OpenGL/Elite3D Performance in Solaris 10?

Hi Folks,
I've searched but can't seem to find anything on this.
I have an Ultra-2 with 2x300MHz, 640MB of RAM, and an Elite3d-M6 framebuffer. Life is good, and Solaris 10 is great.
But - I use a brain modeling application that converts MRI images of the skull to 3d maps of the brain. Under Solaris 9, performance was awesome - fully accelerated, very smooth rotation and scrolling of the 3d models.
Using the same software under solaris 10, its graphics performance is very poor. I've also noticed that the OpenGL plugin for Xmms (which ran great under Solaris 9) performs terribly in Solaris 10. The problem is the same in both CDE and JDS.
To me, this says 'OpenGL' problem. But I've done a full Solaris 10 install (and several re-installs, for other reasons), and performance is equally poor with all versions. OpenGL really seems to be installed and working properly, but performance is really bad. Does anyone have any ideas about this one?
Thanks,
tim

I found out about Update 4 very shortly after posting this, and I have now upgraded. Unfortunately, the problem still persists.

Similar Messages

Poor Elite3D OpenGL performance in Solaris 10

I'm running an old Ultra10 with an Elite3D-m3. Since I installed Solaris 10, I've had a very slow desktop and awful performance in 3D applications. This tells me that there's something wrong between the video card and OpenGL, like it's in software graphics mode. How can I tell if I'm actually running in software OpenGL or if it's the proper performance for the card? Do I need special drivers to run it since the card is so old?

I found out about Update 4 very shortly after posting this, and I have now upgraded. Unfortunately, the problem still persists.

OpenGL and GLX on Solaris Intel

Can OpenGL and GLX be added to Solaris Intel without having to use Xfree86?
I have built the Mesa libraries but still need the GLX extensions.
Does any one know of an easy solution?

Hi Senthilkumar,
Thanks for reply!
I found a couple of intel 10/100 network cards in our town, there are INTEL N232 10/100 PCI, INTEL PRO100+ 10/100 PCI, Intel 8460 10/100BT, I am not sure which one you were talking about, can you give me some detail information please?
Thanks again.
Mark

NFS performance with Solaris 10

Hello,
We have been playing with one of the x4200s running s10u2, or snv_50 for that matter, and are getting terrible numbers from the NFS performance. Initially, we suspected it was just the ZFS filesystem on the back (which it was, though zil_disable made it a lot better), but even after exploring a little I am getting terrible numbers for NFS backed by UFS. Using afio to unafio a file on the disk gives:
Local:
afio: 432m+131k+843 bytes read in 263 seconds. The operation was successful.
Remote:
afio: 432m+131k+843 bytes read in 1670 seconds. The operation was
successful.
I have raised the ncsize to 1000000, and upped the server threads to 1024.
The same thing on a linux box(ext3) turns in local times of 100 seconds and remote at 180 seconds. The differences in the local and remote numbers are just crazy. The difference in the ZFS is way worse:
Local zfs:
afio: 432m+131k+843 bytes read in 137 seconds. The operation was successful.
NFS -> ZFS:
afio: 432m+131k+843 bytes read in 2428 seconds. The operation was
successful.
I have started looking into dtrace for tracking the problem, but don't have much to report yet.
Any suggestions appreciated.

Ask this on the Solaris Forum, not the Java Networking forum.
Edit: typo

Performance on Solaris 10 - Operating system paging

Has anyone experience performance issues after an upgrade from Solaris 9 to Solaris 10 on SAP systems with limited memory?
We have many systems that are on servers with 4 Gig of memory and ran well on Solaris 9. After an upgrade to Solaris 10 we are experiencing very high OS system paging rates. The response times of the SAP systems are very poor when this occurs. It seems to take very little load to cause this.
I realize more memory or decreases in Oracle or SAP memory parms will solve this but am wondering if there is anything on the Solaris OS that could resolve this?
Thanks,
Dan

DISM can be used but in global zone only (according to Sun document "Best Practive for Running Database in Solaris Containers" , the proc_lock_memory privilege which is required to run the ora_dism_ process is not available in non globale zone)
The doc i got is from 2005, so don't know if the Sun recommendations has been updated since then.
In order to activate DISM (if you are in a global zone), sga_max_size should be set up larger than the sum of sga components: db_cache_size, shared_pool_size ...)
Also look for the Sun Blueprint "Dynamic Reconfiguration and Oracle 9i Dynamic Resizable SGA" on http://www.sun.com/blueprints
If you use ISM because in a non-global zone, you can use oracle parameter lock_sga to ensure the SGA is loaded into the RAM and useism_for_pga = true to ensure PGA is loaded into the RAM.
Make sure you have enough RAM to hold filesystem cache (OS memory) , oracle memory, and applications memory
Make sur your PGA and SGA are correctly sized size, since you won't be able to dynamically change the ISM allocation. (see v$shared_pool_advice, v$db_cache_advice, v$pga_target_advice ...)
Take the usual precautions:
- have a successfull backup first
- do the change on a test machine
- and/or ask your vendor before proceeding
Other Doc to read ...
Note 697483 - "Oracle Dynamic SGA on Solaris" (recommends to read Sun doc n°230653)
Note 724713 - parameter settings for Solaris 10, here is an extract :
Only one parameter from SAP note 395438 should remain in file
etc/system
set rlim_fd_cur=8192
As described in SunSolve document 215536, the "Large Page Out Of the Box" (LPOOB) feature of the Solaris 10 memory management, first implemented in Solaris 10 1/06 (Update 1), can lead to performance problems when the system runs out of large chunks of free contiguous memory. This is because the Solaris 10 kernel by default will attempt to relocate memory pages to free up space for creating larger blocks of contiguous memory. Known symptoms are high %system CPU time in vmstat, high number of cross calls in mpstat, and Oracle calling mmap(2) to /dev/zero for getting more memory.
Memory page relocation for satisfying large page allocation requests can be disabled by setting the following Solaris kernel parameter in /etc/system
set pg_contig_disable=1
This will not switch off the LPOOB feature. Large memory pages will still be used when enough free space of contiguous memory is available, so the benefits of this feature will remain
Note 870652 - Installation of SAP in a Solaris Zone
Note 1246022 - Support for SAP applications in Solaris Zones
Edited by: Emmanuel TCHENG on Oct 13, 2009 12:02 PM

Poor I/O Performance on Solaris - v1.4.1_01

Does anyone have any comments on the following?
It's an I/O analysis done to determine which Java
methods might be used to replace an existing C++
platform-specific file re-compression subsystem.
The system has to handle up to 200,000 files per
day (every day).
Java IO test results for converting ZERO_ONE compressed
files to standard compressed files.
Java 1.4.1, 12-04-2002
The input dataset contains 623,230,991 bytes in 1391 files.
The input files are in ZERO_ONE compression format.
For all tests:
1) an input data file was opened in buffered mode.
2) the data was read from the input and expanded
(byte by byte).
3) the expanded data was written to a compressing
output stream as it was created.
4) repeat 1 thru 3 for each file.
64K buffers were used for all input and output streams.
Note: Items marked with "**" hang at random on Solaris
(2.7 & 2.8) when processing a large number of small files. They always hang on BufferedInputStream.read().
There may be a deadlock situation with the 'process
reaper' because we're calling 'exec()' and 'waitFor()'
so quickly. The elapsed times for those items are
estimates based on the volume of data processed up to
the point where the process hung. This 'bug' has been
reported to Sun.
-- elapsed time --
NT Solaris 2.7 Method
n/a 18 min Current C++ code:
fopen(r) -> system(compress)
19 min 19 min ** BufferedInputStream -> exec(compress)
29 min 21 min 1) BufferdInputStream -> file
2) exec(compress file)
24 min 42 min ** BufferedInputStream -> exec(gzip)
77 min 136 min BufferedInputStream -> GZIPOutputStream
77 min -- BufferedInputStream -> ZipOutputStream
The performance of GZIPOutputStream and ZipOutputStream
makes them useless for any Production system. The 2x
performance degradation on Solaris (vs NT) for these two
streams is surprising. Does this imply that the 'libz'
on Solaris is flawed? Notice that whenever 'libz' is
involved in the process stream (exec(gzip),
GZIPOutputStream, ZipOutputStream) the elapsed time
climbs dramatically on Solaris.

Re-submitted Performance Matrix with formatting retained.
Note: for the "-> system()" and "-> exec()" methods, we write to the
STDIN of the spawned process.
-- elapsed time --
NT     Solaris 2.7 Method
n/a    18 min      Current Solaris C++ code:
                     fopen(r) -> system("compress -c >aFile")
19 min 19 min **   BufferedInputStream -> exec("compress -c >aFile")
29 min 21 min      1) BufferdInputStream -> "aFile"
                   2) exec("compress aFile")
24 min 42 min **   BufferedInputStream -> exec("gzip -c aFile")
77 min 136 min     BufferedInputStream -> GZIPOutputStream("aFile")
77 min --          BufferedInputStream -> ZipOutputStream("aFile")

Java2d, Java3d or Opengl best performance

I am planning to write a game in java that must be multi-platform and must have a very good performance. I am making a kind of rts. This rts will only be 2d. I am planning to get as many moving little objects(pics) within a second as i can. What is the best method in doing so. Performance wise and implementing wise. I never worked with Java2d or Java3d and Opengl only with c++. So what is the best choice you reckon?
Also every performance/programming tip is welcom :-)

I would definitly go with java2D. Its really easy to use and get started with. OpenGL in c++ is the best way to go for profesional games, but java can be fast engough for most simple stuff.

OpenGL xscreensaver hacks for Solaris 10

Okay, kind of a silly question... I'm trying to set up a machine to demo Solaris 10 x86 on a Sun Opteron w2100z, and one of the easier ways to attract attention to the machine would be to run GL screensavers. I've successfully installed the nVidia drivers, but I've discovered that the executables for the GL screensavers (atlantis, glplanet, glsnake, etc.) are not present in /usr/openwin/lib/xscreensaver/hacks. Are these executables somewhere on the Solaris 10 x86 media, and I just missed installing them? Or do I need to download them from somewhere, and if so, where? Any help would be appreciated. Thanks!

I don't know the direct answer to your question, but I can tell you the performance will be poor on your W2100z [as it was on mine] without the following patch:
http://groups.yahoo.com/group/solarisx86/message/25071
That workaround is a few months old, and it may have been in a smpatch update, I don't know. But if your performance isn't AGP, use the patch at the link above.
Also, if you join the group at the above link, there are some very talented Solaris graphics guys in that group, John Martin being one.

WebLogic 6.0 performance on Solaris (Excessive polling)

We have a problem with SunOs 5.8, multi (2) CPU, 1GB memory Weblogic 6.0 sp1 and
java 1.3.0.
While running some benchmarks we realized that the tests run much faster (25%-40%
faster) on a NT Pentium III 800 MHz desktop with 256MB RAM than on our Sun Ultra-250
with the above configuration.
So, we decided to do some profiling by turning on hprof (-X:runhprof:cpu=samples,depth=12)
and realized that over 50% of the time is spent in polling the sockets (weblogic.socket.PosixSocketMuxer.poll).
On NT, only 0.23% time is spent on the same activity. I do not know if on NT WebLogic
uses a different IPC model (select vs. interrupt) but the time taken on Solaris
doing polls is outrageous. I am wondering if anybody else out there has experienced
the same problem? Any insights you have to share on this will be appreciated.
TIA,
-Anil Singhal

The Xprof output is misleading - this does not mean that 50% of
the CPU is spent on polling sockets. It simply means that 50% of the thread samples
were in a polling method (probably waiting, not running).
Yes, I believe that an 800MHz PIII would blow away an ES-250 which runs at less
that 300MHz. So you have 2CPU -> 600MHz. Then
(800-600)/800 = 25% which is what you get.
Mike
"Michael Girdley" <----> wrote:
Are you using the performance pack? If not, you should.
Michael Girdley
BEA Systems
Learning WebLogic? http://learnweblogic.com
"Anil Singhal" <[email protected]> wrote in message
news:3adc4f27$[email protected]..
We have a problem with SunOs 5.8, multi (2) CPU, 1GB memory Weblogic6.0
sp1 and
java 1.3.0.
While running some benchmarks we realized that the tests run much faster(25%-40%
faster) on a NT Pentium III 800 MHz desktop with 256MB RAM than onour Sun
Ultra-250
with the above configuration.
So, we decided to do some profiling by turning on hprof(-X:runhprof:cpu=samples,depth=12)
and realized that over 50% of the time is spent in polling the sockets(weblogic.socket.PosixSocketMuxer.poll).
On NT, only 0.23% time is spent on the same activity. I do not knowif on
NT WebLogic
uses a different IPC model (select vs. interrupt) but the time takenon
Solaris
doing polls is outrageous. I am wondering if anybody else out therehas
experienced
the same problem? Any insights you have to share on this will beappreciated.
TIA,
-Anil Singhal

WebLogic 6.0 performance on Solaris vs NT

Has anybody tried comparing the performance of WebLogic 6.0 sp1 on Solaris (SunOS
5.8) with NT 4.0? Please post your experiences.
TIA,
-Anil

The Xprof output is misleading - this does not mean that 50% of
the CPU is spent on polling sockets. It simply means that 50% of the thread samples
were in a polling method (probably waiting, not running).
Yes, I believe that an 800MHz PIII would blow away an ES-250 which runs at less
that 300MHz. So you have 2CPU -> 600MHz. Then
(800-600)/800 = 25% which is what you get.
Mike
"Michael Girdley" <----> wrote:
Are you using the performance pack? If not, you should.
Michael Girdley
BEA Systems
Learning WebLogic? http://learnweblogic.com
"Anil Singhal" <[email protected]> wrote in message
news:3adc4f27$[email protected]..
We have a problem with SunOs 5.8, multi (2) CPU, 1GB memory Weblogic6.0
sp1 and
java 1.3.0.
While running some benchmarks we realized that the tests run much faster(25%-40%
faster) on a NT Pentium III 800 MHz desktop with 256MB RAM than onour Sun
Ultra-250
with the above configuration.
So, we decided to do some profiling by turning on hprof(-X:runhprof:cpu=samples,depth=12)
and realized that over 50% of the time is spent in polling the sockets(weblogic.socket.PosixSocketMuxer.poll).
On NT, only 0.23% time is spent on the same activity. I do not knowif on
NT WebLogic
uses a different IPC model (select vs. interrupt) but the time takenon
Solaris
doing polls is outrageous. I am wondering if anybody else out therehas
experienced
the same problem? Any insights you have to share on this will beappreciated.
TIA,
-Anil Singhal

Mysql performance on solaris

We have planned to run a mysql database on a SUN SPARC box with Solaris 9, but compared to a small linux box the performance seems to be bad. Our test is with only one user and no additional load on the machines. The Linux box is about 5 times faster (response time ) than the SUN box. My questions are:
- will it get better on solaris10??
- can I tune mysql or Solaris for a better performance??

We have planned to run a mysql database on a SUN SPARC box with Solaris 9, but compared to a small linux box the performance seems to be bad. Our test is with only one user and no additional load on the machines. The Linux box is about 5 times faster (response time ) than the SUN box. My questions are:
- will it get better on solaris10??
- can I tune mysql or Solaris for a better performance??

ORACLE 10g PERFORMANCE ON SOLARIS 10

Hi all,
In sunfire v890 we have installed oracle 10g release 2 on solaris 10.
prstat -a command shows :
NPROC USERNAME SIZE RSS MEMORY TIME CPU
105 root 9268M 6324M 20% 1:21:57 0.4%
59 oracle 24G 22G 71% 0:04:33 0.1%
2 nobody4 84M 69M 0.2% 0:11:32 0.0%
2 esbuser 13M 9000K 0.0% 0:00:46 0.0%
1 smmsp 7560K 1944K 0.0% 0:00:00 0.0%
4 daemon 12M 7976K 0.0% 0:00:00 0.0%
and top utility shows :
last pid: 8639; load avg: 0.09, 0.09, 0.09; up 2+06:05:29 17:07:50
171 processes: 170 sleeping, 1 on cpu
CPU states: 98.7% idle, 0.7% user, 0.7% kernel, 0.0% iowait, 0.0% swap
Memory: 32G phys mem, 22G free mem, 31G swap, 31G free swap
therefore from prstat we come to know that memory used by oracle is 71%
where as top says 31.25% used.....
which one is true in this scenario.....
shall we go ahead in trusting top utility????
Advance thanks to you.

therefore from prstat we come to know that memory
used by oracle is 71%
where as top says 31.25% used.....
which one is true in this scenario.....
shall we go ahead in trusting top utility????In this case top is more accurate. prstat pretends all the memory used by each Oracle process is used only by that process. But lots of the memory used by Oracle is shared between several processes. prstat is counting that shared memory over and over for each process... resulting in the higher figure.
http://forum.java.sun.com/thread.jspa?threadID=5114263&tstart=105
Regards,
[email protected]
http://www.HalcyonInc.com

Java 2D OpenGL pipe performance in JDK1.5

I've done some tests with OpenGL pipe in JDK1.5 and the results were not that statisfying...
I downloaded the the J2DBench source code from the sun site and runned the tests. The results:
Summary:
Rendering - Default ppl:
    Number of tests: 226
    Overall average: 60947.21776458781
    Best spread:      0.0% variance
    Worst spread:     2119.32% variance
    (Basis for results comparison)
Rendering - OpenGL:
    Number of tests: 226
    Overall average: 90528.77058480236
    Best spread:      0.0% variance
    Worst spread:     21.97% variance
    Comparison to basis:
      Best result:      15936.15% of basis
      Worst result:     14.68% of basis
      Number of wins:   154
      Number of ties:   0
      Number of losses: 72If I interprete these results correctly this means that the OpenGl pipe is 30% slower than without using the pipe.
Am I misunderstanding these results?
Are there know perfromance issues with the OpenGL pipe?
For the completeness: Test environment:
Processor: Intel Pentium 4 CPU 2.8GHz
Memory: 2048 DDR RAM
Windows XP SP2
NVidia GeForce FX 5200 (+ latest driver from NVidia site)
DirectX 9.0c
BR,
Freggy

Java2Demo does many things which are hard for the OpenGL pipeline to optimize and therefor is maybe not the optimal benchmark.
But in general the improvements made in java-6.0 were significant.
lg Clemens

Perl performance on solaris 10

hi all,
i executed this (http://www.metacard.com/perlbench.html) script on solaris 10 and linux machine. Solaris box took around 15sec to complete the test where as linux box took only 8 seconds.
Is there any way to increase the perfomance of perl on solaris 10.
Regards,
uttam

Yes, I doubt this has much to do with the OS unless they are exactly the same machines with equally good device support.
ultra5 333mhz ultrasparc 2gb ram sol10= 45 seconds
2x750 ultrasparc 2gb ram blade1000 sol10= 22 seconds
3200 amd 1gb ram fedora= 18 seconds
ultra20m2 1.8 ghz opteron 2gb ram sol10= 9 seconds
x2200m2 2x2.8ghz opterons 12gb ram sol10= 6 seconds
I like how the ancient blade1000 is not to far from the much more modern fedora box. Such a lovely machine.

PHP in Solaris 10 and Non-Global Zones: Problem of performance?

Hi friends
We are feeling a poor performance with applications developed with PHP in Solaris 10, with non-global and global zones, while Intel platform (Xeon and Pentium), performance is very good. Difference between both platforms is about 200% aprox, one second in Intel to 9, 12 or 20 seconds in Solaris depending of model.
Our tests were developed in:
1. SF T2000 server Solaris 10 global zone
2. SF T2000 server Solaris 10 non-global zone
3. SF280R server Solaris 10 non-global zone
4. V240 server with 1 GB memory, 1*US III-i 1.0 GHz and Solaris 9 (really this version for test and comparisons)
5. V240 server with 8GB memory, 2*US III-i 1.5Ghz and Solaris 9 (really this version for test and comparisons too)
Intel platforms were:
1. Intel Pentium 4 2GHz 2GB memory, Linux Fedora and PHP 4.4.4
2. Intel Xeon 2 core, 2.33GHz 2GB memory, Linux Fedora and PHP 4.4.3
Versions of products are:
1. Solaris 9 or Solaris 10
2. PHP 4.4.7 downloaded from http://www.php.net/downloads.php
3. Apache 2.0.59
4. MySQL 4.1.15-log
Our php compilation and installation were:
./configure --prefix=/usr/local/php-4.4.7 \
--with-pear \
--with-openssl=/usr/local/ssl \
--with-gettext \
--with-ldap=/usr/local \
--with-iconv \
--enable-ftp \
--with-dom \
--with-mime-magic \
--enable-mbstring \
--with-zlib \
--enable-track-vars \
--enable-sigchild \
--disable-ctype \
--disable-overload \
--disable-tokenizer \
--disable-posix \
--with-gd \
--with-apxs2=/usr/local/apache2.0.53/bin/apxs \
--with-mysql \
--with-pgsql \
--with-oci8=/oracle/product/9.2.0 \
--with-oracle=/oracle/product/9.2.0 \
--with-png-dir=/usr/local \
--with-zlib-dir=/usr/local \
--with-freetype-dir=/usr/local \
--with-jpeg-dir=/usr/local
make
make install
Questions:
Is there any problem of PHP with SunFire T2000 servers or 64-bits platforms?
Is there any flag of PHP would be use to compilarion PHP in 64-bits or multithread?
I wait for any comments or suggestions about our problem with PHP compilation and performance in Solaris 10. Thanks a lot.
Sergio.

I presume you compiled php on the Sun server, was this done using gcc or the Sun One C compiler.
If the latter then you can also use the flag: --enable-nonportable-atomics when you run configure

OpenGL/Elite3D Performance in Solaris 10?

Similar Messages

Maybe you are looking for