Will MAC bring the tower back?

[William Ciani]

I really want to make a sound investment into a better workstation to handle the 6k workflow. I switched to mac years ago after around the time of the infamous windows 2000 OS disaster and I haven't looked back. But with all the TB2 limitations with 3rd party hardware its looking like windows may actually be a better machine to work with this type of content. Apple markets the "trash can" as the ultimate creativity machine for professionals but I feel like its hardly that. In many ways I believe its a failure on that front. Im curious if anyone thinks apple will respond to the professional community and revise the chassis to accommodate important graphics cards that are currently hindered by TB or perhaps be solving the issue soon with TB3. (In any case the current machine would likely not be able to accept the upgraded TB technology anyway so investing in the current mac is likely a bad decision). Looking for others thoughts on the next mac release (if any).

 

[Joachim Hoge]

One word, no

That said, I have not upgraded to nMP. They just got too expensive compared to when I bought my 3.1 MP in 2008.
I might next year

 

[Antony Newman]

If the next Mac had 2 x the CPU grunt, 3 x the GPU grunt, 4 x Memory GB, 6 x TB3 sockets, 4 x USB 3.1, 3 x DP 1.3 (which can host HDMI 2.0) connections ... would this be enough? If not .. what do we actually need?

If you split your needs into 'real time' (editing / experimenting / monitoring) and 'batch' (ie press a button an come back later) .. I guess the question is how much Faster does your real time workflow need to be.

The supercomputing folks are pushing for the GPUs are doing 90% of the work to scale efficiently (performance / watt).. and get the code to 'bypass' the CPU (like in OpenCL 2.0).
I suspect that as more our industry's code is rewritten to multi task efficiently, and leverage GPU wherever possible ... 6K will melt like butter on the next Mac.

2017 is the year that IBM and Nvidia are betting on a 5 x increase in performance per watt. Here's hoping that a 2016 Mac pro will be 3 x faster than their 10xcore 2013.

AJ

 

[L. Langer]

I don't think Apple will bring the tower back. It's clear they want to sell compact systems that rely on GPGPU - via OpenCL - for processor-intensive work and Thunderbolt for peripheral connectivity. I think that the Mac Pro two generations from now will be the system that we all wish today's Mac Pro was because it may incorporate stacked memory and a lot of other advances that could bring 2-3x the raw computing power of today's MacPro while bringing connectivity for displays that are 5K-8K.

 

[Jeff Kilgroe]

Broadwell core Xeons should ship in early to mid 2016 and with the process shrink to 15nm, there will be significant power savings. Desktop class should be here in about 6 months, but no one knows for sure... Desktop Broadwell has been delayed several times and I sometimes wonder if Intel even has a serious ETA right now. 2017~2018 is probably realistic for the massive increase in power per watt as that will bring the following iteration of the 15nm process that provides most of the significant optimizations. As we push to 11nm around 2020~2022, that's when slim notebooks/ 2-in-1's and tablets should start to rival the desktops we have now.

GPUs are on their own path of sorts. Both AMD and nVidia are pushing for 20nm fab processes to launch in mid 2015. Apple, Samsung and others are driving the evolution of ARM and other lightweight CPU platforms ahead at a furious pace. Intel has also jumped into the ARM game and I'm expecting we'll see hybrid ARM/x86 offerings at some point in the not too distant future.

I'm still hoping that AMD wakes up and realizes that they could equip their top GPUs with x86 CPU cores to build all-in-one modules for scalar supercomputing. Kinda odd it hasn't happened yet because they make both products... Surprised they have not combined them. Nor gone the other way and created notebook/ mobile class CPUs with superior integrated GPUs to compete against Intel.

 

[Jeff Kilgroe]

I don't think Apple will bring the tower back. It's clear they want to sell compact systems that rely on GPGPU - via OpenCL - for processor-intensive work and Thunderbolt for peripheral connectivity. I think that the Mac Pro two generations from now will be the system that we all wish today's Mac Pro was because it may incorporate stacked memory and a lot of other advances that could bring 2-3x the raw computing power of today's MacPro while bringing connectivity for displays that are 5K-8K.

Agree. And as we often see with Apple products when they first arrive, they usually have the vision right or have a great concept, only the execution isn't matured. It just takes a few generations to realize the potential. The Mac Pro cylinder is an awesome system and often doesn't get the credit it deserves. OTOH, it doesn't fit well within established hardware environments and is hampered by its own innovations or the limitations of the current connectivity options.

IMO, Apple should have shipped it with 10GE standard, an nVidia GPU option and perhaps also options for 4 and 6 core i7 CPUs to make the system more attractive to a wider range of customers. That said, the nMP has still sold in excess of what Apple had originally anticipated and it is proving to be a very capable system in many ways. It is a very nice and compact workstation any way you look at it. But it doesn't address workflows that need multiple CPUs and/ or peripheral expansion.

 

[Meryem Ersoz]

http://devilanddemon.com/

I wish these had been a hair earlier to market. I would not have bought a new trashcan.