ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность

Changes to VT:

Просмотр исходного кода 
- configure:
		- changed default CUPTI library path to $CUPTI-DIR/lib64
	- VT Libs:
		- corrected prototype of MPI_Get_address in Fortran MPI wrappers (the second parameter should be an MPI_Aint* instead of MPI_Fint*)
		- temporary removed MPI_<Comm|Type|Win>_<get|set>_attr and MPI_Attr_<get|put> from the Fortran MPI wrappers due to missing conversion of the attribute value parameter
	- Docu:
		- latex doc \usepackage[T1]{fontenc} so that _ can be searched and copied
		- smaller font in Environment Variables section
		- some improvements in CUDA section
		- removed GPU idle time as official feature for CUPTI tracing method

This commit was SVN r26161.
Этот коммит содержится в:
Matthias Jurenz 2012-03-19 08:51:05 +00:00
родитель dffc1c5d43
Коммит a40b6dcc42
7 изменённых файлов: 73 добавлений и 54 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

8
ompi/contrib/vt/vt/ChangeLog
Просмотреть файл

@ -2,14 +2,13 @@
- updated version of internal OTF to 1.10.2openmpi
(see extlib/otf/ChangeLog)
- added support for CUDA tracing via CUPTI callbacks and activities
(runtime and driver API, kernels, memory copies, GPU idle time and
GPU memory usage)
(runtime and driver API, kernels, memory copies and GPU memory usage)
- added support for cudaMemcpyDefault and synchronous peer-to-peer
memory copies in CUDA library wrapper
- added configure switches to enable/disable CUPTI and CUDA wrapping
- fixed a bug in CUDA runtime wrapper initialization and thread creation
- fixed a build bug occurred if CUDA and CUPTI found, but support for
library tracing is disabled
library tracing was disabled
- fixed platform detection on BlueGene when configuring for the frontend
- fixed detection of Cray compiler's OpenMP flag in configure
- fixed detection of C++ runtime libraries for Cray and PGI 11.x compilers
@ -18,6 +17,9 @@
it's provided by dlfcn.h but not working (segmentation fault)
- fixed conversion from MPI_Fint-arrays to MPI_Aint-arrays in Fortran
MPI wrappers
- temporary removed MPI_<Comm|Type|Win>_<get|set>_attr and
MPI_Attr_<get|put> from the Fortran MPI wrappers due to missing
conversion of the attribute value parameter
- fixed handling of empty MPI groups (MPI_GROUP_EMPTY)
- fixed handling of MPI groups implicitly generated by MPI_Win_create
- fixed order of OpenMP threads based on its id (omp_get_thread_num)

4
ompi/contrib/vt/vt/config/m4/acinclude.cupti.m4
Просмотреть файл

@ -33,9 +33,9 @@ AC_DEFUN([ACVT_CUPTI],
AC_ARG_WITH(cupti-lib-dir,
AC_HELP_STRING([--with-cupti-lib-dir=CUPTILIBDIR],
[give the path for CUPTI-libraries, default: CUPTIDIR/lib]),
[give the path for CUPTI-libraries, default: CUPTIDIR/lib64]),
[CUPTILIBDIR="-L$withval/"],
[AS_IF([test x"$CUPTIDIR" != x], [CUPTILIBDIR="-L$CUPTIDIR"lib/])])
[AS_IF([test x"$CUPTIDIR" != x], [CUPTILIBDIR="-L$CUPTIDIR"lib64/])])
AC_ARG_WITH(cupti-lib,
AC_HELP_STRING([--with-cupti-lib=CUPTILIB], [use given cupti lib, default: -lcupti CUDALIB]),

10
ompi/contrib/vt/vt/config/mpigen/fortran_dont_generate.txt
Просмотреть файл

@ -22,6 +22,16 @@ MPI_Comm_create_keyval
MPI_Type_create_keyval
MPI_Win_create_keyval
# TODO: Fortran/C conversion of attribute value parameter
MPI_Attr_get
MPI_Attr_put
MPI_Comm_get_attr
MPI_Comm_set_attr
MPI_Type_get_attr
MPI_Type_set_attr
MPI_Win_get_attr
MPI_Win_set_attr
MPI_Comm_spawn
MPI_Comm_spawn_multiple

3
ompi/contrib/vt/vt/config/mpigen/mk_fortran_wrapper.sh.in
Просмотреть файл

@ -97,7 +97,8 @@ cat <<End-of-File >$tmp.awk
aint_convert=1
if (index(mpicall,"MPI_Type_create_hindexed") != 0 ||
if (index(mpicall,"MPI_Get_address") != 0 ||
index(mpicall,"MPI_Type_create_hindexed") != 0 ||
index(mpicall,"MPI_Type_create_hvector") != 0 ||
index(mpicall,"MPI_Type_create_struct") != 0)
{

98
ompi/contrib/vt/vt/doc/UserManual.html
Просмотреть файл

@ -314,20 +314,20 @@ OpenMP events, and performance counters.
<P>
After a successful tracing run, VampirTrace writes all collected data to a
trace file in the Open Trace Format (OTF)<A NAME="tex2html1"
HREF="#foot1530"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1533"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>.
As a result, the information is available for post-mortem analysis and
visualization by various tools.
Most notably, VampirTrace provides the input data for the Vampir analysis
and visualization tool<A NAME="tex2html2"
HREF="#foot1531"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1534"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>.
<P>
VampirTrace is included in OpenMPI&nbsp;1.3 and later versions.
If not disabled explicitly, VampirTrace is built automatically when installing
OpenMPI<A NAME="tex2html3"
HREF="#foot1532"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1535"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>.
<P>
@ -1143,7 +1143,7 @@ in a single file, that
The names in between may contain wildcards as ``?'', ``*', and ``#'', each entry gets a new line.
The lists end with <TT>END[_FILE]_&lt;INCLUDE|EXCLUDE&gt;_LIST</TT>. For further information on selective
profiling have a look at the TAU documentation<A NAME="tex2html4"
HREF="#foot1556"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1559"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>.
To announce the file through the compiler wrapper use the option <TT>-vt:tau</TT>:
<PRE>
@ -1160,7 +1160,7 @@ Binary Instrumentation Using Dyninst
The option <TT>-vt:inst dyninst</TT> is used with the compiler wrapper to
instrument the application during runtime (binary instrumentation), by using
Dyninst<A NAME="tex2html5"
HREF="#foot1557"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1560"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>.
Recompiling is not necessary for this kind of instrumentation,
but relinking:
@ -1311,7 +1311,7 @@ Tracing Calls to 3rd-Party Libraries
VampirTrace is also capable to trace calls to third party libraries, which come with
at least one C header file even without the library's source code. If VampirTrace was
built with support for library tracing (the CTool library<A NAME="tex2html6"
HREF="#foot1558"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1561"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A> is required), the tool <TT>vtlibwrapgen</TT> can be used to
generate a wrapper library to intercept each call to the actual library functions.
This wrapper library can be linked to the application or used in combination with the
@ -1903,7 +1903,7 @@ for the enhanced timer synchronization:
<UL>
<LI>CLAPACK <A NAME="tex2html7"
HREF="#foot1568"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1571"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>
</LI>
<LI>AMD ACML
@ -2165,18 +2165,20 @@ NVIDIA CUDA
When tracing CUDA applications, only user events and functions are recorded,
which are automatically or manually instrumented. CUDA API functions
will not be traced by default.
To enable tracing of CUDA runtime and driver API functions and asynchronous
CUDA device activities (like kernel execution and asynchronous memory copies) build
VampirTrace with CUDA support and set the following environment variable:
To enable tracing of CUDA runtime and driver API functions and CUDA device
activities (like kernel execution and memory copies) build VampirTrace with
CUDA support and set the following environment variable:
<P></P>
<P>
<DIV ALIGN="CENTER">
<TT>export VT_CUDATRACE=[yes|1|2|3|4|no]</TT>
<BR></DIV>
<P>
<P><P>
<BR>
<DIV ALIGN="CENTER"> <TABLE CELLPADDING=3 BORDER="1">
<DIV ALIGN="CENTER"></DIV>
<P>
<DIV ALIGN="CENTER"><TABLE CELLPADDING=3 BORDER="1">
<TR><TD ALIGN="RIGHT">Option</TD>
<TD ALIGN="CENTER">CUDA API</TD>
<TD ALIGN="CENTER">GPU Activity</TD>
@ -2215,9 +2217,10 @@ NVIDIA CUDA
<P>
Since CUDA Toolkit 4.1 the <SPAN CLASS="textbf">CUDA</SPAN> <SPAN CLASS="textbf">P</SPAN>rofiling and <SPAN CLASS="textbf">T</SPAN>ool <SPAN CLASS="textbf">I</SPAN>nterface
(CUPTI) allows capturing of CUDA device activities. Therewith new tracing
abilities (see table above) are available. VampirTrace trace has currently
abilities (see option 2, 3 and 4 in the table above) are available. VampirTrace trace has currently
two methods to trace the CUDA runtime API and corresponding GPU activities:
traditional library wrapping with CUDA events or CUPTI. Several features are
traditional library wrapping with CUDA events for GPU activity measurement and
tracing via the CUPTI interface. Several features are
just implemented in the library wrapping approach, whereas the CUPTI
measurement brings new possibilities and occasionally more accuracy.
<P><P>
@ -2245,8 +2248,9 @@ Tracing of CUDA kernels is enabled/disabled. With '<TT>2</TT>' additional
</DD>
<DT></DT>
<DD><TT>VT_CUDATRACE_IDLE=[yes|<SPAN CLASS="textbf">no</SPAN>]</TT>
<SMALL CLASS="SCRIPTSIZE">(CUDA runtime API wrapper only) </SMALL>
<BR>
Show the GPU compute idle time on first used CUDA stream, if set to <TT>yes</TT>.
Show the GPU compute idle time on CUDA stream zero (default stream), if set to <TT>yes</TT>.
<P>
</DD>
@ -2274,7 +2278,7 @@ Controls how VampirTrace handles synchronizing CUDA API calls, especially
significantly and will not be shown in the trace.
At level 2 the additional synchronization will be exposed to the user.
This allows a better view on the application execution, showing how much
time is actually spent waiting for a kernel to complete during synchronization.
time is actually spent waiting for the GPU to complete.
Level 3 will further use the synchronization to flush the internal task
buffer and perform a timer synchronization between GPU and host. This
introduces a minimal overhead but increases timer precision and prevents
@ -2314,8 +2318,8 @@ Record GPU memory usage as counter ``gpu_mem_usage``, if set to <TT>yes</TT>.
<SMALL CLASS="SCRIPTSIZE">(CUDA runtime API wrapper only) </SMALL>
<BR>
Print out an error message and exit the program, if a function call to a
GPU library does not return successfully. The default is just a warning message
without program exit.
GPU library does not return successfully. The default is just a warning
message.
<P>
</DD>
@ -2324,7 +2328,7 @@ Print out an error message and exit the program, if a function call to a
<SMALL CLASS="SCRIPTSIZE">(CUDA runtime API wrapper only) </SMALL>
<BR>
Do not cleanup all GPU resources (profiling events, contexts, event groups),
as they might have been already implicitly cleaned up by the GPU runtime.
as they might have already been implicitly cleaned up by the GPU runtime.
<P>
</DD>
@ -2375,7 +2379,7 @@ Several new region groups have been introduced:
<TD ALIGN="LEFT" VALIGN="TOP" WIDTH=269>CUDA synchronization</TD>
</TR>
<TR><TH ALIGN="LEFT"><SPAN CLASS="textbf">CUDA_KERNEL</SPAN></TH>
<TD ALIGN="LEFT" VALIGN="TOP" WIDTH=269>CUDA kernels/functions can only appear on
<TD ALIGN="LEFT" VALIGN="TOP" WIDTH=269>CUDA kernels (device functions) can only appear on
``CUDA-Threads''</TD>
</TR>
<TR><TH ALIGN="LEFT"><SPAN CLASS="textbf">CUDA_IDLE</SPAN></TH>
@ -2419,9 +2423,9 @@ CUDA Runtime API Wrapper Particularities</A>
</H3>
To ensure measurement of correct data rates for synchronous CUDA
memory copies, the VampirTrace CUDA runtime library wrapper inserts a CUDA
synchronization before.
Otherwise the CUDA memory copy call would do the synchronization and it was
not possible to get correct transfer rates.
synchronization before the memory copy call.
Otherwise the implicit synchronization of the CUDA memory copy call could not
be exposed and it was not possible to get correct transfer rates.
<P></P>
<P>
@ -2441,7 +2445,7 @@ Counter via CUDA API</A>
<SPAN CLASS="textit">cudaMalloc</SPAN> and <SPAN CLASS="textit">cudaFree</SPAN> functions will be tracked to write
the GPU memory usage counter <TT>gpu_mem_usage</TT>.
This counter does not need space in the CUDA buffer. The counter values
will be directly written to the default CUDA stream '1'. This stream will be
will be written directly to the default CUDA stream '1'. This stream will be
created, if it does not exist and does not have to contain any other CUDA
device activities. If the environment variable is set to <TT>2</TT>, missing
<SPAN CLASS="textit">cudaFree()</SPAN> calls will be printed to stderr.
@ -2547,8 +2551,7 @@ Tracing the NVIDIA CUDA SDK 3.x and 4.x</A>
Use the compiler switches for MPI, multi-threaded
and hybrid programs, if necessary (e.g. the CUDA SDK example
<TT>simpleMultiGPU</TT> is a multi-threaded program, which needs to be linked
with a multi-threaded VampirTrace library - uncomment the compiler switch
in the linker command to use the multi-threaded VampirTrace library).
with a multi-threaded VampirTrace library).
<P><P>
<BR>
@ -2567,21 +2570,22 @@ Multi-threaded CUDA applications</A>
<BR>
<P>
<SPAN CLASS="textbf">Note:</SPAN>
<SPAN CLASS="textbf">Notes:</SPAN>
<BR>
For 32-bit systems VampirTrace has to be configured with the 32-bit
version of CUDA runtime library. If the link test fails, use the
version of the CUDA runtime library. If the link test fails, use the
following configure option :
<PRE>
--with-cuda-lib-dir=$CUDA_INSTALL_PATH/lib
</PRE>
<P>
Since CUDA toolkit version 4.1 the 64-bit CUPTI library is located in the
lib64 directory of CUPTI. If the link test fails, use the
To build CUPTI support on 32-bit systems (or for CUPTI 1.0),
VampirTrace has to be configured with the 32-bit version of the CUPTI library.
If the link test fails, use the
following configure option :
<PRE>
--with-cupti-lib-dir=$CUDA_INSTALL_PATH/extras/CUPTI/lib64
--with-cupti-lib-dir=$CUPTI_INSTALL_PATH/lib
</PRE>
<P>
@ -3441,7 +3445,7 @@ default: automatically by configure.
enable support for Dyninst instrumentation,
default: enable if found by configure.
<SPAN CLASS="textbf">Note:</SPAN> Requires Dyninst<A NAME="tex2html8"
HREF="#foot1596"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1599"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A> version 6.1 or higher!
<P>
@ -3463,9 +3467,9 @@ enable support for automatic source code
instrumentation by using TAU, default: enable if
found by configure.
<SPAN CLASS="textbf">Note:</SPAN> Requires PDToolkit<A NAME="tex2html9"
HREF="#foot1597"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1600"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A> or TAU<A NAME="tex2html10"
HREF="#foot1598"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1601"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A>!
<P>
@ -4034,7 +4038,7 @@ give the path for JVMTI-include files, default:
<P>
To enable support for generating wrapper for 3th-Party libraries the C code parser CTool<A NAME="tex2html11"
HREF="#foot1599"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
HREF="#foot1602"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A> is needed:
<P>
@ -4129,7 +4133,7 @@ give the path for CUPTI-include files, default:
<DD>&nbsp;
<BR>
give the path for CUPTI-libraries, default:
CUPTIDIR/lib
CUPTIDIR/lib64
</DD>
<DT><STRONG><TT>-with-cupti-lib=CUPTILIB</TT></STRONG></DT>
<DD>&nbsp;
@ -5244,69 +5248,69 @@ If you provide us with your additions afterwards we will consider merging them
into the official VampirTrace package.
<BR><HR><H4>Footnotes</H4>
<DL>
<DT><A NAME="foot1530">... (OTF)</A><A
<DT><A NAME="foot1533">... (OTF)</A><A
HREF="#tex2html1"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.tu-dresden.de/zih/otf
</DD>
<DT><A NAME="foot1531">... tool </A><A
<DT><A NAME="foot1534">... tool </A><A
HREF="#tex2html2"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.vampir.eu
</DD>
<DT><A NAME="foot1532">...
<DT><A NAME="foot1535">...
Open MPI </A><A
HREF="#tex2html3"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.open-mpi.org/faq/?category=vampirtrace
</DD>
<DT><A NAME="foot1556">... documentation </A><A
<DT><A NAME="foot1559">... documentation </A><A
HREF="#tex2html4"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.cs.uoregon.edu/Research/tau/docs/newguide/bk05ch02.html#d0e3770
</DD>
<DT><A NAME="foot1557">...
<DT><A NAME="foot1560">...
Dyninst </A><A
HREF="#tex2html5"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.dyninst.org
</DD>
<DT><A NAME="foot1558">... library </A><A
<DT><A NAME="foot1561">... library </A><A
HREF="#tex2html6"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://sourceforge.net/projects/ctool
</DD>
<DT><A NAME="foot1568">... CLAPACK</A><A
<DT><A NAME="foot1571">... CLAPACK</A><A
HREF="#tex2html7"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>www.netlib.org/clapack
</DD>
<DT><A NAME="foot1596">... Dyninst </A><A
<DT><A NAME="foot1599">... Dyninst </A><A
HREF="#tex2html8"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.dyninst.org
</DD>
<DT><A NAME="foot1597">... PDToolkit </A><A
<DT><A NAME="foot1600">... PDToolkit </A><A
HREF="#tex2html9"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://www.cs.uoregon.edu/research/pdt/home.php
</DD>
<DT><A NAME="foot1598">... TAU </A><A
<DT><A NAME="foot1601">... TAU </A><A
HREF="#tex2html10"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://tau.uoregon.edu
</DD>
<DT><A NAME="foot1599">... CTool </A><A
<DT><A NAME="foot1602">... CTool </A><A
HREF="#tex2html11"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]"
SRC="/usr/share/latex2html/icons/footnote.png"></SUP></A></DT>
<DD>http://sourceforge.net/projects/ctool

Двоичные данные
ompi/contrib/vt/vt/doc/UserManual.pdf
Просмотреть файл

Двоичный файл не отображается.

4
ompi/contrib/vt/vt/vtlib/vt_cudart.c
Просмотреть файл

@ -61,7 +61,7 @@ VTThrdMutex* VTThrdMutexCudart = NULL;
* @param _bytes the number of bytes to be transfered
* @param _call the function call of the CUDA Runtime API function
*/
#define CUDA_SEND_RECV(_ptid, _kind, _bytes, _call){ \
#define CUDA_SEND_RECV(_ptid, _kind, _bytes, _call){ \
uint64_t time = 0; \
uint8_t do_traceE = 0; /* is call limit reached */ \
VTCUDADevice* vtDev = NULL; \
@ -344,8 +344,10 @@ static uint8_t trace_gpumem = 0;
/* flag: trace NVIDIA CUPTI events/counters */
static uint8_t trace_cupti_events = 0;
#if defined(VT_CUPTI_EVENTS)
/* flag: sampling for CUPTI counter values enabled? */
static uint8_t cupti_event_sampling = 0;
#endif
/* flag: event based tracing (kernels, memcpyAsync) enabled? */
static uint8_t trace_events = 1;