Godot supports many different compilers and for production releases we
have to support 3 currently: GCC8, Clang6, and MSVC2017. These compilers
all do slightly different things with -ffast-math and it is causing
issues now. See #24841, #24540, #10758, #10070. And probably other
complaints about physics differences between release and release_debug
builds.
I've done some performance comparisons on Linux x86_64. All tests are
ran 20 times.
Bunnymark: (higher is better)
(bunnies) min max stdev average
fast-math 7332 7597 71 7432
this pr 7379 7779 108 7621 (102%)
FPBench (gdscript port http://fpbench.org/) (lower is better)
(ms)
fast-math 15441 16127 192 15764
this pr 15671 16855 326 16001 (99%)
Float_add (adding floats in a tight loop) (lower is better)
(sec)
fast-math 5.49 5.78 0.07 5.65
this pr 5.65 5.90 0.06 5.76 (98%)
Float_div (dividing floats in a tight loop) (lower is better)
(sec)
fast-math 11.70 12.36 0.18 11.99
this pr 11.92 12.32 0.12 12.12 (99%)
Float_mul (multiplying floats in a tight loop) (lower is better)
(sec)
fast-math 11.72 12.17 0.12 11.93
this pr 12.01 12.62 0.17 12.26 (97%)
I have also looked at FPS numbers for tps-demo, 3d platformer, 2d
platformer, and sponza and could not find any measurable difference.
I believe that given the issues and oft-reported (physics) glitches on
release builds I believe that the couple of percent of tight-loop
floating point performance regression is well worth it.
This fixes#24540 and fixes#24841
There are still some left in the Android Java code, even stuff to swap between
GLES1 and GLES2 support from early Godot days... would be good to see some cleanup
there too one day.
The "graphics/api" option for Android exports is removed, as only GLES 3.0 is supported.
It can be readded when GLES 2.0 support comes back. Fixes#13004.
The changes include work done to ensure that GDNative apps and Nim
integration specifically can run on Android. The changes have been
tested on our WIP game, which uses godot-nim and depends on several
third-party .so libs, and Platformer demo to ensure nothing got broken.
- .so libraries are exported to lib/ folder in .apk, instead of assets/,
because that's where Android expects them to be and it resolves the
library name into "lib/<ABI>/<name>", where <ABI> is the ABI matching
the current device. So we establish the convention that Android .so
files in the project must be located in the folder corresponding to
the ABI they were compiled for.
- Godot callbacks (event handlers) are now called from the same thread
from which Main::iteration is called. It is also what Godot now
considers to be the main thread, because Main::setup is also called
from there. This makes threading on Android more consistent with
other platforms, making the code that depends on Thread::get_main_id
more portable (GDNative has such code).
- Sizes of GDNative API types have been fixed to work on 32-bit
platforms.
- The Windows, UWP, Android (on Windows) and Linux builds are
tested with Scons 3.0 alpha using Python 3.
- OSX and iOS should hopefully work but are not tested since
I don't have a Mac.
- Builds using SCons 2.5 and Python 2 should not be impacted.
Fixes GDNative build error on Android.
It's also discouraged by Google to rely on it. In case someone needs to check, use ``__ANDROID__`` instead, provided by the very same compiler.
This commit fixes errors occurring during Android export template
builds.
This required modification in third-party library (libpng) to compile
with NEON. Most likely a similar patch will be applied by them and we
could then replace the modified version.
Tried to organize the configure(env) calls in sections, using the same order
for all platforms whenever possible.
Apart from cosmetic changes, the following issues were fixed:
- Android: cleanup linkage, remove GLESv1_CM and GLESv2
- iPhone: Remove obsolete "ios_gles22_override" option
- OSX:
* Fix bits detection (default to 64) and remove obsolete "force_64_bits" option
(closes#9449)
* Make "fat" bits argument explicit
- Server: sync with X11
- Windows: clean up old DirectX 9 stuff
- X11:
* Do not require system OpenSSL for building (closes#9443)
* Fix typo'ed use_leak_sanitizer option
* Fix .llvm suffix overriding custom extra_suffix
All the warnings are factored out of the platform-specific files and moved to
SConstruct. Will have to check that it does not introduce regressions on some
platforms/compilers.
(cherry picked from commit 31107daa1a)
Fix wrong path for 32-bit Windows, which fixes#7084
Exclude 32-bit Windows from multi-threaded linking because it's not supported by the NDK
Remove 32-bit Linux as there is no NDK variant for it
Done with `autopep8 --select=E7`, fixes:
- E701 - Put colon-separated compound statement on separate lines.
- E702 - Put semicolon-separated compound statement on separate lines.
- E703 - Put semicolon-separated compound statement on separate lines.
- E711 - Fix comparison with None.
- E712 - Fix (trivial case of) comparison with boolean.
- E713 - Fix (trivial case of) non-membership check.
- E721 - Fix various deprecated code (via lib2to3).
Took the opportunity to undo the Godot changed made to the
opus source. The opus module should eventually be built in its
own environment to avoid polluting others with too many include
dirs and defines.
TODO: Fix the platform/ stuff for opus.
Rémi Verschelde
Hein-Pieter van Braam
lupoDharkael
Wojciech Milkowski
Ridwan Abdul Hafidh
Tiago Quendera
Juan Linietsky
Colin Kinloch
Rhody Lugo
AndreaCatania
Elliott Sales de Andrade
Ruslan Mustakov
Matthias Hoelzl
Pedro J. Estébanez
Pedro J. Estébanez
Randy Tan Shaoxian
Rémi Verschelde
Błażej Szczygieł
Anthony Fieroni