Reputation
4,612
Top tag
Next privilege 5,000 Rep.
Approve tag wiki edits
Badges
3 19 22
Newest
 Enthusiast
Impact
~503k people reached

  • 0 posts edited
  • 0 helpful flags
  • 129 votes cast
Jan
6
answered OpenGL, multithreading, and throwing destructors
Jan
6
comment OpenGL, multithreading, and throwing destructors
"The root of the issue is a concurrency issue." No, it isn't. Concurrency is usually not in play here. The current context is a thread-local construct, so it's not really possible to race on it. Either the context is current in the thread where the object is being destroyed or it isn't. The most you could race on would be some other thread destroying it.
Jan
6
comment OpenGL, multithreading, and throwing destructors
"not that I can understand why you'd want to perform multi-threading with an OpenGL context, as the calls needed to do so are expensive enough to negate any benefit you might get out of it from what I understand" There are many reasons for doing so. Constructing objects in the background. Loading textures in the background. Etc. This is usually done by making multiple, shared contexts. That way, each thread has its own graphics context.
Dec
18
awarded  Nice Answer
Oct
5
awarded  Good Answer
Jun
20
awarded  Yearling
Feb
18
awarded  Pundit
Jan
22
awarded  Populist
Jan
22
awarded  Good Answer
Nov
13
awarded  Necromancer
Sep
4
awarded  Good Answer
Jul
31
awarded  Good Answer
Jun
20
awarded  Yearling
Jun
7
awarded  Great Answer
Jul
17
comment Why do game developers prefer Windows?
@Jop: Also, I would like to point out that Valve gets to cheat, because they can basically say to the IHVs, "here's how we're going to render; go make this optimal in your drivers." Other programs don't get to say that. The article you mention even points this out, saying that their work has caused driver changes. They claim that this "benefits all games", but the reality is that it only benefits games that render the way that they do.
Jul
17
comment Why do game developers prefer Windows?
@Jop: ... and? The difference between 270FPS and 303FPS is approximately... 0.4 milliseconds. That's barely more than a rounding error, and in a game that was actually using the hardware (rather than one that throws away 4 out of every 5 frames), it would be an insignificant difference. In short, the performance of D3D nowadays is reasonably comparable to the performance of OpenGL.
Jul
16
comment Why do game developers prefer Windows?
@Jop: Considering how much misinformation is contained in that obvious propaganda piece, I would suggest avoiding that article. How much faster OpenGL's draw calls are than D3D10/11's is very debatable these days; the article cites an NVIDIA PDF from 2006. Saying that Microsoft leaving the ARB is part of a "FUD campaign" is an outright lie. And the rest of the "FUD" stuff is alarmist, anti-Microsoft BS. D3D 9 came to dominate the gaming landscape all on its own, well before the Vista release.
Jun
20
awarded  Yearling
Mar
10
comment Is C++11 Uniform Initialization a replacement for the old style syntax?
"But the syntax using braces doesn't work if type T is an aggregate:" Note that this is a reported defect in the standard, rather than intentional, expected behavior.
Feb
25
awarded  Caucus