Commit Graph

3884 Commits

Author SHA1 Message Date
Hajime Hoshi
8cf3c31cf6 internal/restorable: reuse bytes instead of allocations
Updates #2375
2022-10-06 12:41:47 +09:00
Hajime Hoshi
69cdd2b178 internal/graphicsdriver/metal: refactoring 2022-10-04 00:29:18 +09:00
Hajime Hoshi
b489548963 internal/builtinshader: fix comments 2022-10-03 01:09:28 +09:00
Hajime Hoshi
61cabbf2e7 internal/atlas: optimization 2022-10-03 00:06:50 +09:00
Hajime Hoshi
72983d966b all: remove unused conditions
Now Kage shaders are always used.

The situtation is different from when we fixed for #1355, so we removed
the fast path for #1335. We have to re-check the current performance.

Updates #1355
2022-10-02 23:50:48 +09:00
Hajime Hoshi
534d82c17d internal/builtinshader: move Filter and Address from internal/graphicsdriver 2022-10-02 23:24:15 +09:00
Hajime Hoshi
9c07b20f2b internal/ui: remove an odd cast 2022-10-02 23:20:16 +09:00
Hajime Hoshi
efd91c8b86 internal/ui: remove unused arguments from DrawTriangles
Closes #2369
2022-10-02 23:10:27 +09:00
Hajime Hoshi
d823e22bed internal/mipmap: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 23:06:32 +09:00
Hajime Hoshi
361a1059ee internal/buffered: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 23:06:29 +09:00
Hajime Hoshi
3da30b3b78 internal/atlas: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 23:06:27 +09:00
Hajime Hoshi
de8f85651d internal/restorable: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 22:51:34 +09:00
Hajime Hoshi
0ae2b1bc24 all: remove unnecessary conditions
Now a Kage shader is always used.
2022-10-02 22:30:59 +09:00
Hajime Hoshi
6fca8edc1b internal/graphicscommand: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 22:21:11 +09:00
Hajime Hoshi
080d936d42 internal/graphics: remove unused arguments from DrawTriangles
Updates #2369
2022-10-02 22:13:31 +09:00
Hajime Hoshi
6fc9d9316d internal/graphicscommand: remove unnecessary conditions
Now a Kage shader is always used.
2022-10-02 22:11:20 +09:00
Hajime Hoshi
a9574627e8 internal/graphicsdriver/directx: remove the built-in shaders
Updates #2369
2022-10-02 21:35:23 +09:00
Hajime Hoshi
033997d928 internal/graphicsdriver/opengl: remove the built-in shaders
Updates #2369
2022-10-02 21:26:13 +09:00
Hajime Hoshi
5f7db485f2 internal/graphicsdriver/metal: remove the built-in shaders
Updates #2369
2022-10-02 20:34:38 +09:00
Hajime Hoshi
0b45ca7057 ebiten: reduce duplicated shader compilations 2022-10-02 20:33:50 +09:00
Hajime Hoshi
ff72898b85 internal/builtinshader: use a struct instead of a map
With a map, we might miss typos.
2022-10-02 20:17:47 +09:00
Hajime Hoshi
311aa7dcf8 all: use Kage shaders in internal packages
Updates #2369
2022-10-02 19:17:09 +09:00
Hajime Hoshi
f293a03ab6 internal/mipmap: reland: do not consider filter to determine mipmap level
This is a reland of 42833614fb.

Now the filter argument is not used (a4e9a05b14),
the filter value can be the nearest filter even though the shader is
for the linear filter.

As long as `canSkipMipmap` is set correctly, we don't have to consider
the parameter `filter`. We should ignore it.

Updates #2364
2022-10-02 19:14:23 +09:00
Hajime Hoshi
bcb9844395 Revert "internal/mipmap: bug fix: do not consider filter to determine mipmap level"
This reverts commit 42833614fb.

Reason: test failure: TestImageEdge
2022-10-02 18:53:28 +09:00
Hajime Hoshi
42833614fb internal/mipmap: bug fix: do not consider filter to determine mipmap level
Now the filter argument is not used (a4e9a05b14),
the filter value can be the nearest filter even though the shader is
for the linear filter.

As long as `canSkipMipmap` is set correctly, we don't have to consider
the parameter `filter`. We should ignore it.

Updates #2364
2022-10-02 18:44:06 +09:00
Hajime Hoshi
4b2a9c3243 internal/testing: add the default exit code 1
This confirms that Update is really called.
2022-10-02 18:16:16 +09:00
Hajime Hoshi
3bb21282af internal/builtinshader: bug fix: typo
Updates #2364
2022-10-02 16:16:49 +09:00
Hajime Hoshi
eea11ba6cb internal/ui: use a Kage shader for Fill
Updates #2364
2022-10-02 16:08:33 +09:00
Hajime Hoshi
7b6f726729 internal/ui: skip uniform variables when possible
Updates #2364
2022-10-02 16:02:44 +09:00
Hajime Hoshi
239f9de2ca internal/ui: use Kage shaders
Updates #2364
2022-10-02 15:54:48 +09:00
Hajime Hoshi
a4e9a05b14 ebiten: reland: use Kage shaders instead of built-in shaders for a color matrix
This replaces the built-in shaders with Kage shadres. This is a
refactoring and doesn't degrade performance:

```
go test -bench=^BenchmarkColorMScale$ -run=^$ . -count=5
```

```
name           old time/op  new time/op  delta
ColorMScale-8   978ns ±15%  1184ns ±46%   ~     (p=0.413 n=4+5)
```

A follow-up change to remove the built-in shaders is needed.

Closes #2364
2022-10-02 15:35:40 +09:00
Hajime Hoshi
5e459bbe42 internal/graphicsdriver/opengl, metal, directx: use premultiplied alpha format for color scales
Updates #2365
2022-10-02 13:38:21 +09:00
Hajime Hoshi
8a7d860632 all: unify Windows error handlings for Windows HANDLE
Updates #2366
2022-10-02 02:22:06 +09:00
Hajime Hoshi
a9ba0db3d1 internal/glfwwin, internal/graphicsdriver/directx: bug fix: wrong error handlings for Windows HANDLE
Closes #2366
2022-10-02 01:57:24 +09:00
Hajime Hoshi
eb3c45c8af Revert "internal/graphicsdriver/opengl, metal, directx: skip multiplying a scale when a color matrix is used"
This reverts commit b457dc3307.

Reason: a color scale might be used with ColorM (DrawTriangles)
2022-10-02 00:19:41 +09:00
Hajime Hoshi
520c47f0e8 Revert "ebiten: use Kage shaders instead of built-in shaders for a color matrix"
This reverts commit 7fc79c50c3.

Reason: a color scale might be used with ColorM (DrawTriangles)

Updates #2364
2022-10-02 00:19:37 +09:00
Hajime Hoshi
7fc79c50c3 ebiten: use Kage shaders instead of built-in shaders for a color matrix
Closes #2364
2022-10-02 00:08:14 +09:00
Hajime Hoshi
b457dc3307 internal/graphicsdriver/opengl, metal, directx: skip multiplying a scale when a color matrix is used 2022-10-01 18:14:22 +09:00
Hajime Hoshi
4203a3b68a internal/graphicsdriver/opengl, metal, directx: add comments
Updates #1212
2022-10-01 15:21:02 +09:00
Hajime Hoshi
a10f3d1dad internal/graphicsdriver/opengl, metal, directx: refactoring: clean up the built-in shaders 2022-10-01 14:35:26 +09:00
Hajime Hoshi
cff64894cc internal/affine: refactoring: use slices instead of array pointers
This doesn't change the performance at the test using ColorM:

```
name           old time/op  new time/op  delta
ColorMScale-8  1.11µs ±43%  1.23µs ±70%   ~     (p=1.000 n=5+5)
```
2022-09-30 22:20:32 +09:00
Hajime Hoshi
d00bd1cb16 internal/ui: refactoring: move setVerticesCache to ui.Image
Updates #2362
2022-09-29 01:21:46 +09:00
Hajime Hoshi
9492b4ecf5 internal/mipmap: refactoring
This change replaces the member `volatile` with `imageType` to make
the code more explicit.

In the old code, it was not obvious whether a mipmap was used for a
screen image. Actually a mipmap was not used since `canSkipMipmap` is
always true for a screen image, but this was too tricky.
2022-09-28 03:07:40 +09:00
Hajime Hoshi
6b35ad4a88 internal/mipmap: bug fix: mipmap images were unexpectedly volatile
Closes #2354
2022-09-28 01:40:38 +09:00
Hajime Hoshi
3a0f28ce6b internal/ui: refactoring: reduce global-variable usages 2022-09-26 00:46:03 +09:00
Hajime Hoshi
fa108ca717 internal/ui: bug fix: compiie errors on mobiles and browsers 2022-09-26 00:34:44 +09:00
Hajime Hoshi
4a82a109be internal/ui: use sync/atomic for performance 2022-09-26 00:27:04 +09:00
Hajime Hoshi
909f2fe492 internal/ui: bug fix: make inFrame concurrent-safe
Updates #1816
2022-09-25 23:56:14 +09:00
Hajime Hoshi
12f42544dd internal/ui: refactoring: simplify logic by removing setSizeCallbackEnabled
Closes #1816
2022-09-25 23:48:40 +09:00
Hajime Hoshi
70ebd34d99 internal/ui: refactoring: separate setWindowSizeInDIPImpl to setWindowSizeInDIP and setFullscreen 2022-09-25 22:55:56 +09:00
Hajime Hoshi
f0dbf86799 internal/ui: refactoring: adjust the width earlier 2022-09-25 22:34:14 +09:00
Hajime Hoshi
c7deed6e72 internal/ui: add setFullscreen 2022-09-25 22:31:26 +09:00
Hajime Hoshi
e47f9ac38f internal/ui: bug fix: wrong bit operations
Closes #2349
2022-09-25 20:10:37 +09:00
Hajime Hoshi
9302230fef internal/ui: refactoring 2022-09-25 19:56:53 +09:00
Hajime Hoshi
98076bd512 internal/ui: refactoring: simplify initializing fullscreen 2022-09-25 18:28:57 +09:00
Hajime Hoshi
6167cc4f54 internal/ui: remove unnecessary function calls
updateSize was introduced at 7eff5cbd4a, but
apparently calling setWindowSizeInDIP was originally not needed.
2022-09-25 18:19:35 +09:00
Hajime Hoshi
25405783a7 internal/ui: refactoring
Updates #1816
2022-09-25 02:05:26 +09:00
Hajime Hoshi
b2f3d39acf internal/restorable: add color tests at TestAllowWritePixelsForPartAfterDrawTriangles
Updates #2346
2022-09-24 22:09:08 +09:00
Hajime Hoshi
1703297f37 internal/shader: add a test for type shadowing 2022-09-24 19:00:23 +09:00
Hajime Hoshi
7f91a681e3 internal/shader: check type redeclaration 2022-09-24 18:56:01 +09:00
Hajime Hoshi
c5867c3357 internal/ui: bug fix: avoid 0 as arguments for Layout
Updates #2340
2022-09-20 22:50:39 +09:00
Hajime Hoshi
b5ddee3e4a internal/ui: bug fix: reentering updateImpl caused double unlocking
updateImpl can be invoked in multiple ways. This should have been
protected by a mutex, or this caused unexpected reentrance.

Closes #2339
2022-09-20 13:19:43 +09:00
divVerent
47de8027b9
internal/gamepaddb: fix mapping support for e.g. dpleft:-a0. (#2335)
SDL interprets this as "map -1 to 1, map 0 to -1", so we should do the same.

This fix contains two parts:
* Fix the intended output range.
* Also fix the formula to map a range to a range.
The fix does not change behavior if a "-a" mapping isn't used, as in any other case max-min == 1 or max+min == 0.

Fixes #2334
2022-09-20 00:34:46 +09:00
Hajime Hoshi
c7fcfe5bf7 internal/graphicsdriver/directx: refactoring
This makes more explicit that the pixels are sent after the slice
for the pixels finishes modification.
2022-09-19 00:31:44 +09:00
Hajime Hoshi
cfdf59ef8d internal/graphicsdriver/directx: revert changes to remove GetCopyableFootprints
This reverts these commits
* ccddd4ee20
* 8e40c2ce7e
* 2eb1efe3cc

Without GetCopyableFootprints, the test results on 32bit Windows seemed
much more flaky. This didn't fix the memory issue like #2294 anyway.
2022-09-19 00:26:59 +09:00
Hajime Hoshi
ccddd4ee20 internal/graphicsdriver/directx: refactoring: make the lifetime of uploadingStagingBuffer more explicit 2022-09-19 00:16:53 +09:00
Hajime Hoshi
8e40c2ce7e internal/graphicsdriver/directx: refactoring 2022-09-19 00:06:43 +09:00
Hajime Hoshi
2eb1efe3cc internal/graphicsdriver/directx: stop using GetCopyableFootprints
GetCopyableFootprints seems very conservative and returns a quite
big memory size. Instead of using the info from GetCopyableFootprints
let's use our own minimum footprint.
2022-09-18 23:10:15 +09:00
Hajime Hoshi
cd9477491d internal/gamepad: refactoring 2022-09-18 20:09:10 +09:00
Hajime Hoshi
78802f18d9 internal/glfwwin: remove GetProcAddress 2022-09-18 18:04:02 +09:00
Hajime Hoshi
129f201be4 internal/gamepaddb: use go:embed
Updates #1258
2022-09-17 21:02:07 +09:00
TotallyGamerJet
043397c20e
internal/ui: remove Cgo on darwin in ui_glfw_darwin.go (#2329)
Updates #1162
2022-09-16 11:53:46 +09:00
Hajime Hoshi
45b2bd7b2b internal/graphicsdriver/metal/mtl: remove a TODO comment 2022-09-16 11:48:59 +09:00
Hajime Hoshi
53df5aaaf0 internal/cocoa: rename _cmd to cmd 2022-09-16 02:32:12 +09:00
Hajime Hoshi
963eb2756a all: fix TODO comments 2022-09-16 02:02:02 +09:00
Hajime Hoshi
b13fa2817c internal/gamepad: clean up codes for Darwin
* Removes redundant assignments.
* Fixes types for the Apple libraries.

Updates #1162
2022-09-16 00:25:51 +09:00
TotallyGamerJet
4a569be6f6
internal/gamepad: remove Cgo on darwin (#2330)
Updates #1162
2022-09-16 00:11:59 +09:00
Hajime Hoshi
5404e4d68a all: replace io/ioutil with io and os
Closes #1770
2022-09-15 02:54:25 +09:00
Hajime Hoshi
d0556af8a4 internal/atlas: refactoring 2022-09-15 01:52:05 +09:00
Hajime Hoshi
31945563d8 internal/atlas: optimize adjustDestinationPixel
```
go test -bench=BenchmarkAdjustPixel -run=^$ -count=5 ./internal/atlas/
```

```
name           old time/op  new time/op  delta
AdjustPixel-8  2.59ns ± 1%  2.12ns ± 1%  -18.16%  (p=0.008 n=5+5)
```
2022-09-15 01:50:47 +09:00
Hajime Hoshi
9319266c01 internal/atlas: add a benchmark for adjustDestinationPixel 2022-09-15 01:25:26 +09:00
Hajime Hoshi
2cbc5e7b60 internal/cocoa: remove IsIOS
Use `runtime.GOOS == "ios"` instead.

Closes #1415
2022-09-15 01:04:37 +09:00
Hajime Hoshi
619a2ee4dd internal/gamepaddb: use runtime.GOOS instead of a build tag
Updates #1415
2022-09-15 00:27:11 +09:00
Hajime Hoshi
655cd4bf68 all: remove the build tag
Updates #1415
2022-09-15 00:23:40 +09:00
Hajime Hoshi
95bfa95a91 internal/graphicscommand: refactoring 2022-09-14 23:13:08 +09:00
Hajime Hoshi
86706c0335 internal/graphicsdriver/opengl: add opengles build tag
This enables to use OpenGL ES instead of OpenGL.

Closes #292
2022-09-13 11:10:48 -07:00
Hajime Hoshi
18fa00d25d internal/restorable: update comments 2022-09-14 02:20:11 +09:00
Hajime Hoshi
6179158812 internal/restorable: bug fix: no pixels were read only with DrawTriangles
Closes #2324
2022-09-14 02:17:14 +09:00
Hajime Hoshi
b36c21648f internal/restorable: refactoring: remove a redundant FlushCommands call
readPixelsFromGPU already flushes commands.
2022-09-13 08:43:02 -07:00
Hajime Hoshi
6b906bb813 internal/restorable: extend the staleRegion when the image is already stale on WritePixels
This is a kind of refactoring. There seems no case that this condition
change is really needed, but this is more logical.
2022-09-13 07:29:53 -07:00
Hajime Hoshi
67ceaf5874 internal/atlas: rename replacePixels -> writePixels 2022-09-13 06:39:29 -07:00
Hajime Hoshi
243c224003 internal/restorable: bug fix: wrong panic on mixing DrawTriangles and WritePixels
When DrawTriangles is called and then WritePixels is called on a
sub-image, a panic happened. However, this panic actually happens
only when the graphics driver requires restoring (e.g. OpenGL ES
on Android). The situation was very limited, but this was a real
problem on Android.

This panic was introduced to prevent a rendering bug by a inmature
graphics drivers, but we should no longer need this. This change
just removes the panic.

Updates #292
2022-09-13 06:18:55 -07:00
Hajime Hoshi
40362aa62d internal/restorable: bug fix: readPixelsFromGPU might be called for a non-stale image
Actually readPixelsFromGPUIfNeeded can invoke this.

Closes #2322
2022-09-13 15:34:43 +09:00
Hajime Hoshi
4824cbc755 internal/graphicscommand: bug fix: IsInvalidated was not concurrent-safe
This function was not called actually, so this is not a real problem.
However, this could be a potential problem for a future GLES driver (#292).

Updates #292
Closes #2321
2022-09-13 12:23:19 +09:00
Hajime Hoshi
6ea455f4e1 internal/graphicsdriver/opengl/gles: remove cString 2022-09-13 00:17:33 +09:00
Hajime Hoshi
e46bfd8d30 internal/graphicsdriver: rename files
Updates #292
2022-09-12 23:03:54 +09:00
Hajime Hoshi
7d6ee39c15 internal/gamepad: bug fix: a wireless Xbox controller didn't work correctly on macOS
The button and hat values must be adjusted by their minimum values.
See also GLFW's implementation.

Closes #266
2022-09-12 01:55:05 +09:00
Hajime Hoshi
89598df520 internal/gamepad: refactoring
Updates #2311
2022-09-11 22:44:06 +09:00
Hajime Hoshi
4629306f4b internal/gamepad: bug fix: wrong SDL ID on iOS
The gamepad database already worked well with this fix.

Closes #2311
2022-09-11 22:29:38 +09:00
Hajime Hoshi
14a3a0f6bc internal/graphicsdriver/opengl/gl: better error message 2022-09-10 14:21:59 +09:00