path: root/DOCS
diff options
authorNiklas Haas <>2017-10-07 21:36:16 +0200
committerMartin Herkt <>2017-12-25 00:47:53 +0100
commit286d421666fd812dc4d01e580204cf63b49fec45 (patch)
treeb5fb0257aca4f56a62c0e9ae4211a95c01163214 /DOCS
parenta6aab5dfd6d94be4e2a07c2bf6044f621b28e631 (diff)
vo_gpu: vulkan: allow disabling async tf/comp
Async compute in particular seems to cause problems on some drivers, and even when supprted the benefits are not that massive from the tests I have seen, so it's probably safe to keep off by default. Async transfer on the other hand seems to work better and offers a more substantial improvement, so it's kept on.
Diffstat (limited to 'DOCS')
1 files changed, 17 insertions, 0 deletions
diff --git a/DOCS/man/options.rst b/DOCS/man/options.rst
index f50f67a203..ed27598b8a 100644
--- a/DOCS/man/options.rst
+++ b/DOCS/man/options.rst
@@ -4298,6 +4298,23 @@ The following video options are currently all specific to ``--vo=gpu`` and
1), but it can also slow things down on hardware where there's no true
parallelism between queues. (Default: 1)
+ Enables the use of async transfer queues on supported vulkan devices. Using
+ them allows transfer operations like texture uploads and blits to happen
+ concurrently with the actual rendering, thus improving overall throughput
+ and power consumption. Enabled by default, and should be relatively safe.
+ Enables the use of async compute queues on supported vulkan devices. Using
+ this, in theory, allows out-of-order scheduling of compute shaders with
+ graphics shaders, thus enabling the hardware to do more effective work while
+ waiting for pipeline bubbles and memory operations. Not beneficial on all
+ GPUs. It's worth noting that if async compute is enabled, and the device
+ supports more compute queues than graphics queues (bound by the restrictions
+ set by ``--vulkan-queue-count``), mpv will internally try and prefer the
+ use of compute shaders over fragment shaders wherever possible. Not enabled
+ by default, since it seems to cause issues with some drivers.
Use WARP (Windows Advanced Rasterization Platform) with the D3D11 GPU
backend (default: auto). This is a high performance software renderer. By