[
http://en.wikiquote.org/wiki/The_Wizard_of_Oz#Dorothy ]
Whether the Mac Mini can be a CUDA monster cannot be divined from mere common sense and it's not an all or nothing proposition.
Antonis,
Here're some of the major points to consider:
1.
We Are All Related, Have Something Of Value, Should Be Open To Opposing Opinions, And Are Constantly Changing
I agree with you one 100% that you are allowed to have and to fully express your opinion(s), regardless of whether it/they turn(s) out to be right or wrong. Not only that, I do value your opinion because I did use, even though clearly you did not use, the word "dumb" and I applied it to me, not to you. I learn, more and more every day, to accept other's opinions as being worthy of full consideration. That's what I gave to your opinion initially and am still doing so even as I compose this response. So I mean this from the very bottom of my heart, "Keep it/them coming, cousin." Exposing your opinions to me only acts as catalyst for me to think deeper, broader and with more focus. So I took and take no offense whatsoever and sincerely say to you, "Thank you, Antonis, for sharing."
Here's where we additionally differ: "... . it is not about CUDA at all. I just used the term "CUDA monster" as an opportunity to point out something else; that such an expensive upgrade is not realistic for such an underpowered mahcine, outside of a lab for experiment's shake. And that's just common sense, I thought it was obvious. ... .
Back to the topic, if you (for some reason) are bothered with the word "CUDA", you may replace it with any other term, it's just the same. This Mac Mini will
never be a "monster" in anything, due to the rest of its h/w limitations, so it'd be a total waste to invest in a pricey upgrade for it."
2.
Relative To What(?) Where Specificity Matters
First off, I caution you to avoid using the word "never, for I'm sure that someone once said, "There'll never be {well why don't you fill it in}", only to be proven wrong (sooner or later). For me, the most important thing that came to mind after reading about MVC's project is CUDA. I cannot replace "CUDA" with any other term and expect that "it's just the same" because it more likely won't be the same. That's why I said "CUDA monster," and not "Monster" in general. I have 24 systems that outperform the Mini. In fact, most of them easily outperform the 2013 MacPro. One's use of a system (i.e., their important applications) dictates whether it's a "Monster" at a specific task. I too consider the Mini to be currently underpowered for that which I do, but that's relative to what I have and can build or could purchase if I didn't build my own systems. Moreover, consider this: Could an "expensive upgrade" to the least expensive Apple system be less expensive, in the end, than buying a more expensive system such as the 2013 MP to implement the same mod, if the performance outcome won't be significantly different. In sum, the most important thing that MVC's new lead has to offer that's pertinent to that which I do, is CUDA 3d rendering. That means a lot to me because that my $$$ maker.
3.
In The Technologically Driven World That We Have Created, Are Creating and Will Create, Common Sense Has It's Limitations; Overly Expansive Exclusions Can Prevent Or Hamper New Discoveries
Let's begin here: "Common sense is a basic ability to perceive, understand, and judge things, which is shared by ("common to") nearly all people, and can be reasonably expected of nearly all people without any need for debate." [
http://en.wikipedia.org/wiki/Common_sense ] That which I now know (and admittedly there's a lot more that I can and will learn) about CUDA I could not derive by the application of just common sense. It requires lots of study and participation in Nvidia CUDA developer programs and learning to apply that gained knowledge to real world needs through application development. Consider these four possibilities
*/ posed by the application of CUDA technology to 3d rendering from four different CUDA 3d application providers:
a) "[O]ctanerender is [a] fully GPU accelerated [3d renderer] and renders images at extreme speeds, up to 50x faster than CPU based unbiased renderers. [
http://render.otoy.com ];
b) "Why use GPU?
Up to 15x -100x faster than conventional CPU ray tracing." [
http://furryball.aaa-studio.eu/aboutFurryBall/index.html ];
c) "Redshift is the worlds first fully GPU-accelerated, biased renderer. Redshift renders scenes many times faster than existing CPU-based renderers.
Save time and money, and unleash your creativity!" [ [
https://www.redshift3d.com/products/redshift ]; and
d) TheaRender is a hybrid renderer that has an engine for CPU(s) only, an engine for GPU(s) only and can do both CPU and GPU rendering simultaneously. Thus, with such options, the jury might appear to be still out regarding whether the Mini might also be a CUDA monster here. But my experiences with TheaRender, Blender, Octane Render, FurryBall Render, Redshift 3d, as well as other GPU rendering options, tell me that the Mini would be a CUDA monster here also.
4) Once The GPU Gets The CUDA Data To Compute, The Speed At Which the GPU Performs It's Functions Is Independent of the Computer's Memory, CPU(s), Buss Speed And Long Term Storage
The CPU's role in the GPU's execution of CUDA functions is non-existent, nada, completely absent, 0. To be sure, the computer's CPU does play a part in creating that information on which the GPU will perform compute functions and the system's buss speed does play a role in how long it takes for that information to get to and from the CUDA card. Moreover, the speed of the long-term storage does play a role in how long it takes to read and write inputs and outputs to and from the CUDA card. But it has been my experience that having the same CUDA card in an 8-core 2007 MP (that's simply my oldest one) which has no hyper threads and placing the same CUDA card in one of my 32 core systems with 64 threads total does not result in any significant difference in the speed at which the CUDA card performs. However and surely, there are a few seconds difference in the time elapsed between when I initiate a render project on the 2007 MP and that render project begins, whereas that time is almost nonexistent on my 32-core systems. But as between my Nehalem, Westmere, Sandy and Ivy Bridge systems there is no perceptible time difference (that range of systems covers 4-core single processor systems to quad CPU 8-core systems). So I would surmise that on a recent dual core Mac Mini, that lack of perceptible time lag would maintain.
In sum, CUDA does those compute chores that can be parallelized and once were previously solely or mainly within the domain of CPU(s) or other specialized co-processors to perform.
You can profit from reading this: "5 Technologies Revolutioni[z]ing Animation In 2014." [
http://www.creativebloq.com/audiovisual/technologies-revolutionising-animation-2014-61411961 ].
*/ There are many more areas/applications, in addition to 3d animation, where CUDA excels (which I respectfully submit to you are not mainly within the purview of just "Common sense'):
a) Computational Finance;
b) Defense and Intelligence;
c) Computer Aided Design;
d) Computational Fluid Dynamics;
e) Computational Structural Mechanics;
f) Electronic Design Automation;
g) Color Correction and Grain Management;
h) Video Editing;
i) Encoding and Digital Distribution;
j) On-air Graphics;
k) On-set, Review and Stereo Tools;
l) Simulations;
m) Weather Graphics;
n) Oil and Gas;
o) Computational Chemistry and Biology;
p) Numerical Analytics; and
q) Physics.
[
http://www.nvidia.com/content/PDF/gpu-accelerated-applications.pdf ].
For all of the above reasons, including especially relatively lower costs, I chose to and still say that the full realization of what MVC is doing can make the MacMini a "CUDA monster."
P.S. Giving credit to the one whose tutoring of me first sent me into the CUDA cave, I shall always be grateful to my comrade in hacking [
http://www.insanelymac.com/forum/topic/282584-top-20-hackintosh-geekbench-scores/ ]: the one, the only, the builder of the most beautiful systems, the spiritual and the generous - PunkNugget.
