zdoom/gzdoom-gles
0
0
Fork 0
mirror of https://github.com/ZDoom/gzdoom-gles.git synced 2024-12-15 06:41:27 +00:00
Code Issues Projects Releases Packages Wiki Activity
17271 commits 6 branches 6 tags 107 MiB
Commit graph

189 commits

Author SHA1 Message Date
Magnus Norddahl
188714dc14 - the light storage buffer is not dynamic 2019-04-28 23:51:09 +02:00
Christoph Oelckers
90cc0cf35b Merge remote-tracking branch 'remotes/origin/vulkan2' 2019-04-22 11:43:43 +02:00
Christoph Oelckers
0640a86c1e - removed InitPalette call from VulkanFrameBuffer constructor. 
It is as wrong here as in OpenGL because it destroys already set up data.
 2019-04-22 11:00:40 +02:00
Magnus Norddahl
09883431bf - fix wrong clamp mode used in OpenGL 
- fix sampler array size on Vulkan
 2019-04-20 19:19:34 +02:00
Magnus Norddahl
a3587009e7 - the semaphore should never be added when its the last submit before waiting 2019-04-20 16:42:52 +02:00
Magnus Norddahl
401a4944b4 - fix Queue is signaling semaphore that has not been waited on by any queue 2019-04-20 16:35:26 +02:00
Magnus Norddahl
6699cd5462 - change FRenderState to store directly to the StreamData struct. This simplifies the vulkan backend and also allows the OpenGL backend to use the same uniform block transfer strategy in the future. 2019-04-20 04:16:01 +02:00
Magnus Norddahl
da7a4ceb34 - change drawcalls to measure the Apply time rather than the individual draw calls (Apply is what costs something on vulkan while the draw call queuing is so cheap its an uninteresting thing to measure) 2019-04-20 00:38:38 +02:00
Magnus Norddahl
a8e7f38150 - minor code cleanup 2019-04-19 23:56:54 +02:00
Magnus Norddahl
779cb42578 - remove vk_submit_multithread again as it seemed to have no effect on performance and only complicated the code 2019-04-19 23:26:06 +02:00
Magnus Norddahl
458da39c39 - add vk_submit_multithread for doing command buffer submit calls on a worker thread 
- add vk_submit_size for testing various command buffer sizes before flushing them
- add submitted command buffer count to renderstats
 2019-04-19 22:42:32 +02:00
Magnus Norddahl
3957a19bd0 - flush the commands for every 1000th Apply call 2019-04-19 21:08:15 +02:00
Magnus Norddahl
8fadf3d9bd - add support for flushing commands during drawing 2019-04-19 20:55:15 +02:00
Magnus Norddahl
faac0805f0 - implement VKBuffer::Resize 2019-04-18 22:01:42 +02:00
Magnus Norddahl
e332011995 - fix wrong model culling 2019-04-18 17:56:43 +02:00
Magnus Norddahl
47fa7dafe3 - fix shader timer not always being applied correctly 2019-04-18 17:14:26 +02:00
Magnus Norddahl
47f056e882 - improve shader error handling and attempt to remove some bogus declarations 2019-04-18 01:20:28 +02:00
Magnus Norddahl
d63513ec14 - fix normal vectors on models 2019-04-17 20:42:00 +02:00
Magnus Norddahl
06222a1fbf - update glslang to 7.11.3113 (stable release February 8, 2019). This fixes the SPIR-V validation error reported for the shadowmap shader 2019-04-15 04:53:43 +02:00
Rachael Alexanderson
482e9c0f2c - fixed: Intel OpenGL did not quite recognize the "layout(location=0)" shader uniform definitions, so this had to be removed for OpenGL mode only so that Intel's compiler will continue to function normally for custom shaders. 2019-04-11 13:21:40 -04:00
Magnus Norddahl
40a72c71f5 - fix shutdown crash 2019-04-11 18:21:51 +02:00
Magnus Norddahl
1e7911d1f9 - we cannot recover from this 2019-04-11 05:36:36 +02:00
Magnus Norddahl
59904faff4 - slightly adjust AcquireImage to avoid border cases in the spec 2019-04-11 05:28:37 +02:00
Magnus Norddahl
a0f618311a - change SubmitCommands to handle all parts of the present 2019-04-11 04:52:57 +02:00
Magnus Norddahl
095ea3ce76 - apply vk_hdr immediately and delay initial swapchain creation until first image needs to be acquired 2019-04-11 04:26:43 +02:00
Magnus Norddahl
5f70ce2148 - fix SubmitCommands bug if it no swapchain image was acquired 2019-04-10 16:39:33 +02:00
Magnus Norddahl
85b754b9ca - fix model rendering glitch 2019-04-09 18:19:14 +02:00
Magnus Norddahl
d413581ee2 - fix model shutdown crash 2019-04-09 17:06:54 +02:00
Magnus Norddahl
eb9f6ec313 - remove the 6 layer texture descriptor set limitation 2019-04-09 16:30:49 +02:00
Magnus Norddahl
c9dbb589e2 - once the Vulkan backend booted all the remaining vulkan calls are unrecoverable (unless the code calling it gets actively involved, which means that particular action is not exceptional in nature and shouldn't be done as an exception) 2019-04-09 14:25:18 +02:00
Magnus Norddahl
d47891d20a - allocate new texture descriptor pool if it is full 2019-04-09 14:06:24 +02:00
Magnus Norddahl
2cbd1c4736 - Fix inverted check in last commit 2019-04-09 12:41:05 +02:00
Magnus Norddahl
d3dacfc2cb - improve error handling during vulkan initialization 2019-04-09 12:28:25 +02:00
Christoph Oelckers
ab256945aa - use I_Error for throwing errors in the Vulkan backend and print the message if one gets thrown during init. 2019-04-08 23:48:46 +02:00
Magnus Norddahl
2694b0a167 - workaround for drivers that bug by never acquiring an image 2019-04-08 21:01:40 +02:00
Magnus Norddahl
c98dfd1790 - improve swap chain resize and handle the edge cases for the swap chain 2019-04-08 19:23:37 +02:00
Magnus Norddahl
20fde9f8be - clean up swapchain class 2019-04-08 18:14:07 +02:00
Magnus Norddahl
d114575bd1 - implement custom post process shaders for vulkan backend 2019-04-08 00:47:55 +02:00
Magnus Norddahl
f7069c4ddc - there is no need to track a transfer family - yet another pointless detour thanks to vulkan-tutorial.. 2019-04-04 15:58:48 +02:00
alexey.lysiuk
df8fa90a34 - fixed compilation warnings reported by Clang 
src/rendering/vulkan/renderer/vk_renderpass.cpp:44:22: warning: comparison of integers of different signs: 'std::__1::vector<FVertexBufferAttribute, std::__1::allocator<FVertexBufferAttribute> >::size_type' (aka 'unsigned long') and 'int' [-Wsign-compare]
src/rendering/vulkan/system/vk_framebuffer.cpp:860:55: warning: format specifies type 'int' but the argument has type 'VkDeviceSize' (aka 'unsigned long long') [-Wformat]
src/rendering/vulkan/system/vk_objects.h:471:23: warning: suggest braces around initialization of subobject [-Wmissing-braces]
 2019-03-30 11:15:49 +02:00
Magnus Norddahl
9f0f659db0 - remove old vid_maxfps implementations as they were garbage anyway and the new one works on all the platforms 2019-03-26 11:10:17 +01:00
Magnus Norddahl
6078428b84 - fix typo in sleep 2019-03-26 10:46:31 +01:00
Magnus Norddahl
9f758b0032 - implement vid_maxfps 2019-03-26 10:40:43 +01:00
Christoph Oelckers
b40983be61 - fixed bad Printf formatter. 2019-03-26 08:48:10 +01:00
Magnus Norddahl
1c9bf262e6 - hook up VMA_ALLOCATOR_CREATE_KHR_DEDICATED_ALLOCATION_BIT 2019-03-25 21:30:03 +01:00
Magnus Norddahl
7256af0b32 - fix sort bug 2019-03-25 20:41:53 +01:00
Magnus Norddahl
2d8516b2ca - make shaderClipDistance optional 2019-03-25 19:44:46 +01:00
Magnus Norddahl
be74675e5e - change vk_hdr default to false 2019-03-25 01:41:16 +01:00
Magnus Norddahl
c010c5e818 - fully enable vk_hdr as a feature since it works now after the queue family selection was improved 2019-03-25 01:39:27 +01:00
Magnus Norddahl
954b72915c - improve the queue family selection process to pick first entry in the list over later ones 2019-03-24 18:33:59 +01:00