out the window before even making the transition make the convergence even bigger.
...
Apple themselves defines the obsolete windows as 5 years. And they have overtly reinforced that pattern last 3 years will basically 5 year rolls off to OS support.
...
If the high priority goal was to get them onto Apple silicon why was Apple shipping Intel Mini 2 YEARS after the first M1 Mini ! You cannot both BLOCK folks from getting off the x86 train (no alternative to switch to) and then punish them because they got off later than the 2 year schedule and seriously expect to be taken as being reasonable.
...
To stop supporting Intel Mac is relatively necessary prequistie should be to provide something to transition to. "Get off the Train" but don't open the door so they can get off. WTF?

Speculating here, but could the pandemic/lockdown and subsequent supply chain issues have screwed up their carefully constructed timeline?
 
Last edited:

ZombiePhysicist

Suspended
Original poster
May 22, 2014
2,884
2,794
Boil pointed this post from a different forum on another thread. I have no idea if this guy's math is right, but if so, it's at least a step in the right direction, other issues notwithstanding.

Here's the quote below:

Just some napkin math with blender open data (v 3.6.0): M2 Ultra scores 3447.15; naively extrapolating 20% GPU compute improvement with 4x raytracing speed results in a score of ~16546.32. RTX 4090 scores 13100.33. Even without any exciting M3 surprises, that's looking pretty good.

4x raytracing seems dubious.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
Boil pointed this post from a different forum on another thread. I have no idea if this guy's math is right, but if so, it's at least a step in the right direction, other issues notwithstanding.

Here's the quote below:

Just some napkin math with blender open data (v 3.6.0): M2 Ultra scores 3447.15; naively extrapolating 20% GPU compute improvement with 4x raytracing speed results in a score of ~16546.32. RTX 4090 scores 13100.33. Even without any exciting M3 surprises, that's looking pretty good.

The math there is a bit flawed; at least in context.

A16 5 GPU cores.
A17 6 GPU cores.

Adding one more GPU core to a set of five is 6/5 ==> 1.2 ( 20% more ).

That doesn't look like a huge win there. I suspect they are somewhat throttled by squatting on LPDDR5 for the memory ( they have no 'raw' bandwidth increase, but there is probably some caching upgrades. )

It looks like Apple has built some bigger cores that do more stuff, but baseline stuff isn't a big win.

The M2 used two of those 5 core clusters ( 10). But binned model had one core off in each cluster ( 8 )
M2 Pro 19 cores max for four of those clusters. So at least one off.
M2 Max 38 cores max for eight of those clusters . So at least two off.

M3 Max could be in range of 40 to 46 cores ( one off in each cluster or just similar two clusters with one down) .
The 46 would be 20% increase. The 40 cores would be just 6% . The midpoint of downgraded clusters is 43 cores at 13% .

The folks who believe TSMC N3B still has horrible yield, the 40 count could be closer. I doubt it is that bad, but may get a bit less than 20% because bigger dies means lower yields. ( Adding LPDDR5X would likely help more at larger die scale ( but much smaller unit volume) and incrementally bigger power budgets. Even more so in the Ultra configuration where some GPU cores are going to be even further away from the RAM controllers and there is more 'traffic' on the internal bus network. ) . The M2 was done a relatively very mature N5 variant and still had to bin. There is going to be a defect impact here with N3B ( or N3E if want to hand wave those into existence. )




The 4x is likely very highly dependent upon the scene. Some of Apple's hardware RT patents are targeting 'better efficiency' to get more throughput. If it is pointless to trace a ray, they abandoned sooner rather than later. That saves lots of energy and time not wasted on computation that will have no material gain. But if there is fewer 'pointless' rays then it will also 'save' less 'wasted' computations ( i.e., will have to do more work to complete the scene. )
 

mattspace

macrumors 68040
Jun 5, 2013
3,342
2,975
Australia
Well, it is an "up to" statement. Which, I guess, depends a lot on how well a particular software is optimised.

View attachment 2260423

Ahh yes software optimisations, the "it'll be here when tomorrow arrives" of performance.

I can't recall any example in the almost 30 years I've been in this game when promised software optimisations actually made something faster on the same hardware. The optimisation job always seems to slip to the next purchase round.

I note with some (delicious) schadenfreude that Unity has chosen the last 24 hours to completely implode their ecosystem and developer relations story... right in time for Apple to have bet the farm on it as a platform for Apple's AR & Game ambitions.
 

chfilm

macrumors 68040
Nov 15, 2012
3,423
2,108
Berlin
Ahh yes software optimisations, the "it'll be here when tomorrow arrives" of performance.

I can't recall any example in the almost 30 years I've been in this game when promised software optimisations actually made something faster on the same hardware. The optimisation job always seems to slip to the next purchase round.

I note with some (delicious) schadenfreude that Unity has chosen the last 24 hours to completely implode their ecosystem and developer relations story... right in time for Apple to have bet the farm on it as a platform for Apple's AR & Game ambitions.
What happened with unity?
 

chfilm

macrumors 68040
Nov 15, 2012
3,423
2,108
Berlin
So, now three months later, has anybody taken the plunge and bought the 14,8?

Happy? Regrets?
I went to an m2 Mac Studio from my 2019 and am very happy indeed. It’s definitely much noticeably faster, both in day to day and high end use cases.. for example I could never playback neighboring clips smoothly with grade and noise reduction applied in resolve and now I can. It renders a 30 minute ProRes timeline with full grading applied in 1,5 minutes 😵‍💫. The only thing that bothers me is the external raid enclosure which doesn’t seem to be nearly as stable or fast as the internal pcie raid that I had in the Mac Pro.
 
Last edited:

Mac3Duser

macrumors regular
Aug 26, 2021
183
139
I went to an m2 Mac Studio from my 2019 and am very happy indeed. It’s definitely much noticeably faster, both in day to day and high end use cases.. for example I could never playback neighboring clips smoothly with grade and noise reduction applied in resolve and now I can. It renders a 30 minute ProRes timeline with full grading applied in 1,5 minutes 😵‍💫. The only thing that bothers me is the external raid enclosure which doesn’t seem to be nearly as stable or fast as the internal pcie raid that I had in the Mac Pro.
My Macbook Pro M2 max is faster than my Mac Pro 7.1, but the Mac Pro 7.1 is not the same kind of machine. The 7.1 is a workstation, slow, but with lots of PCI lanes, different OS possible.
Indeed, as it was said above or in the other thread, Apple could very well have kept a Xeon/Nvidia workstation alongside the other Macs, but I suppose they don't want to be workstation seller like Dell, HP or Lenovo, but rather Vision Pro. Now, if you need a workstation, the Mac Pro 7.1 is quite versatile.
 

Boil

macrumors 68040
Oct 23, 2018
3,477
3,173
Stargate Command
112 Gen5 lanes across seven single-width slots, so x16 each; but if loaded with the almighty RTX 4090, a bunch of slots go unused...?

One might assume that a M3 Ultra Mac Pro would have more PCIe lanes available; we all realize the M2 Ultra Mac Pro (Mac14.8) was rushed out to "complete the transition"...

I, for one, look forward to seeing what the M3 Ultra (and who knows, maybe we do see a configuration that is more than an Ultra) with hardware ray-tracing can do...!

Still hoping for a M3 Extreme Mac Pro Cube though... ;^p
 

avro707

macrumors 68020
Dec 13, 2010
2,263
1,654
My Macbook Pro M2 max is faster than my Mac Pro 7.1, but the Mac Pro 7.1 is not the same kind of machine. The 7.1 is a workstation, slow, but with lots of PCI lanes, different OS possible.
Indeed, as it was said above or in the other thread, Apple could very well have kept a Xeon/Nvidia workstation alongside the other Macs, but I suppose they don't want to be workstation seller like Dell, HP or Lenovo, but rather Vision Pro. Now, if you need a workstation, the Mac Pro 7.1 is quite versatile.

So what is your 7,1? 28 core 2.5ghz, dual 6800 Duos, 1.5TB ram? Or base spec?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.