summaryrefslogtreecommitdiff
path: root/indra/newview/llglsandbox.cpp
AgeCommit message (Collapse)Author
2018-04-30MAINT-8610 Fixed GPU Benchmark appears to be broken - "Failed to allocate ↵AndreyL ProductEngine
buffer during benchmark"
2018-01-29Merged in lindenlab/viewer-bearAndreyL ProductEngine
2018-01-24MAINT-8197 Fix for crash in gpu_benchmark()AndreyL ProductEngine
including safety checks and some refactoring
2018-01-23MAINT-8197 Account for dummy tex unitandreykproductengine
2018-01-23MAINT-8197 Check buffer before using itandreykproductengine
2017-11-17MAINT-7977: Per Graham, unbind textures as well as deleting.Nat Goodspeed
2017-11-17MAINT-7977: Additional cleanup per code reviews.Nat Goodspeed
Introduce helper classes to manage paired initProfile() / finishProfile() calls and gBenchmarkProgram.bind() / unbind() calls. Make TextureHolder a class instead of a struct. Per Henri Beauchamp, since gpu_benchmark() takes a very early exit if (!gGLManager.mHasTimerQuery), subsequent tests of mHasTimerQuery are redundant. Remove. One of those tests controls the busted_finish bool, which can never become true. Remove that and all tests on it.
2017-11-16MAINT-7977: Release the LLRenderTargets when done.Nat Goodspeed
A classic-C array doesn't destroy its individual elements, but a std::vector does. Use a std::vector<LLRenderTarget> for dest, so each LLRenderTarget will be destroyed. ~LLRenderTarget() calls its release() method.
2017-11-16MAINT-7977: If getVertexStrider() returns false, abandon benchmark.Nat Goodspeed
Ruslan tracked the observed crash to assignments (to create a dummy triangle) through an LLStrider<LLVector3> obtained from getVertexStrider(). When getVertexStrider() returns false, produce a warning and just skip the rest of the benchmark test. The one bit of explicit cleanup apparently required by that early exit is a call to LLImageGL::deleteTextures() to match the preceding generateTextures() call. Wrap both in a new TextureHolder class whose destructor takes care of cleanup. The only other references to the corresponding U32 array are a couple calls to LLTexUnit::bindManual(); add a bind() method to support that. Also fix apparent bug in the LL_DARWIN special case for "improbably high and likely incorrect": the code assigned -1.f (the "couldn't compute" value) to gbps, overlooking the fact that gbps is unconditionally recomputed below. In the "likely incorrect" stanza, simply return -1.f instead.
2017-11-16MAINT-7977 [Alex Ivy] Feature Table crashesruslantproductengine
In case of buff->getVertexStrider(v) return false it mean that glMapBufferRange() return NULL The next three lines can be the reason of this crash.
2015-11-10remove execute permission from many files that should not have itOz Linden
2014-12-15merge changes for 3.7.22-releaseOz Linden
2014-12-10Rename llround(..) to ll_round(..) because of a collision with MS llround ↵callum_linden
(long long round) in VS2013
2014-12-01remove unused gpu table code, improve logging for gpu benchmarksOz Linden
2014-11-19Roll back changes to geo drawn in gpu_bench to stop punishing cards with low ↵Graham Linden
fill rate
2014-11-18Adjust gpu_bench thresholds and move VB set outside loop again for speedGraham Linden
2014-11-18Fix for gpu_bench crash without parallels knock-onGraham Linden
2014-11-17Attempt wider disable of using profile API when timer query is not presentGraham Linden
2014-11-17Avoid using GL_ARB_timer_query functionality when not supported and protect ↵Graham Linden
against probably-not-NULL current bound shader pointer
2014-09-19MAINT-3131 Discard improbably high memory bandwidth measurements on OSX and ↵davep
default to Class 3
2014-06-19Automated merge with http://bitbucket.org/lindenlab/viewer-releaseDave Parks
2014-05-07Pull and merge viewer-release as version 3.7.8simon
2014-03-24MAINT-3131 Attempt to work around drivers that lack timer query and ignore ↵Dave Parks
glFinish calls.
2014-03-21MAINT-3131 fallback to CPU timers for return value of benchmark if ↵Dave Parks
ARB_timer_query is unavailable.
2014-03-19MAINT-3131 Use benchmark to determine GPU class instead of GPU table.Dave Parks
2014-03-12merge with releaseRichard Linden
2013-12-06MAINT-3530 : Add viewer checkbox to extend parcel entry limits to a higher ↵simon
ceiling
2013-10-30Fix transplated codesimon
2013-09-10MAINT-3131 Add a GPU memory bandwidth benchmark.Dave Parks
2013-04-19merge changes for DRTVWR-294Oz Linden
2013-03-29Update Mac and Windows breakpad builds to latestGraham Madarasz
2013-03-12SH-3944 WIP CHUI merge fixingNyx Linden
re-introduced don's refactor of low-level openGL calls pulling out of llui and putting them into llrender. Took the new code from their updated versions from the CHUI merge, but put them in a place accessible to appearance utility.
2011-11-02mergeDave Parks
2011-10-19storm-1622 modify wind vector visualization to always be 25m over avatarOz Linden
2011-10-19storm-1622: fix viewer-side wind turbulence simulationOz Linden
2011-10-06SH-2240 Better fix for beacon rendering -- let LLRender take care of ↵Dave Parks
optimization around joining chunks of line segments together into one draw call
2011-10-06SH-2240 Fix for crash when rendering beacons and Debug GL enabled -- flush ↵Dave Parks
every 128 beacons to keep from hitting the end of the immediate mode wrapper buffer.
2011-10-06SH-2240 Fix for beacons not rendering (or crashing when rendered)Dave Parks
2011-09-14SH-2243 work in progress -- application side matrix stack managementDave Parks
2011-08-11SH-2181 Fix for occlusion culling going bonkers when ban lines are present.Dave Parks
2010-10-15sync up to viewer-developerTofu Buzzard
2010-10-10VWR-21967 There's lots of ::renderForSelect() code that we don't need any more.Tofu Linden
VWR-22088 Remove underlying ::renderForSelect() code (render-pipeline)
2010-09-03Post-convert merge by convert_monolith.py from ↵Aimee Linden
/Users/Aimee/Documents/Work/Linden-Lab/Development/viewer/convert/viewer-identity-evolution
2010-08-13Change license from GPL to LGPL (version 2.1)Oz Linden
2010-05-11merge from dessie/viewer-public right before SLE code landedJames Cook
2010-03-23EXT-2959 : Full out camera functions from llagent to llagentcameraLoren Shih
First check-in; only compiles, nothing more.
2010-03-08Merge with viewer 2 beta 4James Cook
2010-02-26Removed unnecessary include llhudtext.h from llviewerobject.hJames Cook
2010-02-25Support multiple fonts, colors and styles in name tagsJames Cook
Simplified LLHUDText string API to be UTF8 only Added per-text segment font support Changed LLVOAvatar invalidateNameTag to clearNameTag
2010-02-03Optimization pass.Dave Parks
-- Further instrumentation of idle update -- Removed mActiveQ from LLPipeline, instead making calm objects static during LLVOVolume::idleUpdate -- Further instrumentation of flexible object update -- Converted LLDynamicArray and LLMap members of LLViewerObjectList to stl counterparts -- Increased number of update bins in LLViewerObjectList from 16 to 128 (objects will be processed every 128 frames instead of every 16, or ~ 90% fewer objects processed per frame) -- Removed unused "renderObjectsForSelect" code. -- Reenabled LOD updates for objects further than 24m from camera
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-27 13:42:36 +0000