Re: [r200] Viewperf-7.1.1 results on SMP
Dieter Ntzel schrieb: Argh, what a regression. most tests show mostly empty and/or black and/or white windows... Strange, I just ran Viewperf 8 on my r200 and didn't notice any problems (expect ensight-01 that didn't even run, but it seems that's a problem with a Viewperf script). Philipp Klaus Krause --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
Philipp Klaus Krause schrieb: Dieter Ntzel schrieb: Argh, what a regression. most tests show mostly empty and/or black and/or white windows... Strange, I just ran Viewperf 8 8.0.1? I've downloaded it in the morning. ;-) on my r200 and didn't notice any problems SMP-system (EnvLINUX.c reports num=2) -DMP? -lpthread? Some more (hardware) specs, please. Without that 7.1.1 finished. But NOT that fast. And I see some garbage from KDE (Konsole content) superimposed in the resulted *.pngs. Anyone? (expect ensight-01 that didn't even run, but it seems that's a problem with a Viewperf script). I'll have a look. -Dieter --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
I did use a slightly older cvs version with only my vertex program patch applied and built in the Mes tree without optimizations. Strange, I just ran Viewperf 8 8.0.1? Yes. I've downloaded it in the morning. ;-) on my r200 and didn't notice any problems SMP-system (EnvLINUX.c reports num=2) -DMP? -lpthread? gcc-3.5 -O2 -march=athlon-mp -msse -DMP Some more (hardware) specs, please. Dual Athlon MP, Radeon 9000 Pro I'll retry with new drivers and rebuilt -O3, -mfpmath=sse Viewperf. Philipp --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
gcc-3.5 -O2 -march=athlon-mp -msse -DMP Just saw that the it seems to take the options from the Makefile, I just set them in options (included by Makefile, but the CFLAGS defined there is overwrittem) Philipp --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
Philipp Klaus Krause schrieb:
gcc-3.5 -O2 -march=athlon-mp -msse -DMP
Just saw that the it seems to take the options from the Makefile,
I just set them in options (included by Makefile, but the CFLAGS
defined there is overwrittem)
;-)
That's why I 'included' my changed EnvLINUX.c (taken from EnvSGI.c) files.
src/EnvLINUX.c
[-]
int
GetNumProcessors()
{
return (2);
}
[-]
src/Makefile
[-]
CDEBUGFLAGS = -O -mcpu=athlon-mp -march=athlon-mp # -mfpmath=sse
INCLUDES = -I$(AUX_DIR) -I$(X_DIR)/include/ -I$(LIBGL)/include/
-I$(LIBPNG)/include
# Add -DMP to the following line to build a multithreaded viewperf
DEFINES = -DXWINDOWS -DSEARCHPATH -DLINUX -DPNG -DMP
# edit EnvXXX.c to be your version of Env.c, such as EnvDEC.c
ENV_C=EnvLINUX.c
CCFLAGS =
CFLAGS = $(CCFLAGS) $(CDEBUGFLAGS) $(INCLUDES) $(DEFINES)
LIBS = -L$(X_DIR)/lib -Lobjs -L$(AUX_DIR) -L$(TK_DIR) -L$(LIBGL)/lib
-L$(LIBPNG) /lib -lvp -lm -lX11 -lXext -laux -lGL -lGLU -lz -lpng -lpthread
[-]
Do you see any performance win with GCC-3.5 (auto vectorization)
compared to the old -mfpmath=sse?
-Dieter
---
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click
--
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
Dieter Nützel schrieb: Philipp Klaus Krause schrieb: gcc-3.5 -O2 -march=athlon-mp -msse -DMP Just saw that the it seems to take the options from the Makefile, I just set them in options (included by Makefile, but the CFLAGS defined there is overwrittem) ;-) That's why I 'included' my changed EnvLINUX.c (taken from EnvSGI.c) files. OK, adding -DMP makes some tests turn completely black (tried with ugs). Still using the old cvs driver with only vertex programs added and built in non-optimizing Mesa tree (driver built with gcc-3.3). Do you see any performance win with GCC-3.5 (auto vectorization) compared to the old -mfpmath=sse? I've never compiled Viewperf test with an older gcc version. Philipp --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
[r200] Viewperf-7.1.1 results on SMP
System:
dual Athlon MP 1900+
MSI K7D Master-L (aka AMD 768MPX)
1 GB DDR266 RAM, CL2 (2x 512 MB)
ATI Radeon 8500 QL, AGP (r200), 64 MB DDR RAM
all U160/320 ;-)
SuSE 9.0
* with latest 2.6.5-104 (official released 9.1) kernel
* glibc-2.3.3-73 (NPTL)
based on SuSE's 9.0 XFree86 4.3.0.1-46
DRM CVS
DRI CVS
Mesa CVS
+ r200-maybe-flush-less-3.diff
+ DRI-TLS-01.patch Mesa-TLS-01.patch
+ r200_vertex.patch (test.patch by Philipp Klaus Krause)
viewperf was compiled with threading enabled.
src/EnvLINUX.c
[-]
int
GetNumProcessors()
{
return (2);
}
[-]
src/Makefile
[-]
CDEBUGFLAGS = -O -mcpu=athlon -march=athlon -mfpmath=sse
INCLUDES = -I$(AUX_DIR) -I$(X_DIR)/include/ -I$(LIBGL)/include/
-I$(LIBPNG)/include
# Add -DMP to the following line to build a multithreaded viewperf
DEFINES = -DXWINDOWS -DSEARCHPATH -DLINUX -DPNG -DMP
# edit EnvXXX.c to be your version of Env.c, such as EnvDEC.c
ENV_C=EnvLINUX.c
CCFLAGS =
CFLAGS = $(CCFLAGS) $(CDEBUGFLAGS) $(INCLUDES) $(DEFINES)
LIBS = -L$(X_DIR)/lib -Lobjs -L$(AUX_DIR) -L$(TK_DIR) -L$(LIBGL)/lib
-L$(LIBPNG) /lib -lvp -lm -lX11 -lXext -laux -lGL -lGLU -lz -lpng -lpthread
[-]
OLD results
* all as above but UP-compiled 'cause the 'r200 multiple app lockups'
* NO DRI-TLS-01.patch Mesa-TLS-01.patch
* NO r200_vertex.patch (test.patch by Philipp Klaus Krause)
Run All Summary
-- sum_results/3dsmax/summary.txt
3dsmax-02 Weighted Geometric Mean = 6.352
-- sum_results/drv/summary.txt
drv-09 Weighted Geometric Mean = 4.235
-- sum_results/dx/summary.txt
dx-08 Weighted Geometric Mean = 29.728
-- sum_results/light/summary.txt
light-06 Weighted Geometric Mean = 6.921
-- sum_results/proe/summary.txt
proe-02 Weighted Geometric Mean = 4.412
-- sum_results/ugs/summary.txt
ugs-03 Weighted Geometric Mean = 2.457
And now the GREAT new ones:
Run All Summary
-- sum_results/3dsmax/summary.txt
3dsmax-02 Weighted Geometric Mean = 5.944
-- sum_results/drv/summary.txt
drv-09 Weighted Geometric Mean = 10.626
-- sum_results/dx/summary.txt
dx-08 Weighted Geometric Mean = 39.704
-- sum_results/light/summary.txt
light-06 Weighted Geometric Mean = 6.741
-- sum_results/proe/summary.txt
proe-02 Weighted Geometric Mean = 4.324
-- sum_results/ugs/summary.txt
ugs-03 Weighted Geometric Mean = 47.214
Cheers,
Dieter
---
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click
--
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
Am Dienstag, 31. August 2004 22:12 schrieb Dieter Ntzel:
System:
dual Athlon MP 1900+
MSI K7D Master-L (aka AMD 768MPX)
1 GB DDR266 RAM, CL2 (2x 512 MB)
ATI Radeon 8500 QL, AGP (r200), 64 MB DDR RAM
all U160/320 ;-)
SuSE 9.0
* with latest 2.6.5-104 (official released 9.1) kernel
* glibc-2.3.3-73 (NPTL)
based on SuSE's 9.0 XFree86 4.3.0.1-46
DRM CVS
DRI CVS
Mesa CVS
+ r200-maybe-flush-less-3.diff
+ DRI-TLS-01.patch Mesa-TLS-01.patch
+ r200_vertex.patch (test.patch by Philipp Klaus Krause)
viewperf was compiled with threading enabled.
src/EnvLINUX.c
[-]
int
GetNumProcessors()
{
return (2);
}
[-]
src/Makefile
[-]
CDEBUGFLAGS = -O -mcpu=athlon -march=athlon -mfpmath=sse
INCLUDES = -I$(AUX_DIR) -I$(X_DIR)/include/ -I$(LIBGL)/include/
-I$(LIBPNG)/include
# Add -DMP to the following line to build a multithreaded viewperf
DEFINES = -DXWINDOWS -DSEARCHPATH -DLINUX -DPNG -DMP
# edit EnvXXX.c to be your version of Env.c, such as EnvDEC.c
ENV_C=EnvLINUX.c
CCFLAGS =
CFLAGS = $(CCFLAGS) $(CDEBUGFLAGS) $(INCLUDES) $(DEFINES)
LIBS = -L$(X_DIR)/lib -Lobjs -L$(AUX_DIR) -L$(TK_DIR) -L$(LIBGL)/lib
-L$(LIBPNG) /lib -lvp -lm -lX11 -lXext -laux -lGL -lGLU -lz -lpng -lpthread
[-]
OLD results
* all as above but UP-compiled 'cause the 'r200 multiple app lockups'
* NO DRI-TLS-01.patch Mesa-TLS-01.patch
* NO r200_vertex.patch (test.patch by Philipp Klaus Krause)
Run All Summary
-- sum_results/3dsmax/summary.txt
3dsmax-02 Weighted Geometric Mean = 6.352
-- sum_results/drv/summary.txt
drv-09 Weighted Geometric Mean = 4.235
-- sum_results/dx/summary.txt
dx-08 Weighted Geometric Mean = 29.728
-- sum_results/light/summary.txt
light-06 Weighted Geometric Mean = 6.921
-- sum_results/proe/summary.txt
proe-02 Weighted Geometric Mean = 4.412
-- sum_results/ugs/summary.txt
ugs-03 Weighted Geometric Mean = 2.457
And now the GREAT new ones:
Run All Summary
-- sum_results/3dsmax/summary.txt
3dsmax-02 Weighted Geometric Mean = 5.944
-- sum_results/drv/summary.txt
drv-09 Weighted Geometric Mean = 10.626
-- sum_results/dx/summary.txt
dx-08 Weighted Geometric Mean = 39.704
-- sum_results/light/summary.txt
light-06 Weighted Geometric Mean = 6.741
-- sum_results/proe/summary.txt
proe-02 Weighted Geometric Mean = 4.324
-- sum_results/ugs/summary.txt
ugs-03 Weighted Geometric Mean = 47.214
Argh,
what a regression.
most tests show mostly empty and/or black and/or white windows...
Must rerun, when I can stay on the machine.
Expect a new try, soon.
-Dieter
---
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_idP47alloc_id808op=click
--
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [r200] Viewperf-7.1.1 results on SMP
Dieter Ntzel schrieb: src/Makefile [-] CDEBUGFLAGS = -O -mcpu=athlon -march=athlon -mfpmath=sse INCLUDES = -I$(AUX_DIR) -I$(X_DIR)/include/ -I$(LIBGL)/include/ -I$(LIBPNG)/include # Add -DMP to the following line to build a multithreaded viewperf DEFINES = -DXWINDOWS -DSEARCHPATH -DLINUX -DPNG -DMP # edit EnvXXX.c to be your version of Env.c, such as EnvDEC.c ENV_C=EnvLINUX.c CCFLAGS = CFLAGS = $(CCFLAGS) $(CDEBUGFLAGS) $(INCLUDES) $(DEFINES) LIBS = -L$(X_DIR)/lib -Lobjs -L$(AUX_DIR) -L$(TK_DIR) -L$(LIBGL)/lib -L$(LIBPNG) /lib -lvp -lm -lX11 -lXext -laux -lGL -lGLU -lz -lpng -lpthread [-] -mfpmath=sse will be ignored, since -march=athlon doesn't allow it (the original athlon didn't have sse) you have to use -march=athlon-mp. Philipp Klaus Krause --- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047alloc_id=10808op=click -- ___ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
