Apple announces Mac studio desktop equipped with M1 Ultra Chip & Studio Display Monitor

[saltycaramel]

Interesting tidbit that explains why the M1 Ultra Studio is 1kg/2lbs heavier than the M1 Max Studio. 

 

They have the same 370W power supply. The additional weight is due to M1 Ultra having a larger copper thermal module, whereas M1 Max has an aluminum heatsink.

[HomeBoi]

20 hours ago, KWelz said:

It was shown powering multiple displays.  

multiple monitors, sadge

[GhostRoadieBL]

I should not be this mad about m1 ultra but, for me, it's a bad day when Apple pantses intel nvidia and amd with what appears to be a price competitive option for professionals.

It's totally irrational but Apple is supposed to be overpriced vs performance but when prices are this silly, it's a tough choice now even with inflation.

 

Performance per watt and performance per $ aren't supposed to be in Apple's favour.

[Video Beagle]

19 hours ago, Paul Thexton said:

I can understand the desire/requirement for that much screen real estate, but I recently tried vertical stacking of monitors so I could use a single desk setup for my work + personal hardware and the results were … not great. Maybe there’s something about the higher monitor tilt angle I’ve not gotten quite right, anybody aware of any good information about getting that kind of setup right?

Well assuming you're talking about positioning, rather than the mechanics of it (which is "get stands/arms that support it") it really comes down to fiddiling around til you find what you like.  I have a verticle stack dual monitor on my PC (i have 3 monitors for my mac.. well it's an imac, so to additional on left and right)... I haven't really worked out the best way to use the upper monitor....i mainly set it up because I had the monitor arm, and I had an extra monitor, so I COULD do it, and figured the overkill of it made me happy.

[Deadpool2onBlu-Ray]

19 hours ago, Obioban said:

My guess is...

 

3) Mac Pro gets an M2 based chip. The M2 has additional mesh points, that allow them to do 2-4X M1 Ultra (so 40-80 cpu cores and 128-256 GPU cores), plus whatever generation over generation improvement they get going from M1 to M2 (which is likely 10-25%, assuming the M2 is A15 based, since that was the A14 (which the M1 is based on) to A15 improvement.

 

... unless the M2 is A16 based, since the M2 is still months out and the A16 is (probably) only 6 months away, in which case maybe a little more?

I am guessing this is more likely. The A15 from A14 wasn't that big of change. A16 will likely be more significant. 

[Deadpool2onBlu-Ray]

11 hours ago, GhostRoadieBL said:

I should not be this mad about m1 ultra but, for me, it's a bad day when Apple pantses intel nvidia and amd with what appears to be a price competitive option for professionals.

It's totally irrational but Apple is supposed to be overpriced vs performance but when prices are this silly, it's a tough choice now even with inflation.

 

Performance per watt and performance per $ aren't supposed to be in Apple's favour.

I wonder if this will affect Nvidias 4090 pricing. Doubtful, I know, but still. 

[GhostRoadieBL]

3 hours ago, Ryan829 said:

I wonder if this will affect Nvidias 4090 pricing. Doubtful, I know, but still. 

If the rumors are true (huge doubt) that 4090 is 2x 3090 performance and a ~600w tdp it won't change anything.

It's likely that the rumor of 30 and 40 series being manufactured and sold at the same time might cause a small tick down for 30 series and could line back up with M1 Ultra vs 3080 pricing for gpu only tests.

 

[Dracarris]

On 3/9/2022 at 6:20 PM, WereCat said:

For example, A6000 will definitely demolish the M1 Ultra in rendering performance maybe even by an order of magnitude but only as long as the scene fits into its 48GB VRAM buffer. Once the scene becomes much larger the A6000 simply has no chance. 

also consider how much that A6000 48GB costs..

[MrSmiles]

Is all this restricted to certain uses eg. Particular programs?

Just saying, it is Apple.

[Obioban]

23 minutes ago, MrSmiles said:

Is all this restricted to certain uses eg. Particular programs?

Just saying, it is Apple.

The Mac is extremely open, software wise. Almost no restrictions, if you turn off the safeties.

 

That said, programs will need to be compiled to take full advantage-- aka not X86 code, GPU code running through Metal, ML tasks running on the ML cores. This happens automatically if you use XCode, but if you're using, say, Electron... not going to be nearly as optimized.

[MrSmiles]

Hmmm. I guess I’m a skeptic still. 

[WereCat]

1 hour ago, Dracarris said:

also consider how much that A6000 48GB costs..

I literally did in the next paragraph 

[Kisai]

48 minutes ago, Obioban said:

The Mac is extremely open, software wise. Almost no restrictions, if you turn off the safeties.

 

That said, programs will need to be compiled to take full advantage-- aka not X86 code, GPU code running through Metal, ML tasks running on the ML cores. This happens automatically if you use XCode, but if you're using, say, Electron... not going to be nearly as optimized.

There is a Tensorflow plugin

https://developer.apple.com/metal/tensorflow-plugin/

There however is not a pytorch module.

 

https://coremltools.readme.io/docs

However there is a suggestion to convert pytorch and others models to CoreML. Which is of no help if the software is written to use pytorch.

 

[hishnash]

On 3/11/2022 at 4:36 AM, Ryan829 said:

I am guessing this is more likely. The A15 from A14 wasn't that big of change. A16 will likely be more significant. 

There was quite a big change from A14 to A15 apple however mostly put this change into reducing power draw as that was more important on the phones than increasing perf. Perf increase was about 10% but at the same time power draw was reduced by over 30% in many tasks!  This is in fact a much larger change than apple typically do year on year, the entire story about A15 not bieng a big gains of A14 comes from the fact Apple did not compare them when they shipped the new phones with A15 but people who have looked at them since have been very impressed with the improvements.