Cool VL Viewer forum

View unanswered posts | View active topics It is currently 2024-04-18 14:44:00



Reply to topic  [ 53 posts ]  Go to page 1, 2, 3, 4, 5, 6  Next
Feedback for the NEW experimental branch with v3.3 renderer. 
Author Message

Joined: 2012-02-10 15:12:54
Posts: 80
Reply with quote
I started this topic so people can say how this branch works for them.

I just tested it for a bit and i`m not to happy with the 3.3 renderer. Max fps now with all on low is 20 and with the old renderer it was 60-70.
Also the first version before the 3.3 renderer used to crash a lot for no reason.
My sistem:

CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 5200+ (2712.37 MHz)
Memory: 4095 MB
OS Version: Microsoft Windows 7 64-bit Service Pack 1 (Build 7601)
Graphics Card Vendor: ATI Technologies Inc.
Graphics Card: ATI Radeon HD 3600 Series
Windows Graphics Driver Version: 8.17.0010.1124
OpenGL Version: 3.3.11631 Compatibility Profile Context

This is just the feedback that you asked for.
Please keep v1.26.4 maintained because not all of us have new CPU & GPU.


2012-07-29 19:55:48
Profile

Joined: 2009-03-17 18:42:51
Posts: 5545
Reply with quote
TheADX wrote:
Also the first version before the 3.3 renderer used to crash a lot for no reason.
If you get crashes, please open a topic in the Bug reports forum, following that forum's rules and providing all relevant logs and repro steps.
Please also note that v1.26.4.23 will actually include all the changes that went into v1.26.5.0 because I need to switch to the new build systems used by LL (i.e. glibc 2.7 or later and VS2010) so to have a chance to be able to keep maintaining v1.26.4 in the future. So, if you crash with v1.26.5.0, better telling and properly reporting the bugs so that I can fix them !


2012-07-29 20:32:32
Profile WWW

Joined: 2011-08-27 17:31:05
Posts: 98
Reply with quote
Here is my experience with v1.26.5.1 on Windows 7 64-bit:

At first I had problems enabling deferred rendering. Took me a while to find out that I not only needed to enable the checkbox for it on the graphics tab, but also needed to move the slider to Ultra first. After I found that out, it worked fine.

I wanted to compare FPS with v1.26.4.22, but I had also just updated to the latest Nvidia driver (from 285.62 to 301.42), and now v1.26.4.22 will only give me about 1 (!) FPS with deferred rendering. I *think* deferred rendering seems to work more smoothly now. I think I was getting about 15 FPS with v1.26.4.22 before updating the driver, while I now get 25-30 FPS with v1.26.5.1. This is with an Nvidia 580 on a 30" monitor and the viewer running in a window with about 2440x1500 pixel.

I also got a crash of v1.26.5.1 at one point, but I forgot to keep the logfile. If it happens again I promise to remember to keep and post the logfile. :D


2012-07-29 20:41:05
Profile

Joined: 2009-03-17 18:42:51
Posts: 5545
Reply with quote
Nicolette Lefevre wrote:
Here is my experience with v1.26.5.1 on Windows 7 64-bit:

At first I had problems enabling deferred rendering. Took me a while to find out that I not only needed to enable the checkbox for it on the graphics tab, but also needed to move the slider to Ultra first. After I found that out, it worked fine.
What kind of troubles did you have, exactly ?... As long as the Atmospheric shaders (and all lower shaders, i.e. "Bump, Mapping & Shiny" and "Basic shaders") are not enabled, the "Deferred rendering" checkbox stays grey (which is normal and simply indicates you need to enable the proper shaders first); as soon as all shaders are enabled, you just have to check the "Deferred Rendering" box and "Apply" the graphics settings.

Quote:
I wanted to compare FPS with v1.26.4.22, but I had also just updated to the latest Nvidia driver (from 285.62 to 301.42), and now v1.26.4.22 will only give me about 1 (!) FPS with deferred rendering.
Wow !... It looks like the new drivers are pretty f*cked up then... Strange, because here, under Linux and with 304.22 (beta driver) I get the best frame rates I ever had with my GTX460 (in non-deferred mode, my best case scenario FPS rate even jumped from 130 to 170fps with v1.26.4).

Quote:
I *think* deferred rendering seems to work more smoothly now. I think I was getting about 15 FPS with v1.26.4.22 before updating the driver, while I now get 25-30 FPS with v1.26.5.1. This is with an Nvidia 580 on a 30" monitor and the viewer running in a window with about 2440x1500 pixel.
Impressive figure for such high a resolution... What about non-deferred mode ?

Quote:
I also got a crash of v1.26.5.1 at one point, but I forgot to keep the logfile. If it happens again I promise to remember to keep and post the logfile. :D
Yes, please, do report crashes !


2012-07-29 20:52:55
Profile WWW

Joined: 2011-08-27 17:31:05
Posts: 98
Reply with quote
Henri Beauchamp wrote:
What kind of troubles did you have, exactly ?... As long as the Atmospheric shaders (and all lower shaders, i.e. "Bump, Mapping & Shiny" and "Basic shaders") are not enabled, the "Deferred rendering" checkbox stays grey (which is normal and simply indicates you need to enable the proper shaders first); as soon as all shaders are enabled, you just have to check the "Deferred Rendering" box and "Apply" the graphics settings.


The problem was that by checking deferred-rendering and clicking Apply very little happened. There was a short drop in framerate, but that went back to normal almost instantly. I did notice some difference in the rendering though. Invisiprims stopped working and the anti-aliasing looked slightly different. But there were no shadows. None. Once I moved the slider to Ultra the shadows were immediately there.

Henri Beauchamp wrote:
Wow !... It looks like the new drivers are pretty f*cked up then... Strange, because here, under Linux and with 304.22 (beta driver) I get the best frame rates I ever had with my GTX460 (in non-deferred mode, my best case scenario FPS rate even jumped from 130 to 170fps with v1.26.4).


The slowdown is now happening in v1.26.5.1 too. GPU-Z shows that the driver has set the core-clock of the graphics-card to very low and won't speed up again. Time to install a different driver. And then check again...

Henri Beauchamp wrote:
Impressive figure for such high a resolution... What about non-deferred mode ?


It was about 80-85 fps at the same place.

I will try again later and post updated figures once I have a working graphics driver again.


2012-07-29 22:38:37
Profile

Joined: 2011-08-27 17:31:05
Posts: 98
Reply with quote
Ok. The driver was causing the slowdown in v1.26.5.1 that I saw later on. The clock-rate of the GPU was fixed at a very low setting. 50MHz instead of the normal 772MHz. I'm using driver version 304.79 Beta now. Both v1.26.4.22 and v1.26.5.1 work fine with that one.

The low framerate with deferred rendering in v1.26.4.22 was not caused by the driver, but by some borked debug settings that I had changed. With the new driver, clean install and deleted settings-file it is now back to normal.

v1.26.4.22:
Non-deferred 120fps
Deferred 45 fps (only 4.9 (!) before the clean install)

v1.26.5.1:
Non-deferred 95fps
Deferred 48fps

So the new renderer is noticeably slower with non-deferred rendering, but slightly faster with deferred rendering.

I just wish it would be easier to find a good-working Nvidia driver... :D


2012-07-29 23:21:56
Profile

Joined: 2009-03-17 18:42:51
Posts: 5545
Reply with quote
Nicolette Lefevre wrote:
So the new renderer is noticeably slower with non-deferred rendering, but slightly faster with deferred rendering.
Yes, this confirms my own experience, the question being: "is the deferred render speed gain worth the non-deferred render speed loss ?"... At 120+ fps, it's not much of an issue, but more testing needs to be done in heavy environments (with loads of avatars and complex objects) and then see how v3.3 fares when compared to v2.6, both being in non-deferred render mode...


2012-07-30 00:18:40
Profile WWW

Joined: 2012-02-10 15:12:54
Posts: 80
Reply with quote
Henri Beauchamp wrote:
Nicolette Lefevre wrote:
So the new renderer is noticeably slower with non-deferred rendering, but slightly faster with deferred rendering.
Yes, this confirms my own experience, the question being: "is the deferred render speed gain worth the non-deferred render speed loss ?"... At 120+ fps, it's not much of an issue, but more testing needs to be done in heavy environments (with loads of avatars and complex objects) and then see how v3.3 fares when compared to v2.6, both being in non-deferred render mode...



I will help you with the low end side of things... i`l keep testing it and i will post here the debug for every crash i have. Also... can u make a buton with "copy log to clipboard" or "open log file"? Or some kind of way that will make posting logs easy.

btw i noticed something strange on the v1.26.4 version (and i think it dose it on the new one also) , when i login its full power fps like avrage 30 but after some time of use like 2-3 hours it gets slow and frames drop to like 10 or even in a 8h scenario it droped to 5fps. I hang out in info hubs with lots of people around so i keep my avatar impostors to 1 and also basic shaders are disabled. Is it a problem with the GPU memory that gets full and stays like that? I`l just post a log from just a short login and tp`s from sim to sim so you can see what i get and also the drop in frames.

http://pastebin.com/77NeNKFL


2012-07-30 06:40:20
Profile

Joined: 2011-09-27 11:18:31
Posts: 172
Reply with quote
Well, I've got my usual crash when the viewer tries to load shaders on an Intel card. I don't expect you can do much about it, because Intel isn't supported on Linux, which (is I suppose) why no viewer3 works on my system, but for sake of completeness I am filing the report.

System information


Cool VL Viewer 1.26.5 (1) Jul 29 2012 17:27:49 (Cool VL Viewer)
Release Notes

CPU: Intel(R) Celeron(R) CPU E3400 @ 2.60GHz (2592.92 MHz)
Memory: 1878 MB
OS Version: Linux 3.4.6-1-ARCH #1 SMP PREEMPT Fri Jul 20 08:21:26 CEST 2012 x86_64
Graphics Card Vendor: Tungsten Graphics, Inc
Graphics Card: Mesa DRI Intel(R) G41 x86/MMX/SSE2
OpenGL Version: 2.1 Mesa 8.0.4

libcurl Version: libcurl/7.21.1 OpenSSL/1.0.0d zlib/1.2.5 c-ares/1.7.1
J2C Decoder Version: KDU
Audio Driver Version: FMOD version 3.750000
Qt Webkit Version: 4.7.1 (version number hard-coded)

Built with GCC version 40302

Compile flags used for this build:
-O2 -fomit-frame-pointer -frename-registers -fweb -ftree-vectorize -fexpensive-optimizations -DLL_RELEASE=1 -DLL_RELEASE_FOR_DOWNLOAD=1 -DLL_SEND_CRASH_REPORTS=1 -DNDEBUG -Wall -Wno-sign-compare -Wno-trigraphs -Werror -Wno-reorder -Wno-non-virtual-dtor -Wno-deprecated -g -fexceptions -fno-strict-aliasing -fvisibility=hidden -fsigned-char -march=pentium4 -msse2 -mfpmath=sse -fno-math-errno -fno-trapping-math -pthread -fno-stack-protector -m32 -D_FORTIFY_SOURCE=2 -DLL_USE_TCMALLOC=1 -DLL_LINUX=1 -D_REENTRANT -DAPPID=secondlife -DLL_IGNORE_SIGCHLD -DLL_DBUS_ENABLED=1 -DLL_ELFBIN=1 -DOV_EXCLUDE_STATIC_CALLBACKS -DCARES_STATICLIB -DLL_SDL=1 -DLIB_NDOF=1 -DLL_GTK=1 -DLL_X11=1


Stacktrace
#0 0xf7fdb430 in __kernel_vsyscall ()
#1 0xf685888f in raise () from /usr/lib32/libc.so.6
#2 0xf685a173 in abort () from /usr/lib32/libc.so.6
#3 0xf196f154 in ir_validate::visit(ir_dereference_variable*) ()
from /usr/lib32/libglsl.so
#4 0xf19698e5 in ir_dereference_variable::accept(ir_hierarchical_visitor*) ()
from /usr/lib32/libglsl.so
#5 0xf1969383 in visit_list_elements(ir_hierarchical_visitor*, exec_list*, bool) () from /usr/lib32/libglsl.so
#6 0xf1969b16 in ir_call::accept(ir_hierarchical_visitor*) ()
from /usr/lib32/libglsl.so
#7 0xf1969a54 in ir_assignment::accept(ir_hierarchical_visitor*) ()
from /usr/lib32/libglsl.so
#8 0xf1969383 in visit_list_elements(ir_hierarchical_visitor*, exec_list*, bool) () from /usr/lib32/libglsl.so
#9 0xf1969544 in ir_function_signature::accept(ir_hierarchical_visitor*) ()
from /usr/lib32/libglsl.so
#10 0xf1969383 in visit_list_elements(ir_hierarchical_visitor*, exec_list*, bool) () from /usr/lib32/libglsl.so
#11 0xf19695bc in ir_function::accept(ir_hierarchical_visitor*) ()
from /usr/lib32/libglsl.so
#12 0xf1969383 in visit_list_elements(ir_hierarchical_visitor*, exec_list*, bool) () from /usr/lib32/libglsl.so
---Type <return> to continue, or q <return> to quit---
#13 0xf19692d9 in ir_hierarchical_visitor::run(exec_list*) ()
from /usr/lib32/libglsl.so
#14 0xf196f4ea in validate_ir_tree(exec_list*) () from /usr/lib32/libglsl.so
#15 0xf1bfe304 in _mesa_glsl_compile_shader () from /usr/lib32/libdricore.so
#16 0xf1ae24e0 in ?? () from /usr/lib32/libdricore.so
#17 0xf1ae2507 in _mesa_CompileShaderARB () from /usr/lib32/libdricore.so
#18 0x08d616d7 in ?? ()
#19 0x089c20a9 in ?? ()
#20 0x089c6da3 in ?? ()
#21 0x087ca0f7 in ?? ()
#22 0x0817d13d in ?? ()
#23 0x0817ecf7 in ?? ()
#24 0x08b8f436 in ?? ()
#25 0xf6843605 in __libc_start_main () from /usr/lib32/libc.so.6
#26 0x0810a921 in _start ()


and the logs are attached.


EDIT: Ah, by the way, at least on my system, the viewer takes a few moments to start. That is, the main window opens instantanously, but it takes some minutes to draw. It's not important, just a little annoying and I thought I'd report it.


Attachments:
logs.tar.bz2 [8.88 KiB]
Downloaded 209 times
2012-07-30 12:35:06
Profile

Joined: 2012-02-10 15:12:54
Posts: 80
Reply with quote
Just had a new crash on v1.26.5.1 after i loged in and tp`d to a new sim: here is the log.

http://pastebin.com/3pzk3NJ0


/later edit: the textures do a strange thing... they randomly become blurry like they try to load again every few seconds, so i opened the live debug log on my screen and it happens when the LLMemory info apears.

Example:

2012-07-30T18:31:51Z INFO: LLMemory::logMemoryInfo: --- Private memory pool information --
2012-07-30T18:31:51Z INFO: LLMemory::logMemoryInfo: Total reserved (KB): 0
2012-07-30T18:31:51Z INFO: LLMemory::logMemoryInfo: Total allocated (KB): 0
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Current max physical memory (KB): 1524740
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Current allocated physical memory (KB): 1243988
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Current allocated virtual memory (KB): 1639136
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Current available physical memory (KB): 1524740
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: --- Private memory pool information --
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Total reserved (KB): 0
2012-07-30T18:31:55Z INFO: LLMemory::logMemoryInfo: Total allocated (KB): 0


2012-07-30 17:55:44
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 53 posts ]  Go to page 1, 2, 3, 4, 5, 6  Next

Who is online

Users browsing this forum: No registered users and 41 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group
Designed by ST Software.