01288be937
Use the same X display name for VDPAU that SDL is using
2025-11-23 13:20:36 -06:00
66b96e75af
Convert all time-based stats to be microsecond-based, improving accuracy of very fast events. Add bitrate average/peak.
2025-11-15 12:17:59 -06:00
8324f61db0
Reduce per-frame GL calls in EGLRenderer
...
Uniforms are attached to the program object, so they don't need to
be set each time we swap between the YUV shader and the overlay
shader.
Since 0 overlays is by far the most common case and > 1 is highly
unlikely, move the glViewport() call into renderOverlay() to let
us skip it in the common case (and be no worse than today in the
2nd most common case of 1 overlay).
2025-11-06 17:52:30 -06:00
a2670dd568
Use hasFrameFormatChanged() in DRMRenderer
2025-11-06 00:04:08 -06:00
c54e6e4459
Use hasFrameFormatChanged() in VDPAU
2025-11-05 23:58:51 -06:00
dd6f1c5873
Use hasFrameFormatChanged() in SDLRenderer
2025-11-05 23:51:51 -06:00
7fab5007a8
Switch Metal to the shared functions for CSC matrix generation and chroma co-siting
2025-11-05 23:22:15 -06:00
7f7cc89e61
Switch EGLRenderer to the shared functions for CSC matrix generation and chroma co-siting
2025-11-05 22:27:31 -06:00
7116efd8de
Switch D3D11VA to the shared functions for CSC matrix generation and chroma co-siting
2025-11-05 22:26:27 -06:00
229f5e4cea
Add functions for color matrix generation and chroma co-siting in the renderer interface
2025-11-05 21:59:13 -06:00
af37002a60
Select correct color matrix for 10-bit SDR
2025-11-05 00:13:25 -06:00
181dba5864
Use correct chroma co-siting offsets in Metal
2025-11-04 23:51:39 -06:00
fd314db27e
Use correct chroma co-siting offsets in D3D11
...
Introduced in f67272b153
2025-11-04 23:50:27 -06:00
f67272b153
Handle chroma co-siting in the D3D11 shaders
2025-11-01 22:42:51 -05:00
9b3050514b
Handle chroma co-siting in the Metal shaders
2025-11-01 21:57:13 -05:00
2c12ad297f
Fix accuracy issues in YUV to RGB conversion of 10-bit content in Metal
...
See #1667 for details.
2025-11-01 19:19:19 -05:00
50223dbdb8
Fix accuracy issues in YUV to RGB conversion of 10-bit content in D3D11VA
...
See #1667 for details.
2025-11-01 18:45:02 -05:00
ccaca68570
Don't use CAMetalDisplayLink on Intel Macs
...
When in Direct mode, skipping a frame will cause the display to flash black on Tahoe.
2025-10-23 21:16:53 -05:00
57db20016a
Rewrite frame pacing mechanism for VTMetalRenderer
...
This should hopefully improve performance and avoid performance issues on Tahoe.
2025-10-23 19:43:10 -05:00
200cab9d17
Attach EDR metadata to Metal layer
...
This improves the accuracy of HDR streaming and enables HDR->SDR tonemapping.
2025-10-20 22:28:27 -05:00
a20d429bc1
Remove usage of deprecated Metal API
2025-10-11 11:41:28 -05:00
1fd545ae1f
build: fix no <linux/dma-buf.h> for BSDs
2025-08-31 15:05:03 -05:00
bf51577787
Fix macOS build
2025-02-18 19:52:33 -06:00
fd70865026
Avoid retrying renderer init if we know the error was not transient
2025-02-18 19:02:12 -06:00
351aaa6759
Add an enum type for each renderer
2025-02-18 18:58:04 -06:00
dd2a99a96b
Prepare for SDL3 support
2025-01-25 16:20:20 -06:00
4af9623727
Replace C++ VLAs with std::vector
2024-10-15 22:41:21 -05:00
3531fe0a4f
Fix typo in version check for new Vulkan queue API
2024-10-02 00:37:22 -05:00
6ce02616f0
Use new Vulkan queue API in FFmpeg 7.1
2024-09-29 23:04:23 -05:00
901cbd255c
Disable Vulkan debug layers by default
...
The vast majority of Vulkan code running in Moonlight
is part of FFmpeg or libplacebo, so the debug layers
really just slow things down without finding any bugs
in our code.
Additionally, there are some overzealous checks firing
constantly on libplacebo and FFmpeg with certain Vulkan
drivers that we can't do anything about.
2024-09-28 00:05:17 -05:00
023b6b2772
Fix DRM FD leak with Vulkan windows
2024-09-21 21:41:43 -05:00
6f39d120cb
Unify handling of DRM devices between DRM and VAAPI
...
SDL may not be able to give us a DRM FD for Vulkan windows.
2024-09-21 20:46:11 -05:00
6b11f43302
Don't set the output rect until after modesetting
...
It's possible that modesetting will also change the resolution
not just the refresh rate. This can happen in cases where the
CRTC is currently set to 4K 30 Hz and we choose 1080p 60 Hz as
a better mode match for displaying a 1080p 60 FPS stream.
2024-09-19 22:43:34 -05:00
76deafbd7b
Fix build for Xcode < 14
2024-09-18 22:12:04 -05:00
f786e94c7b
Fix excessive latency with Vulkan video decoding on Nvidia
...
For yet unknown reasons, high frame queue delay leads to massive decode latency.
2024-09-16 21:59:17 -05:00
4838a75c58
Suppress spammy libplacebo messages
...
Streaming 10-bit content on Nvidia GPUs spams the log with the following for every rendered frame:
Masking `blit_src` from wrapped texture because the corresponding format 'rx10' does not support PL_FMT_CAP_BLITTABLE
Masking `blit_src` from wrapped texture because the corresponding format 'rxgx10' does not support PL_FMT_CAP_BLITTABLE
2024-09-15 22:01:39 -05:00
c5ec8c0fdb
Add libplacebo prefix to log messages
2024-09-15 21:58:19 -05:00
b266f61ba6
Disable direct rendering of YUV444 using vaPutSurface()
...
It doesn't work as of libva 2.21 + iHD 24.2.5 despite the fact that
the VAEntrypointVideoProc advertises VA_RT_FORMAT_YUV444.
2024-09-15 21:03:00 -05:00
96fb6ee5e9
Share SDL's EGL display with EGLRenderer
...
Fixes #1396
2024-09-12 22:37:25 -05:00
92ee4a3046
Don't use the bind path for UMA GPUs
...
It turns out to be very marginally slower on RDNA2 iGPUs.
2024-09-12 21:32:36 -05:00
6ec0c79899
Revert "Remove FL 11.0 restriction for D3D11VA"
...
Even with a fix for the Intel IVB rendering issue, there are still problems on
other older cards that make it not really worth it to enable D3D11VA.
The big benefit of D3D11VA is independent flip in windowed mode using
overlays and that isn't supported on those old cards anyway.
This reverts commit 7074463d0f
2024-09-12 21:29:40 -05:00
67e89d9e16
Add workaround for incorrect color range in AMF AV1 encoding output
2024-09-04 23:13:48 -05:00
25132a1f7b
Enable decoder texture binding for all FL11.1 iGPUs
...
Since those GPUs don't benefit from the fence hack, restrict that to Intel only.
2024-08-31 00:36:25 -05:00
7074463d0f
Remove FL 11.0 restriction for D3D11VA
2024-08-30 23:03:36 -05:00
99311403fa
Move dummy fence into renderVideo()
...
The workaround only seems to work with a new fence each time.
2024-08-30 22:42:53 -05:00
8b50eea485
Wait on a dummy fence before using textures shared with the decoder
2024-08-30 22:22:30 -05:00
f756be87ff
Fix YUV444 hardware decoding on Windows systems without an Intel GPU present
2024-08-30 16:31:58 -05:00
4290a54ffa
Log the chosen Metal device
2024-08-29 23:20:52 -05:00
caf322d934
Provide valid default preferred formats for YUV444
2024-08-29 01:03:39 -05:00
76fd502262
Reintroduce support for FFmpeg 4
...
Some SBCs still depend on patched FFmpeg 4.x builds for hardware acceleration.
2024-08-24 16:50:27 -05:00
Cameron Gutman
Andy Grundman
ns6089
armin-25689
Matt Tannahill