rooty/dxvk
1
0
Fork 0
mirror of https://github.com/doitsujin/dxvk.git synced 2024-12-04 16:24:29 +01:00
Code Issues Projects Releases Wiki Activity
5,469 Commits 24 Branches 83 Tags 24 MiB
0e36a07a93
Commit Graph

5119 Commits

Author SHA1 Message Date
Joshua Ashton
0e36a07a93 [d3d9] Don't mark DS as hazardous if depth write is disabled 2023-06-24 03:55:16 +01:00
Joshua Ashton
ff65599dba [d3d9] Use m_activeRTs in BindFramebuffer 
Can roll in the anyColorWrites this way.
 2023-06-24 03:42:54 +01:00
Joshua Ashton
6b60de2d31 [d3d9] Fix unbinding RTs 
Fixes a regression with 8560efa3c7 in ND1.

Also more optimized.
 2023-06-24 03:41:31 +01:00
Joshua Ashton
362743c1d6 [d3d9] Update DS hazards when PS shader masks change 2023-06-24 02:55:38 +01:00
Joshua Ashton
7f302fc350 [d3d9] Don't mark DS hazards if not used by shader 2023-06-24 02:53:14 +01:00
Philip Rebohle
8704ed7af6 [dxvk] Ignore some pipeline flags when ending render pass 2023-06-22 23:41:24 +02:00
Philip Rebohle
0895858901 [dxvk] Only decrement pending submission count for command submissions 
Otherwise we'll underflow the integer and break the D3D9 flush heuristic.
 2023-06-22 20:15:42 +02:00
Joshua Ashton
80b27f95bc [d3d9] Fix active hazards RT only being 4 bits 2023-06-22 16:37:09 +01:00
Joshua Ashton
a791493d14 [d3d9] Don't rebind AlphaTest when changing RT if not necessary 2023-06-22 16:37:09 +01:00
Joshua Ashton
c768196251 [d3d9] Compress hazard state going into CS 
This can potentially happen per-draw so make it as small as possible.
 2023-06-22 16:37:09 +01:00
Joshua Ashton
3625c5d481 [d3d9] Optimize DS active hazard check 
There can only be one DS, so no need to loop over.
 2023-06-22 16:37:09 +01:00
Philip Rebohle
b3cbe36c08 [dxvk] Apply frame rate limit on presentation timeline 
This may reduce latency as we no longer end up stalling subsequent
GPU submissions.
 2023-06-22 16:17:03 +02:00
Philip Rebohle
166d90b04c [d3d11] Fix frame latency based on buffer count 
We need to ignore the front buffer here since we synchronize after
presentation.
 2023-06-22 16:16:30 +02:00
Robin Kertels
60b6e98529 [d3d9] Transition DS too if there's a feedback loop 2023-06-21 15:16:43 +01:00
Robin Kertels
a20869fb93 [d3d9] Track textures in m_activeHazardsRT instead of RT 
There's 21 textures and only 4 RTs.
Tracking the textures allows us to mask off the active texture bitfield
instead of the active render target one, potentially resulting in fewer iterations.
 2023-06-21 15:16:43 +01:00
Philip Rebohle
a287566c65 [dxgi] Implement frame statistics based on IDXGIVkSwapChain1 2023-06-21 15:16:37 +02:00
Philip Rebohle
28f48f9fdc [dxgi] Initialize output refresh counts with non-zero values 
Matches Windows behaviour.
 2023-06-21 15:16:37 +02:00
Philip Rebohle
e02a800c33 [d3d11] Implement IDXGIVkSwapChain1 interface for D3D11 swap chain 2023-06-21 15:16:37 +02:00
Philip Rebohle
d7fa39c4eb [util] Be more robust against timing weirdness when computing vblank count 2023-06-21 15:16:37 +02:00
Philip Rebohle
d1e39be7e7 [dxgi] Add IDXGIVkSwapChain1 interface definition 2023-06-21 15:16:37 +02:00
Philip Rebohle
438c535fe7 [d3d11] Always release frame latency semaphore in Present 2023-06-21 15:16:37 +02:00
Philip Rebohle
7dbe4abb48 [d3d9] Use new presenter signal mechanism for frame pacing 2023-06-21 15:16:37 +02:00
Philip Rebohle
e99bc591df [d3d11] Use new presenter signal mechanism for frame latency event 2023-06-21 15:16:37 +02:00
Philip Rebohle
08363edb05 [dxvk] Only synchronize with presents with FIFO present modes 2023-06-21 15:16:37 +02:00
Philip Rebohle
5d1196733b [dxvk] Implement waiting for specific present requests 2023-06-21 15:16:37 +02:00
Philip Rebohle
ca3492570c [dxvk] Add functionality to wait for a given present operation 2023-06-21 15:16:37 +02:00
Philip Rebohle
215c4f8f6f [dxvk] Enable VK_KHR_present_id and VK_KHR_present_wait if supported 2023-06-21 15:16:37 +02:00
Robin Kertels
5a1ebfa4ee Revert "[d3d9] Only use direct buffer mapping for DYNAMIC buffers" 
This reverts commit 1850819483.
 2023-06-20 23:08:52 +01:00
Joshua Ashton
6f87ccdafc [d3d11] Use FORCE_UNORM depth bias representation for UNORM formats 2023-06-20 13:31:48 +01:00
Joshua Ashton
b30a4f0cc7 [d3d9] Use FLOAT or FORCE_UNORM depth bias representation 2023-06-20 13:31:48 +01:00
Joshua Ashton
4e9853f608 [dxvk] Expose depth bias representation/exact controls 2023-06-20 13:31:48 +01:00
Joshua Ashton
5fbb0dd4ba [dxvk] Enable EXT_depth_bias_control 2023-06-20 13:31:48 +01:00
Alpyne
0b9acf3a25 [util] Add str::split 
Used by d3d8
 2023-06-20 13:31:17 +01:00
Timo Gurr
211d095ee4 [util] Limit STEINS;GATE ELITE to 60 fps 
Intros, menu and ui animations are playing way too fast, also causing
input issues.
 2023-06-20 13:27:52 +02:00
Trevonn
77e7e8bfba [util] Report NVIDIA for Star Wars Battlefront as it errors out on AMD 2023-06-20 13:23:52 +02:00
Alpyne
5d29140f74 [util] replaceNaN: Use unaligned SIMD _mm_loadu_ps 
There is no good reason to expect games will have aligned the data they're passing in.
 2023-06-19 21:13:37 +01:00
Alpyne
0236e780a7 [d3d9] Fix std::hex in D3D9CommonTexture 2023-06-19 21:02:10 +01:00
Alpyne
24dbcf8fd8 [dxso] Shared code for texbem and bem 2023-06-19 20:59:18 +01:00
Alpyne
404c984f9c [dxso] Implement bem instruction 2023-06-19 20:59:18 +01:00
Alpyne
a79772322b [d3d9] Initialize m_mapFlags to 0 
Can sometimes get weird default values in there.
 2023-06-19 19:52:03 +01:00
Trevonn
36e6a7c2e5 [util] Set maxChunkSize to 1 for GOG and fix EA App exe name 
Similar to the other launchers the VRAM goes 600MB+ to 100MB+
 2023-06-16 16:28:31 +01:00
Trevonn
c5ab5be48d [util] Set maxChunkSize to 1 for EA App 
Reduces GPU VRAM usage of EADesktop.exe from 162MB to 6MB
 2023-06-15 19:39:28 +01:00
Paul Gofman
af9bd16b8d [dxvk] Always reference library in VR extension providers 2023-06-15 19:07:26 +02:00
Winter Snowfall
2c014fdb34 [util] Match on the GOG version of KoF XIII as well 2023-06-15 19:06:45 +02:00
Timo Gurr
6478c10a18 [util] Set maxChunkSize to 1 for Battle.net 
Currently when Diablo 4 is running the Battle.net launcher it uses around
600MB of VRAM. Reducing the max chunk size to 1MB reduces the memory usage
to around 150MB.
 2023-06-15 19:06:26 +02:00
Philip Rebohle
7388c243d2 [dxvk] Fix incorrect fullScreenExclusive feature check 2023-06-15 17:32:05 +02:00
Lilium
bd575a4a46 [d3d11] implement 'clampNegativeLodBias' as a conf option 2023-06-14 20:22:27 +01:00
Lilium
3fce9886f5 [d3d9] implement 'clampNegativeLodBias' as a conf option 2023-06-14 20:22:27 +01:00
Lilium
5d134b877a [d3d9] implement 'samplerLodBias' as a conf option 2023-06-14 20:22:27 +01:00
Robin Kertels
c75ed86909 [d3d9] Reset vertex & index buffer in Reset 2023-06-13 22:47:55 +01:00