mirrors/looking-glass
2
0
Fork 0
mirror of https://github.com/gnif/LookingGlass.git synced 2024-12-26 23:37:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
248 commits 15 branches 37 tags 7.8 MiB
Commit graph

248 commits

This branch
This branch
All branches
Author SHA1 Message Date
Geoffrey McRae
0ad1f21ffb [client] renderer interface improvements and use gl lists for opengl 2017-12-12 03:59:58 +11:00
Geoffrey McRae
fcfea1b65d [client] fixed basic renderer 2017-12-12 03:59:58 +11:00
Geoffrey McRae
e6a21d1049 [client] implemented SGI_video_sync for better frame sync 2017-12-12 03:59:58 +11:00
Geoffrey McRae
31c9693cbf [client] move remaining opengl code into the opengl renderer 2017-12-12 03:59:58 +11:00
Geoffrey McRae
5e9885bec1 [client] apply fps limiter when presentation time is too low 2017-12-12 03:59:58 +11:00
Geoffrey McRae
786b9ec723 [client] altered makefile to link fontconfig 2017-12-12 03:59:58 +11:00
Geoffrey McRae
421b091448 [client] Added fontconfig for font path detection 2017-12-12 03:59:58 +11:00
Geoffrey McRae
a4d8fd2237 [client] added rate limiting to failed kicks (guest not ready) 2017-12-12 03:59:58 +11:00
Geoffrey McRae
3899556cbf [client] added extra sanity checks to the header's data format 2017-12-12 03:59:58 +11:00
Geoffrey McRae
81f1b43581 [client] makefile, enable -O3 as the default now 2017-12-12 03:59:58 +11:00
Geoffrey McRae
79dffa047f [client] dont mipmap when enlarging smaller screens 2017-12-12 03:59:58 +11:00
Geoffrey McRae
0ccb38940e [client] make autoResize work again 2017-12-12 03:59:58 +11:00
Geoffrey McRae
ad6c2ab544 [host] Cropping is still required on NvFBC... :( 2017-12-12 03:59:58 +11:00
Geoffrey McRae
59a894ae04 [client] switched srcRect to srcSize and type to SDL_Point 2017-12-12 03:59:57 +11:00
Geoffrey McRae
7f3657e5f0 [host] reversed the cropping logic, likely not needed anymore 2017-12-12 03:59:57 +11:00
Geoffrey McRae
a61293b551 [host] fixed support for high DPI desktops (4K+) 2017-12-12 03:59:57 +11:00
Geoffrey McRae
65ba649b36 [host/test] vertical align the text too 2017-12-12 03:59:57 +11:00
Geoffrey McRae
7d7074e9f6 [host/test] center text in the window 2017-12-12 03:59:57 +11:00
Geoffrey McRae
3336f27b5a [host] added sync-test application 2017-12-12 03:59:57 +11:00
Geoffrey McRae
34f07ec77f [client] minor logical order optimization of the render loop 2017-12-12 03:59:57 +11:00
Geoffrey McRae
b1d48943c7 [host] the client doesn't use interrupts anymore, no need to ring it 2017-12-12 03:59:57 +11:00
Geoffrey McRae
1d5f98db32 [client] adjusted presentation time to be more accurate 2017-12-12 03:59:57 +11:00
Geoffrey McRae
032602f336 [client] reworked the polling logic yet again 2017-12-12 03:59:57 +11:00
Geoffrey McRae
eb52ee9412 [client] added retard/resync counts for debugging 2017-12-12 03:59:57 +11:00
Geoffrey McRae
d09bb1146d [client] removed unused useBufferStorage app state variable 2017-12-12 03:59:57 +11:00
Geoffrey McRae
ea5d0ddfe0 [client] removed deprecated option to disable GL_ARB_buffer_storage 2017-12-12 03:59:57 +11:00
Geoffrey McRae
852309464f [client] more polling improvements 2017-12-12 03:59:57 +11:00
Geoffrey McRae
32c6989654 [client] removed debugging print 2017-12-12 03:59:57 +11:00
Geoffrey McRae
7349334811 [client] reworked polling logic and added GPU magic 
SDL_RenderPresent calls SwapBuffers which doesn't block even with vsync
enabled until the driver/gpu has buffered several frames, this is no good
for our use case where low latency is of ultimate importance. This patch
measures the swap interval, ensures the time is consumed and then calls
glFinish to ensure that there is no buffered data. The results on NVidia
hardware are exceptional, but will need verification on AMD hardware.

Polling logic has also been improved, the client will hunt for the best
delay rather then resorting to waiting on an event from the guest, avoiding
scheduler latency caused by waiting on the irq event.
 2017-12-12 03:59:57 +11:00
Geoffrey McRae
01d4d52355 [client] removed useless delay functions 2017-12-12 03:59:57 +11:00
Geoffrey McRae
e1a566bcad [client] reworked sync code to use polling for short delays 2017-12-12 03:59:57 +11:00
Geoffrey McRae
a1f574ff2f [client] added 1ms delay loop for incoming polling support 2017-12-12 03:59:57 +11:00
Geoffrey McRae
c26842cb60 [client] fixed minor issue in the makefile 2017-12-12 03:59:57 +11:00
Geoffrey McRae
51bc7631ab [host] fixed incorrect checking of difference map 2017-12-12 03:59:57 +11:00
Geoffrey McRae
2d5b633397 [client] changed ivshmem wait timeout to avoid stalls on startup 2017-12-12 03:59:57 +11:00
Geoffrey McRae
ff81f5df64 [host] removed gitmodule from testing 2017-12-12 03:59:57 +11:00
Geoffrey McRae
0b290d83d3 [host] Numerous performance improvements and minor bug fixes 2017-12-12 03:59:57 +11:00
Geoffrey McRae
c1a82e853d [client] added renderer abstratction 
This moves the bulk of the rendering code into seperate rendering
modules cleaning up much of intertwined SDL & OpenGL mess.
 2017-12-12 03:59:57 +11:00
Geoffrey McRae
5c335fca67 [client] Makefile rewrite from intial lazy version 2017-12-12 03:59:57 +11:00
Geoffrey McRae
310dba3620 [client] fixed incorrect texSize calculation 2017-12-12 03:59:57 +11:00
Geoffrey McRae
ec13fc161d [host] rename project to "Looking Glass" 2017-12-12 03:59:57 +11:00
Geoffrey McRae
10e6408022 [misc] common and client project rename to "Looking Glass" 
See: https://forum.level1techs.com/t/headless-pci-passthrough-program-name-vote/121779/36
 2017-12-12 03:59:57 +11:00
Geoffrey McRae
3156c0e0cb [client] corrected event processing 
SDL_WaitEvent is useless as it runs in 10ms increments, which is way
too slow for high rate of input applications.
 2017-12-12 03:59:57 +11:00
Geoffrey McRae
b61ba15a9c [client] fixed realignment when in server mode 2017-12-12 03:59:56 +11:00
Geoffrey McRae
62ab543720 [client] fixed incorrect kick timing 2017-12-12 03:59:56 +11:00
Geoffrey McRae
14efdf7314 [client] cleanup of vbo and texture allocation 2017-12-12 03:59:56 +11:00
Geoffrey McRae
7638925387 [client] code cleanup 2017-12-12 03:59:56 +11:00
Geoffrey McRae
a54f3000f5 [host] reworked capture to capture at the hosts's frame rate 
This will help frame delivery to the host arrive on time, as well as
allow lower latency with a higher host refresh rate.

Data offset addresses are now also pre-calculated rather then
updated on every frame.
 2017-12-12 03:59:56 +11:00
Geoffrey McRae
aa191f6c90 [client] added option to show host cursor for latency testing 2017-12-12 03:59:56 +11:00
Geoffrey McRae
d408c6c5bb [misc] added extra contributors that donated via paypal 2017-12-12 03:59:56 +11:00