niels
b009fb2819
small copy
2015-07-19 13:22:19 +02:00
niels
87dc9169e6
fix EMMS use in fast_memset (causes illegal instruction on older archs)
2015-07-19 13:10:07 +02:00
niels
ba23e10342
fix warning
2015-07-11 22:52:18 +02:00
niels
a178da1509
correctly report which memcpy function is used
2015-06-16 19:30:11 +02:00
niels
3cdd0f5e70
add sse memcpy function that copies 128 bytes per time, update with aclib fast_memcpy, review xine lib from previous commit
2015-06-16 00:02:55 +02:00
niels
0a9faf2e94
enable AVX support, update veejay -B output
2015-06-14 00:52:16 +02:00
niels
d3cc9c0c90
fix compile warnings
2015-06-14 00:07:33 +02:00
niels
7f03bf9423
fix compile warnings mostly, some bugfixes.
2015-06-13 23:54:51 +02:00
niels
b5f5132ab5
update optimized memcpy (sync with xine-lib), profile heap allocation, use runtime cpu flags to select best memcpy/memset functions, update internal benchmark test, updated info dump (-B), report selected best memset
2015-06-07 23:22:17 +02:00
niels
2ba715f9c4
replace strdup for vj_strdup (faster), fix magic tracer, clear task banks after completing multithread task
2015-06-07 00:13:51 +02:00
niels
2494d9df26
remove old stuff, fix ssize_t, align multiple of 32 bytes, fix Map B to A by threshold, fix some leaks at exit, pre-allocate memory for fx-chain to eliminate dynamic memory allocation when switching samples or manipulating the fx chain, add docs/README.memory, add commandline option to disable static fx chain, take size of static chain in account when combined with -m option, update man page, clear font worker after samples are deleted, fix possible oob write in srt get list, lock static chain to RAM (non fatal), dont clear job and task argument in multithreader (no effect)
2015-06-05 02:05:40 +02:00
niels
2a9e6c1b9f
drop veejay_strncpy, replace for strncpy
2015-05-12 23:26:04 +02:00
niels
8518d0eef2
fix CFLAGS in sendVIMS for puredata, minor updates to documentation, fixed small memory leak, removed all assertions except on libhash and lzo selftest, fix some console log messages
2015-04-23 21:21:56 +02:00
niels
0d861dfc39
plane length for frame copy function, allocate sufficient space for VJ22 and VJ20 recorders, fix audio enable/disable
2015-04-22 00:08:35 +02:00
niels
df2b5710f2
initialize debug level for benchmark, consolidate task start/stop, reduce number of mutex calls (less contention), lock then signal.
2014-12-31 17:21:46 +01:00
niels
394db6786b
added commandline option --benchmark NxN. fixes to last commit, cleanup, smaller data structures in task queue, signal wakes all threads fix
2014-12-31 02:31:16 +01:00
niels
35d54efa49
do not define HAVE_ASM_AVX (untested code)
2014-12-28 23:33:17 +01:00
niels
6e4b5cc835
benchmark threading when VEEJAY_MULTITHREAD_TASKS is set by user, fix messages, fixed uncessary mutex lock in task manager
2014-12-28 19:13:10 +01:00
niels
6b49830264
add mmx ext to yuv resampling, have mmx macros in one location, added faster yuv 4:2:2 to YUYV routine (optimizes put to SDL display), added mmxext optimized slow motion
2014-12-28 14:58:03 +01:00
niels
da30916774
add MMX optimizd version of slow motion ( alt+ a to l )
2014-12-28 12:42:44 +01:00
niels
f0a7e1df37
disable avx_memcpy for now, open avi file using mmap (default to 1% of total memory per AVI, set with VEEJAY_MMAP_PER_FILE, drop map-from-file commandline parameter, add mmap to AVI open
2014-12-28 01:33:41 +01:00
niels
3eed42c910
remove debug line
2014-12-27 23:20:53 +01:00
niels
4b6e573d4e
detect if femms is working on newer amd, add detection for avx, added avx optimized memcpy, added mmx and mmxext optimized memcpy from xine, added optmized sub and super sampling routines
2014-12-27 21:07:13 +01:00
niels
8dd3693fe7
fix cflags for 32 bit systems, rewrite slow motion (alt+a,s,d,f,g,h,j,k,l) in simpler terms for better auto vectorization with gcc, fixes to libsubsample, bump copyright year in notice, added -O3 gcc flag to enable auto vectorization for libvje,libyuv and libvjmem.
2014-12-22 23:57:10 +01:00
niels
8ff5a33007
fix more compile warnings
2014-12-21 20:57:05 +01:00
niels
c31f40fcdf
fix many compile warnings
2014-12-21 17:05:25 +01:00
niels
5ad0965a0c
disable ac_memcpy, some systems may benchmark this as fastest but it is not working properly
2014-12-20 20:40:14 +01:00
niels
e649dfb5a8
arch build fixes (ffmpeg-compat)
...
some bugfixes
removed changelog from PKGBUILD
bumped version
2014-02-01 17:45:23 +01:00
niels
708c722f30
bugfixes to a few FX, fixed multi thread task balancer,fixed removing shm segment on close, do not fail opening v4l2 device that report no input sources or channels.
2012-12-28 21:30:51 +01:00
niels
e5ac37a5b8
fix reloaded build against newer ffmpeg, add large file support (avilib still limited to 4gb), workarround bug in lzo compressor,
...
refactored all memcpy into (multithreaded) frame copy, refactored all memset into (multirheaded) frame clear, multithreaded super and sub sampling (only tested 4:2:2 ), added new mode to pencil sketch, fixed bug in lens correction, multithreaded 36 effects, updated minilzo to newer version, multhreaded yuv functions for format conversion and pixel scaling, multithreaded fx chain fader, multithreaded fx apply in libvje. Fix bug in encoder when recording in high resolution,
2012-11-25 08:29:07 +01:00
niels
59db8c5106
move multithread code to vj-task.c, added tasks for sub- and supersampling, refactor multithreaded memcpy methods
2012-11-22 23:00:40 +01:00
niels
83b70813e2
increase thread pool limit to 32, limit VEEJAY_MULTITHREAD_TASKS
...
add override norm option to dummy stream (used to always start in PAL@25fps)
2012-11-22 19:23:50 +01:00
niels
0a922175c7
fix uninitialized value of lparam, moved all dynamic structures to static. max threads is 16.
2012-11-22 00:40:32 +01:00
niels
0f8bcf1cc9
stop veejay when VEEJAY_MULTITHREAD_TASKS is too high
2012-11-21 21:29:29 +01:00
niels
d995a41240
fix segfault on 64bit machines, put limit and error checking on task_start in threadpool
2012-11-21 21:19:23 +01:00
niels
7eac3bee69
change VIMS 114, VIMS 128 and set_speed (run when playing streams). added improved support for SMP and multicore cpu. scheduling N threads over N cpus. experimental support auto enabled for resolutions above 720x576. spans off a pool of threads for video frame copy and slowmotion. lock pages into RAM, realtime priority to render thread
2012-11-20 23:10:26 +01:00
niels
846bfa1d40
merge new vj-client
...
added threadsafe gethostbyname
fix bug in reloaded set keyframes
drop vj_client_send_buf* functions
drop vj_client_setup_timeout
keep thread local buffer to store incoming frame (unicast stream)
recover gracefully when lock or unlock fails (unicast stream)
adapt poll frequency to find acceptable delay times
recover gracefully from network errors
retry to create connection when it fails while playing
destroy thread when locking functions fail
simpler get_frame for tag
dropped multicast stream for now
fix veejay_strncpy, null terminate string
drop use of bzero, use memset instead
rewrite part of cmd, socket peer-to-peer
rewrite VIMS verify message
rewrite VIMS parsing message
2011-09-11 00:48:18 +02:00
niels
aaa614fa92
drop send devices
...
fix v4l2 device list
2011-07-10 21:56:18 +02:00
Niels Elburg
c415a1b697
fixing bugs eeh
...
git-svn-id: svn://code.dyne.org/veejay/trunk@1376 eb8d1916-c9e9-0310-b8de-cf0c9472ead5
2010-02-05 21:22:32 +00:00
Niels Elburg
56d83622f0
sync
...
git-svn-id: svn://code.dyne.org/veejay/trunk@1349 eb8d1916-c9e9-0310-b8de-cf0c9472ead5
2009-11-15 16:17:20 +00:00
Niels Elburg
8546f9f897
fix instability issues in FX
...
git-svn-id: svn://code.dyne.org/veejay/trunk@1320 eb8d1916-c9e9-0310-b8de-cf0c9472ead5
2009-03-14 18:26:38 +00:00
Niels Elburg
d8e6f98d53
Initial checkin of veejay 1.4
...
git-svn-id: svn://code.dyne.org/veejay/trunk@1172 eb8d1916-c9e9-0310-b8de-cf0c9472ead5
2008-11-10 20:16:24 +00:00
