We are currently in the process of converting the website to the new design. Some pages, like this one, are still broken. We appreciate your patience.
Jari Komppa
Also, some compilers actually optimize zero-setting for-loops into memset calls. Googling about…
»
Livet Ersomen Strøm
d7samurai If you're not using SIMD instructions (https://software.intel.com/sites/landingpage/I…
»
Andrew Bromage
Good point. I didn't think about small blocks.
»
Livet Ersomen Strøm
cmuratori In general, I don't like the tone of your posts, so I will not be reading them in the …
»
David Owens II
Yeah, my input is 440bytes per frame. I haven't looked to see how big Casey's version is. But eve…
»
Casey Muratori
Kladdehelvete But have you actually meassured the difference of "bad alignment" on real code? Lik…
»
Casey Muratori
The file is large because it is actually saving all the memory that is allocated, not just the 64…
»
Rowan Crawford
It would be helpful to have more information. Depending on how you are handling events you might …
»
mojobojo
cmuratori Err, that's what the codebase already does? Not sure what you guys are saying here... …
»
d7samurai
Kladdehelvete Code is so fast, that it will only be limited by how often there is data available …
»
David Owens II
The point is less memory reads per instruction.
»
Livet Ersomen Strøm
cmuratori But assuming you are writing your code to use full-width loads (SSE, AVX, or AVX 2), t…
»
David Owens II
Maybe the OP is not calling your Win32RecordInput(&Win32State, NewInput) in the game loop? @…
»
Livet Ersomen Strøm
d7samurai I was just about to mention something to Kladdehelvete about SIMD, but I see Casey took…
»
Livet Ersomen Strøm
cmuratori Kladdehelvete I have tested jumpaligning, codealigning and data alignment on my entire …
»
Casey Muratori
Err, that's what the codebase already does? Not sure what you guys are saying here... when you s…
»
Casey Muratori
Yeah, just to follow up on this further: it's a bad idea to do something like "I tested to see if…
»
Casey Muratori
Actually, the answer is usually "it depends" :) Sometimes the system memset is actually slower b…
»
d7samurai
I was just about to mention something to Kladdehelvete about SIMD, but I see Casey took care of it.
»
Casey Muratori
Kladdehelvete I have tested jumpaligning, codealigning and data alignment on my entire codebase, …
»
Livet Ersomen Strøm
rathersleepy @Kladdehelvete On intel there may be no real world consequences to ignoring alignme…
»
Livet Ersomen Strøm
5sw It doesn't matter what you perceive the speed to be, you can only trust actual measured dat…
»
Patrick Lahey
It would be helpful to have more information. Depending on how you are handling events you might…
»
Patrick Lahey
Clearly I misunderstood what you were trying to say. My mistake.
»
Andrew Bromage
Kladdehelvete Yeah. But who decide if it looks right? The artist, usually. Certainly someone who …
»
Andrew Bromage
It would buy us a lot of performance. System memset() functions are far better tuned to take adva…
»
David Owens II
That's what I'm doing. I have a snapshot.mem file that contains the initial state, then an input.…
»
d7samurai
rathersleepy Not quite. The compiler automatically pads out any struct that contains something re…
»
Patrick Lahey
@Kladdehelvete On intel there may be no real world consequences to ignoring alignment. On other…
»
Patrick Lahey
@d7samurai You imagined if we perhaps had a Foo like this (with no padding), right? struct Foo …
»