This project is a fork of OpenArena with specific changes to its renderer module. I am naive programmer, this repository is mainly for myself learning Vulkan and Quake3's engine.
For people who want to try the vulkan based renderer on quake3's map, go to https://github.com/suijingfeng/vkQuake3
Install the build dependencies.
$ sudo apt-get install libcurl4-openssl-dev libsdl2-dev libopenal-dev libvulkan-dev libgl1-mesa-dev
$ sudo apt-get install clang gcc make git
$ git clone https://github.com/suijingfeng/vkOpenArena.git
$ cd vkOpenArena
$ make -j4Please note that vulkan renderer requires at least SDL 2.0.6. The precompiled versions in some of the distribute repositories do not ship with Vulkan support, you therefore may come cross the following problem when you launch OA:
Vulkan support is either not configured in SDL or not available in video driver.
This problem can be solved easily by compiling the SDL 2.0.9 from source:
Get a copy of the source code from https://www.libsdl.org/, then extract it.
cd SDL
mkdir build
cd build
../configure
make -j4
sudo make install
To build 64-bit binaries, follow these instructions:
- Install msys2 from https://msys2.github.io/ , following the instructions there.
- Start "MinGW 64-bit" from the Start Menu, NOTE: NOT MSYS2.
- Install mingw-w64-x86_64, make, git and necessary libs.
pacman -S mingw-w64-x86_64-gcc make git- Grab latest openarena source code from github and compile. Note that in msys2, your drives are linked as folders in the root directory: C:\ is /c/, D:\ is /d/, and so on.
git clone https://github.com/suijingfeng/vkOpenArena.git
cd vkOpenArena
make -j4- Find the executables and dlls in build/release-mingw64-x86_64 .
First, download the map packages from http://openarena.ws/download.php Second, extract the data files at ~/.OpenArena/ (on linux) C:\Users\youname\AppData\Roaming\OpenArena\ (on windows)
$ cd /build/release-linux-x86_64/
$ ./openarena.x86_64Q: How to enable vulkan support from the pulldown console ?
\cl_renerer vulkan
\vid_restart
# then play
# Enable renderergl2 ported from ioq3
\cl_renerer opengl2
\vid_restart
# Enable renderergl1 ported from ioq3
\cl_renerer opengl1
\vid_restart
# Enable openarena
\cl_renerer openarena
\vid_restartWhen you start OpenArena, you can switch witch dynamic library to load by passing its name.
Example:
# New vulkan renderer backend, under developing,
# work on ubuntu 18.04, ubuntu16.04, win10, win7.
# associate code located in code/renderer_vulkan, see readme there.
$ ./openarena.x86_64 +set cl_renderer vulkan
# Enable renderergl2( borrowed from ioq3 ):
$ ./openarena.x86_64 +set cl_renderer opengl2
# Enable the renderergl1( borrowed from ioq3 ):
$ ./openarena.x86_64 +set cl_renderer opengl1
# Enable the default OpenArena renderer:
# This renderer module is similiar to the renderergl1 code.
$ ./openarena.x86_64 +set cl_renderer openarenaQ: How to use check FPS or using it as a benchmarking tool?
\timedemo 1
\demo demo088-test1
# Frames TotalTime averageFPS minimum/average/maximum/std deviation
# tested on win10 OS:
render_vulkan: 3398 frames 4.3 seconds 793.6 fps 1.0/1.3/4.0/0.5 ms
render_gl2: 3398 frames 24.6 seconds 138.3 fps 1.0/7.2/18.0/1.8 ms
render_gl1: 3398 frames 8.1 seconds 417.2 fps 1.0/2.4/8.0/0.9 ms
render_oa: 3398 frames 8.0 seconds 423.7 fps 1.0/2.4/8.0/0.9 ms
# tested on Ubuntu18.04:
render_vulkan: 3398 frames 3.6 seconds 935.6 fps 0.0/1.1/2.0/0.3 ms
render_gl2: 3398 frames 6.8 seconds 503.2 fps 1.0/2.0/6.0/0.6 ms
render_gl1: 3398 frames 5.4 seconds 629.7 fps 1.0/1.6/4.0/0.6 ms
render_mydev: 3398 frames 5.0 seconds 686.2 fps 1.0/1.5/3.0/0.5 ms
render_oa: 3398 frames 5.3 seconds 646.9 fps 1.0/1.5/4.0/0.5 ms
# render_vulkan on Ubuntu18.04
\com_speed 1
frame:36203 all: 1 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 1
frame:36204 all: 0 sv: 0 ev: 7 cl: 0 gm: 0 rf: 0 bk: 0
frame:36205 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36206 all: 1 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 1
frame:36207 all: 0 sv: 0 ev: 7 cl: 0 gm: 0 rf: 0 bk: 0
frame:36208 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36209 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36210 all: 0 sv: 0 ev: 7 cl: 0 gm: 0 rf: 0 bk: 0
frame:36211 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36212 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36213 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36214 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
frame:36215 all: 1 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 1
frame:36216 all: 0 sv: 0 ev: 7 cl: 0 gm: 0 rf: 0 bk: 0
frame:36217 all: 0 sv: 0 ev: 8 cl: 0 gm: 0 rf: 0 bk: 0
# TODO : make the game run at 1000 FPS.
# Without loading some non-existing resource which waste some time, I guess this game can got 1000 FPS.
\timedemo 1
\demo demo088-test1
# Testing on UBUNTU 16.04, driver: 384.130, tested on 2019.05.29,
vulkan : 3398 frames 21.0 seconds 162.2 fps 3.0/6.2/11.0/1.1 ms
opengl2: 3398 frames 25.1 seconds 135.5 fps 4.0/7.4/14.0/1.5 ms
opengl1: 3398 frames 10.1 seconds 337.0 fps 1.0/3.0/6.0/0.8 ms
# Testing on WIN10 Pro 64-bit, Driver version 397.31
vulkan : 3398 frames 14.5 seconds 233.6 fps 1.0/4.3/14.0/1.6 ms
opengl2: 3398 frames 83.4 seconds 40.8 fps 8.0/24.5/331.0/9.6 ms
opengl1: 3398 frames 22.2 seconds 152.8 fps 2.0/6.5/17.0/1.9 ms
# opengl2 renderer acieve best virsual result but consume the compute power most
\com_speed 1
# opengl2 backend
frame:24071 all: 13 sv: 0 ev: 0 cl: 0 gm: 0 rf: 1 bk: 12
frame:24072 all: 19 sv: 0 ev: 0 cl: 0 gm: 0 rf: 1 bk: 18
frame:24073 all: 9 sv: 1 ev: 0 cl: 0 gm: 0 rf: 1 bk: 7
frame:24074 all: 20 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 19
frame:24075 all: 9 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 8
frame:24080 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24081 all: 8 sv: 0 ev: 1 cl: 0 gm: 0 rf: 1 bk: 7
frame:24083 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24084 all: 17 sv: 0 ev: 0 cl: 1 gm: 0 rf: 1 bk: 15
frame:24085 all: 9 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 8
frame:24086 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24090 all: 9 sv: 1 ev: 0 cl: 0 gm: 0 rf: 0 bk: 8
frame:24092 all: 8 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 7
frame:24093 all: 9 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 8
frame:24094 all: 8 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 7
frame:24095 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24096 all: 9 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 8
frame:24101 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24102 all: 10 sv: 0 ev: 0 cl: 2 gm: 0 rf: 0 bk: 8
frame:24104 all: 11 sv: 0 ev: 0 cl: 0 gm: 1 rf: 0 bk: 10
frame:24105 all: 11 sv: 0 ev: 1 cl: 1 gm: 0 rf: 0 bk: 9
frame:24106 all: 20 sv: 0 ev: 0 cl: 0 gm: 0 rf: 1 bk: 19
frame:24107 all: 13 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 12
frame:24108 all: 15 sv: 0 ev: 1 cl: 1 gm: 0 rf: 0 bk: 14
frame:24109 all: 10 sv: 1 ev: 0 cl: 0 gm: 0 rf: 0 bk: 9
frame:24110 all: 12 sv: 1 ev: 0 cl: 1 gm: 0 rf: 0 bk: 10
frame:24111 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24113 all: 11 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 10
frame:24114 all: 10 sv: 0 ev: 0 cl: 0 gm: 1 rf: 0 bk: 9
frame:24116 all: 11 sv: 0 ev: 1 cl: 0 gm: 0 rf: 0 bk: 11
frame:24117 all: 11 sv: 0 ev: 0 cl: 0 gm: 1 rf: 0 bk: 10
frame:24118 all: 10 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 9
frame:24119 all: 14 sv: 0 ev: 1 cl: 0 gm: 0 rf: 0 bk: 14
frame:24120 all: 12 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 11
frame:24121 all: 11 sv: 0 ev: 0 cl: 1 gm: 0 rf: 0 bk: 10
# ON Ubuntu
\com_speed 1
# vulkan backend
frame:76190 all: 3 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 3
frame:76191 all: 3 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 3
frame:76192 all: 2 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 2
frame:76193 all: 2 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 2
frame:76194 all: 2 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 2
frame:76195 all: 3 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 3
frame:76196 all: 3 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 3
frame:76197 all: 3 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 3
frame:76198 all: 2 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 2
# Testing on ubuntu 18.04 gnome
render_vulkan: 3398 frames 15.0 seconds 225.9 fps 2.0/4.4/9.0/0.7 ms
frame:62212 all: 2 sv: 0 ev: 6 cl: 1 gm: 0 rf: 0 bk: 1
frame:62213 all: 2 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 2
frame:62214 all: 2 sv: 0 ev: 6 cl: 1 gm: 0 rf: 0 bk: 1
frame:62215 all: 2 sv: 0 ev: 6 cl: 1 gm: 0 rf: 0 bk: 1
frame:62216 all: 3 sv: 0 ev: 7 cl: 1 gm: 0 rf: 0 bk: 2
frame:62217 all: 2 sv: 0 ev: 4 cl: 0 gm: 0 rf: 1 bk: 1
frame:62218 all: 1 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 1
frame:62219 all: 2 sv: 0 ev: 7 cl: 0 gm: 0 rf: 1 bk: 1
frame:62220 all: 2 sv: 0 ev: 6 cl: 0 gm: 0 rf: 0 bk: 2
frame:62221 all: 2 sv: 0 ev: 6 cl: 0 gm: 0 rf: 1 bk: 1
frame:62222 all: 3 sv: 0 ev: 6 cl: 0 gm: 0 rf: 1 bk: 2
frame:62223 all: 2 sv: 0 ev: 5 cl: 0 gm: 0 rf: 0 bk: 2
frame:62224 all: 1 sv: 0 ev: 6 cl: 1 gm: 0 rf: 0 bk: 0
vulkan: 3398 frames 4.6 seconds 731.2 fps 1.0/1.4/3.0/0.5 ms
opengl1: 3398 frames 5.8 seconds 589.2 fps 1.0/1.7/6.0/0.6 ms
opengl2: 3398 frames 7.2 seconds 469.7 fps 1.0/2.1/21.0/0.7 ms
Q: How to turn the intensity/gamma of the drawing wondow ?
-
use r_gamma in the pulldown window, which nonlinearly correct the image before the uploading to GPU.
\r_gamma 1.5thenvid_restart -
you can also use r_intensity which turn the intensity linearly.
# 1.5 ~ 2.0 give acceptable result
$ \r_intensity 1.8
$ \vid_restart
note: original gamma setting program setting the gamma of entire destop window. which works on newer computer but not works on some machine. It is buggy and embarrasing when program abnormal quit or stall.
Q: How to check that Vulkan backend is really active ?
\vkinfobesides, the following new cmd only exist in vulkan renderer.
-
pipelineList: list the pipeline we have created;
-
gpuMem: image memmory allocated on GPU;
-
printOR: print the value of backend.or;
-
pipelineList: list the number of pipelines created (about 100, seem too much ?)
-
displayResoList: list of the display resolution you monitor supported
-
printDeviceExtensions: list the device extensions supported on you device/GPU
-
printInstanceExtensions: list the instance extensions supported on you device/GPU
-
printImgHashTable: print the image hash table usage, which also list the image creaeted and is size info etc. For example:
$ \displayResoList
Mode 0: 320x240
Mode 1: 400x300
Mode 2: 512x384
Mode 3: 640x480 (480p)
Mode 4: 800x600
Mode 5: 960x720
Mode 6: 1024x768
Mode 7: 1152x864
Mode 8: 1280x1024
Mode 9: 1600x1200
Mode 10: 2048x1536
Mode 11: 856x480
Mode 12: 1280x720 (720p)
Mode 13: 1280x768
Mode 14: 1280x800
Mode 15: 1280x960
Mode 16: 1360x768
Mode 17: 1366x768
Mode 18: 1360x1024
Mode 19: 1400x1050
Mode 20: 1400x900
Mode 21: 1600x900
Mode 22: 1680x1050
Mode 23: 1920x1080 (1080p)
Mode 24: 1920x1200
Mode 25: 1920x1440
Mode 26: 2560x1080
Mode 27: 2560x1600
Mode 28: 3840x2160 (4K)
$ \r_mode 12
$ \vid_restart
Type \vkinfo in the console reports information about active rendering backend. It will report something as following:
Active 3D API: Vulkan
Vk api version: 1.0.65
Vk driver version: 1637679104
Vk vendor id: 0x10DE (NVIDIA)
Vk device id: 0x1B80
Vk device type: DISCRETE_GPU
Vk device name: GeForce GTX 1080
Total Device Extension Supported:
...
Vk instance extensions:
...
Image chuck memory(device local) used: 8 M
You can also get the information from the UI: SETUP >> SYSTEM >> DRIVER INFO
It's non engine part of OA, includes game, cgame and ui. In mod form it is referred as OAX.
cd linux_scripts/
./supermake_native
Development documentation is located here: https://github.com/OpenArena/gamecode/wiki
The development board on the OpenArena forum: http://openarena.ws/board/index.php?board=30.0
In particular the Open Arena Expanded topic: http://openarena.ws/board/index.php?topic=1908.0
- Initial testing on Ubuntu16.04 and Ubuntu18.04, Win7 and Win10
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
- About com_hunkmegs
When i playing CTF on :F for stupid server with default com_hunkmegs = 128 setting, the following errors occurs:
ERROR: Hunk_Alloc failed on 739360: code/renderergl2/tr_model.c, line: 535 (sizeof(*v) * (md3Surf->numVerts * md3Surf->numFrames)).
OpenGL2 renderer seems use more memory, Upping com_hunkmegs to 256 will generally be OK.
- Different results compile OA without -fno-strict-aliasing I am using GCC7.2 and clang6.0 on ubuntu18.04.
Build OA using clang with -fno-strict-aliasing removed:
WARNING: light grid mismatch, l->filelen=103896, numGridPoints*8=95904
This is printed by renderergl2's R_LoadLightGrid function.
Problem solved with following line added in it.
ri.Printf( PRINT_WARNING, "s_worldData.lightGridBounds[i]=%d\n", s_worldData.lightGridBounds[i]);
However this line do nothing but just printed the value of s_worldData.lightGridBounds[i]. I guess its a bug of clang.
Build OA with GCC without this issue.
- E_AddRefEntityToScene passed a refEntity which has an origin with a NaN component
- Unpure client detected. Invalid .PK3 files referenced!
- Shader rocketThrust has a stage with no image
- merge rendergl1, rendereroa, renderer_mydev to one module
- r_gamma shader
- have issues with \minimize when use vulkan renderer in fullscreen. recreate the swapchain ?
- flare support
- Implement RB_SurfaceAxis();
- Split the program with the data.
- Use gprof to examine the performance of the program
gprof openarena.x86_64 gmon.out > report.txt
\printInstanceExtensions
----- Total 12 Instance Extension Supported -----
VK_EXT_debug_report
VK_EXT_display_surface_counter
VK_KHR_get_physical_device_properties2
VK_KHR_get_surface_capabilities2
VK_KHR_surface
VK_KHR_win32_surface
VK_KHR_device_group_creation
VK_KHR_external_fence_capabilities
VK_KHR_external_memory_capabilities
VK_KHR_external_semaphore_capabilities
VK_NV_external_memory_capabilities
VK_EXT_debug_utils
----- ------------------------------------- -----
\printDeviceExtensions
--------- Total 40 Device Extension Supported ---------
VK_KHR_swapchain
VK_KHR_16bit_storage
VK_KHR_bind_memory2
VK_KHR_dedicated_allocation
VK_KHR_descriptor_update_template
VK_KHR_device_group
VK_KHR_get_memory_requirements2
VK_KHR_image_format_list
VK_KHR_maintenance1
VK_KHR_maintenance2
VK_KHR_maintenance3
VK_KHR_multiview
VK_KHR_push_descriptor
VK_KHR_relaxed_block_layout
VK_KHR_sampler_mirror_clamp_to_edge
VK_KHR_sampler_ycbcr_conversion
VK_KHR_shader_draw_parameters
VK_KHR_storage_buffer_storage_class
VK_KHR_external_memory
VK_KHR_external_memory_win32
VK_KHR_external_semaphore
VK_KHR_external_semaphore_win32
VK_KHR_win32_keyed_mutex
VK_KHR_external_fence
VK_KHR_external_fence_win32
VK_KHR_variable_pointers
VK_KHX_device_group
VK_KHX_multiview
VK_EXT_blend_operation_advanced
VK_EXT_depth_range_unrestricted
VK_EXT_discard_rectangles
VK_EXT_shader_subgroup_ballot
VK_EXT_shader_subgroup_vote
VK_NV_dedicated_allocation
VK_NV_external_memory
VK_NV_external_memory_win32
VK_NV_glsl_shader
VK_NV_win32_keyed_mutex
VK_NVX_device_generated_commands
VK_NVX_multiview_per_view_attributes
--------- ----------------------------------- ---------
total 2 Queue families:
Queue family [0]: 16 queues, Graphic, Compute, Transfer, Sparse, presentation supported.
Queue family [1]: 1 queues, Transfer,
ubuntu 18.04 gnome Aspire v3-772G i7-4702MQ GTX760M 1920x1080
----- Total 14 Instance Extension Supported -----
VK_EXT_acquire_xlib_display
VK_EXT_debug_report
VK_EXT_direct_mode_display
VK_EXT_display_surface_counter
VK_KHR_display
VK_KHR_get_physical_device_properties2
VK_KHR_get_surface_capabilities2
VK_KHR_surface
VK_KHR_xcb_surface
VK_KHR_xlib_surface
VK_KHR_external_fence_capabilities
VK_KHR_external_memory_capabilities
VK_KHR_external_semaphore_capabilities
VK_EXT_debug_utils
----- ------------------------------------- -----
\printDeviceExtensions
--------- Total 33 Device Extension Supported ---------
VK_KHR_swapchain
VK_KHR_16bit_storage
VK_KHR_bind_memory2
VK_KHR_dedicated_allocation
VK_KHR_descriptor_update_template
VK_KHR_get_memory_requirements2
VK_KHR_image_format_list
VK_KHR_maintenance1
VK_KHR_maintenance2
VK_KHR_push_descriptor
VK_KHR_relaxed_block_layout
VK_KHR_sampler_mirror_clamp_to_edge
VK_KHR_sampler_ycbcr_conversion
VK_KHR_shader_draw_parameters
VK_KHR_storage_buffer_storage_class
VK_KHR_external_memory
VK_KHR_external_memory_fd
VK_KHR_external_semaphore
VK_KHR_external_semaphore_fd
VK_KHR_external_fence
VK_KHR_external_fence_fd
VK_KHR_variable_pointers
VK_KHX_device_group
VK_KHX_multiview
VK_EXT_depth_range_unrestricted
VK_EXT_discard_rectangles
VK_EXT_display_control
VK_EXT_shader_subgroup_ballot
VK_EXT_shader_subgroup_vote
VK_NV_dedicated_allocation
VK_NV_glsl_shader
VK_NVX_device_generated_commands
VK_NVX_multiview_per_view_attributes
--------- ----------------------------------- ---------