I have been writing a lot of code in OpenGL in the past few years, and I've made sure that the code builds with no errors on Windows as well (coming mainly from Linux and Mac). I want this code to run efficiently on Windows machines, and get re-aquainted with DirectX again though.
Would it be more efficient for the Windows build of my code to use DirectX and Windows API over OpenGL/OpenAL/BSD Sockets/pthreads, etc for graphics, sounds, networking, multithreading etc?
I suggest changing the topic name because current one looks like asking for a meaningless war 
Rewriting your whole code from openGL to DirectX would be time consuming, I would advise you to only change the strictly operating system bound functions (i think it's obvious one, and I can see you already achieved that). But if your aim is to learn (or revise) directx, it might be a good idea to do it ;)
Would it be more efficient for the Windows build of my code to use DirectX and Windows API over OpenGL/OpenAL/BSD Sockets/pthreads, etc for graphics, sounds, networking, multithreading etc?
Nope. Why? Because the only part of DirectX that's really meaningful these days is D3D. There's no reason to change your OpenAL/socket/pthread, etc. code to use the old and (hopefully) dying parts of DirectX. Unless you need some kind of platform specific functionality, there's nothing to be gained here.
Now, as for whether you should use OpenGL or D3D... It's up to you. D3D won't be magically more efficient if you just do a straight port.
I suggest changing the topic name because current one looks like asking for a meaningless war
100% agree with this!
I suggest changing the topic name because current one looks like asking for a meaningless war.
I second this suggestion - I was seriously considering reporting this thread for mod action before I opened it.
Regarding your question, if you really want to learn DirectX, then porting it may be a good way of doing so, but be aware that - with the exception of D3D - almost all of DirectX is gone nowadays. You could port just the renderer, but if you've got it set up to work nicely with some of GL's quirks (bind-to-modify, I'm looking at you, and don't feel safe shader-management because you're next) you may have a fairly torrid time. Maybe working through some tutorial material might be a better idea, until you establish a better level of comfort with the API differences.
It is a dark path you're walking on OP with such thread title.
OpenGL vs DirectX wars were a long time ago but many still have scars from them.
I echo the responses that have been provided so far. The only real difference between OpenGL and Direct3D would be related to the driver support. Check your expected GPU vendor support, and make sure they have well supported drivers for the API of your choosing. If they do, then don't switch. If they don't, then do switch.
If you want to learn D3D, there is no reason to re-implement your previous projects - just do your next one in D3D to freshen up your knowledge.
I was concerned with whether this title would cause some sort of flame war lol. I was actually starting to learn programming and DirectX when that was dying down (I think). Can anyone explain on how to change the title?
I'm aware of what they both do, and I've got experience with both, I'm just wondering if there's performance issues when using OpenGL in a Windows environment. I've read in different things online regarding API performance since 2010 concerning DirectX and OpenAL/GL/CL libraries. For example, I've heard that Windows drivers for graphics cards seem to support D3D's max capabilities much better than OpenGL in terms of things like shader processing performance, max texture units/texture size/lights/stacks/etc. There was an article mentioning OpenGL dying because of lack of support, which I don't believe is true, however, it's not the first I've heard of it.
To generalize... on Windows D3D is more stable/reliable because it's largely implemented by one entity (Microsoft) with the rest implemented by the driver (nVidia/AMD/Intel/etc) according to a strict D3D driver specification from Microsoft.
On the other hand, GL is almost entierly implemented by the driver, and Khronos do not test implementations for compliance with the specification.
To be fair, you do find driver bugs occasionally in both D3D and GL, but in my experience GL has the worse reputation in this regard.
With features/extensions, sometimes one API provides more capabilities than the other, and it goes both ways. Intel in particular have often lagged behind with their GL driver support, with their D3D driver providing more capabilities than their GL driver... but there are also cases of the opposite.
In the D3D9 vs GL days, GL's CPU-side overheads tended to be slightly smaller... however, D3D10/11 are much more efficient than 9 in this regard, so I'd guess that they've closed this gap with GL, and probably even beaten it now, seeing as GL is much more complex due to backwards compatibility whereas D3D makes a clean break with each version, discarding old cruft. In any case, these performance differences should be very small (e.g. a millisecond per frame...).
There is one annoying case with GL that can cause poor performance -- many drivers when asked to perform a function that is not supported in hardware, will resort to CPU-emulation of that feature. For example, I added array indexing to a pixel shader once, but my GPU only supported this feature in vertex shaders, so my driver decided that instead of failing to draw anything, it would instead run my pixel shader in software emulation on the CPU, at 1 frame per second... GL has a lot more of these kinds of fast-path/slow-path pitfalls compared to D3D.
Well written GL is probably slower than well written D3D 11 code. (Source) But there are a lot of intersecting factors at play, and optimizing your GL code might yield bigger speedups than a D3D rewrite in the same amount of time. Stuff like using your buffers properly, batching, vertex arrays, uniform buffers, texture formats, avoiding slow path GL stuff, etc. There are also driver headaches, but no sense worrying about those unless Intel is really big for you and in that case testing is the correct approach.
