1 of 1 people found this helpful
It depend from which point of view, one ask the question, if from a hardware or a software point of view.
From a hardware point of view, i would say it doesn't matter. Nowadays gpu and cpu pack an astonishing transistor count, with astonishing features.
From a low level hardware point of view either AMD, Nvidia, Intel, etc, pack enough power to provide satisfying 3D processing to the software.
From a software point of view, not being a 3D designer, i honestly don't know.
I would assume a professional 3D designer being divided between the Nvidia proprietary API or the AMD open source API.
In this case i would say Nvidia for sure have a bigger grip on the desktop market with it's proprietary API, AMD more focused on consoles API.
Nevertheless i feel the gap between low level hardware languages and higher level of languages becoming a real performance barrier.
Don't get me wrong, low level hardware language is hard to deal with, but if understood give much more performance flexibility.
High level language API are easier to deal with, but also add increasing performance overhead over time.
On the other hand even if low level language is hard to manage, it doesn't produce performance overhead, hence no need to build complex API.
Yes I understand, AMD and Nvidia have their respective advantages
Obviously add to that the general economic market rules, as any other fields.
Who is bigger can often push more aggressively it's own IP technology, hence pushing the field toward a direction instead another more easily.
Especially if one also provide extended support to 3D designers making their job, life easier.
However imo, i would prefer that 3D developers would come back to the hardware computation basics, instead going forward only using high level API in future.
As said i feel the gap between low level hardware and high level of software growing exponentially, to a point that only a few know really the process as a whole.
Hence i would rather push a reciprocal teaching approach between high end hardware and software department, rather than segmenting the two fields.
Slower approach for sure but imo give more and better result over time.