MS Flight Simulator 2020 update should fix CPU bottleneck and increase perf (mentions DX12)

Published by

Click here to post a comment for MS Flight Simulator 2020 update should fix CPU bottleneck and increase perf (mentions DX12) on our message forum
data/avatar/default/avatar33.webp
yasamoka:

Have you done parallel programming (multi-threading / CUDA / OpenCL)?
I have done no programming at all. However, I am an IT professional and it does not take a genius to know it is not going to remain the same. The fact that things are as they are now is based upon how the hardware was in the beginning but, it clearly will not remain that way. Or more specifically, can you see a quantum mechanics based processor doing one things at a time in serial fashion? 😀
https://forums.guru3d.com/data/avatars/m/220/220214.jpg
ManofGod:

The code is not serial by nature, in my opinion but, it is serial because things were created with single core cpu's in mind. You do not parallelize old code or the old way of doing things but create new things and new ways of doing things, that will not be based upon the old code. And the old code is not so much normal code is it is just based upon the old way of doing things, which, once again, will eventually change. Edit: Call and order bread, clean up your place or prepare your other food well waiting, have bread delivered and eat your food and bread. The old way of doing things is going to die, eventually and there is simply no way around that.
Possibly... but in my experience (when in college learning code) most people were average coders at best and barely competent enough to figure out how to do something in traditional serial code - trying to understand or write parallel code (forked threads) was beyond their comprehension. I mean even recursive code is beyond most codes abilities, including myself, never use that much in real world. Anyway there is entire branched of computer science dedicated to trying to parellelise algorithms - it might be possible to partially split to two or three threads for one part of task but trying to split to 32 threads is very difficult unless it is a majorly SIMD task. Most tasks are not.
ManofGod:

The fact that things are as they are now is based upon how the hardware was in the beginning but, it clearly will not remain that way.
Not really because of hardware... but because it was, and still is, much harder to code and conceptualize and debug with all sorts of new issues and problems with synchronization and stuff like that.