Sweet Spot for PT w/ 2020 Mac Pro?

[rlhinc]

Would love some comments about where the economic 'sweet spot' lies with the new (expensive!) Mac Pro for use with PT Ultimate.

My sessions can be 50-60 tracks, lots of plugs, outboard analog, etc. With my current 2010 chessegrater, I ALWAYS have to deactivate all plugs, unused tracks, etc. to achieve low enough latency for tracking, and would LOVE never to have to do that again!

Not sure how efficiently PT uses multiple cores, etc. So, where might the 'sweet spot' be in terms of how to 'spec' a new Mac Pro?

Many thanks in advance...

Robert

[uptheoctave]

I have the new Mac Pro.

I believe I hit the sweet spot for me, which is 3.2ghz 16 Core with base memory and a 4TB SSD.
I have Pro Tools installed- it seems to be working even though it isn't supported but I haven't tried HDX yet.

Logic is supported and I am getting crazy amounts of processing.
The Logic benchmark test gives me 309 tracks with the new Mac Pro.
My previous machine would "only" give me about 100 tracks.

To go up to the next CPU would be an additional £3600.
I cannot justify that.

I've bought 128GB RAM from OWC that turns up in a few days time.

I've decided though that I will return this unit and get the rack when it is released.
It is just a better fit for me.

[TNM]

Your machine under load is at 2.9ghz which is not only below base clock, but way below the chip's rated all core turbo. I would not recommend the 16 core to anyone right now. So is this topic going to be moved to general right now also, like mine was within seconds, which was also talking about the new mac pro?

[TNM]

BTW to the OP, i re examined the 12 core video and it seems to be above base clock at all times but still not at the proper all core turbo (it can do around 3.6 under load). That's the mac pro to get right now for audio as single core performance also matters for us. So.. the 12 core. If you want the 8 core get the imac pro and save money. If the 16 and 28 cores have a power limit issue and can't use the full cpu potential, I doubt the 24 core can either.

[RobertDorn]

When I have to believe the benchmark MAC chart, the 28-core Xeon will be best single-core and multi-core processor. It tops out my 10-core Xeon iMac Pro on single core speeds. Which already is quite a beast for Pro Tools. And then x 28 cores. Sounds like a charm :)


However, it's a little weird because both the 16 core 3.2HGz and 24 core 2.7HGz are a little faster than the 12-core 3,3GHz on single core speeds. And the 28-core with lowest clock-speed per core tops out all other 3 and sits more in the range of the iMac 2019, which excels in single core speeds because of the highest clock-speed per core.

Not sure if there's a firmware update or something needed because some checks in the field show that the CPU's are not running on their base-clock speed. Otherwise, if these score's are correct, the 28-core is the CPU to chose for best single core and multicore performances

https://browser.geekbench.com/mac-benchmarks/

[johnnyv]

I just ordered the first of my pair of Mac Pro's, and for me the sweet spot for price/performance is:

16 Core
32GB RAM (less expensive to upgrade from third parties)
Base graphics card (for music and light post work, it should be fine and can easily be upgraded at a later time)
2TB SSD (for me it's better to buy the OWC accelerator card and load it up with the SSD's I already own)

The Geekbench 5 benchmarks place the 16-core as the third fastest multi-core Mac ever tested (bested only by the 24 and 28-core 2019 Mac Pro's), so it should be a solid performer. I do however find it curious that on single core performance, the iMac Pro Intel i7 processor running at 3.2ghz bests the 2019 16-core's Xeon running at the same speed. (It's only benching at 3% faster, but still...). If this is some sort of a power limiting issue as TNM is suggesting, I'm confident that Apple will address it. And regardless, I still think the 16-core is the sweet spot for my budget and performance needs, even as the benchmarks stand today.

Overall, I'm crazy excited, as my 2012 12-cores were showing their age, stuck at Sierra and being without Thunderbolt, which seems to be a standard with some longevity.

Arrives in about two weeks.

[rlhinc]

I'm the OP here, and just wanted to thank all of you for the very excellent and valuable info. Not sure what to make of what seems the general consensus that some of Mac Pro processors aren't actually running at their advertised speeds. Everyone seems to feel this will be corrected in a firmware update. Is this normal for new Apple machines? Seems very weird to me, thought I admittedly know nothing about that world. Anyway, thanks! Will probably go 16-core and - hopefully - never look back!

[Darryl Ramm]

Originally Posted by rlhinc

I'm the OP here, and just wanted to thank all of you for the very excellent and valuable info. Not sure what to make of what seems the general consensus that some of Mac Pro processors aren't actually running at their advertised speeds. Everyone seems to feel this will be corrected in a firmware update. Is this normal for new Apple machines? Seems very weird to me, thought I admittedly know nothing about that world. Anyway, thanks! Will probably go 16-core and - hopefully - never look back!

Apple had a very public problem with the 2018 Macbook Pro with broken CPU thermal throttling. Fixed/problem reduced with software updates (and I'm typing this post on a later version of a MacBook Pro i9 system that works great for me under load), see https://www.macrumors.com/2018/07/24...o-improvements or just google for it. Here is the original report of issues for that https://www.youtube.com/watch?time_c...ature=emb_logo.

I would hope that Apple is super careful about thermal design and throttling issues on any new system, but the Mac Pro especially. The unified cooling and overall design approach of the Mac Pro means that Apple must have spend a ton of time doing thermal modeling of that system.

Could there be a general underlying mistake of trying to optimize for cooling quietness not performance, maybe but I'd hope not. Could there be more of an implementation bug, similar to the 2018 MacBook Pro, maybe more likely? Could Intel have screwed up something, maybe. Could there be systems that have specific problems (faulty CPUs/thermal sensors/fan controllers, incorrect heatsink connections/thermal bonding, etc.), sure.

[RobertDorn]

I'm eagerly awaiting real world tests with the 2019 Mac Pro. The benchmark results so far don't seem to lie. The link I posted are averages, I searched for individual MacPro7,1 results and the 28 core tops out with 1207 on single-core speed, none of the other W series XEON processors -with higher clockspeed-per-core counts- get above 1200. And it's common known music production relies heavily on single-core speed performance. It's a whole different story with video and image rendering which is not real-time and benefits hugely from multi-core performance.

If everything is officially supported with Catalina, not only Pro Tools itself but also all major third party plugin vendors, I'd love to see how the 12 - 16 - 24 and 28 core versions compare with heavily loaded Pro Tools sessions

[rlhinc]

"...If everything is officially supported with Catalina, not only Pro Tools itself but also all major third party plugin vendors, I'd love to see how the 12 - 16 - 24 and 28 core versions compare with heavily loaded Pro Tools sessions[/QUOTE]..."

Yeah, you and me both.