2003-04-01 02:30:57 +08:00
|
|
|
/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
|
|
|
|
|
* Copyright by the Board of Trustees of the University of Illinois. *
|
|
|
|
|
* All rights reserved. *
|
|
|
|
|
* *
|
|
|
|
|
* This file is part of HDF5. The full HDF5 copyright notice, including *
|
|
|
|
|
* terms governing use, modification, and redistribution, is contained in *
|
|
|
|
|
* the files COPYING and Copyright.html. COPYING can be found at the root *
|
|
|
|
|
* of the source code distribution tree; Copyright.html can be found at the *
|
|
|
|
|
* root level of an installed copy of the electronic HDF5 document set and *
|
|
|
|
|
* is linked from the top-level documents page. It can also be found at *
|
|
|
|
|
* http://hdf.ncsa.uiuc.edu/HDF5/doc/Copyright.html. If you do not have *
|
|
|
|
|
* access to either file, you may request a copy from hdfhelp@ncsa.uiuc.edu. *
|
|
|
|
|
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
|
1997-07-31 05:17:56 +08:00
|
|
|
|
|
|
|
|
/*
|
1997-08-16 00:51:34 +08:00
|
|
|
* This file contains public declarations for the H5F module.
|
1997-07-31 05:17:56 +08:00
|
|
|
*/
|
1997-08-16 00:51:34 +08:00
|
|
|
#ifndef _H5Fpublic_H
|
|
|
|
|
#define _H5Fpublic_H
|
|
|
|
|
|
|
|
|
|
/* Public header files needed by this file */
|
2001-04-06 01:29:14 +08:00
|
|
|
#include "H5public.h"
|
2005-04-29 00:40:07 +08:00
|
|
|
#include "H5Cpublic.h"
|
|
|
|
|
#include "H5ACpublic.h"
|
2001-04-06 01:29:14 +08:00
|
|
|
#include "H5Ipublic.h"
|
1997-07-31 05:17:56 +08:00
|
|
|
|
1998-02-20 02:19:48 +08:00
|
|
|
/*
|
|
|
|
|
* These are the bits that can be passed to the `flags' argument of
|
|
|
|
|
* H5Fcreate() and H5Fopen(). Use the bit-wise OR operator (|) to combine
|
1998-09-03 05:16:24 +08:00
|
|
|
* them as needed. As a side effect, they call H5check_version() to make sure
|
1998-04-23 01:26:01 +08:00
|
|
|
* that the application is compiled with a version of the hdf5 header files
|
|
|
|
|
* which are compatible with the library to which the application is linked.
|
|
|
|
|
* We're assuming that these constants are used rather early in the hdf5
|
|
|
|
|
* session.
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
*
|
|
|
|
|
* NOTE: When adding H5F_ACC_* macros, remember to redefine them in H5Fprivate.h
|
|
|
|
|
*
|
1998-02-20 02:19:48 +08:00
|
|
|
*/
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
|
2004-12-29 22:26:20 +08:00
|
|
|
/* When this header is included from a private header, don't make calls to H5check() */
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
#undef H5CHECK
|
2004-12-29 22:26:20 +08:00
|
|
|
#ifndef _H5private_H
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
#define H5CHECK H5check(),
|
2004-12-29 22:26:20 +08:00
|
|
|
#else /* _H5private_H */
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
#define H5CHECK
|
2004-12-29 22:26:20 +08:00
|
|
|
#endif /* _H5private_H */
|
[svn-r6252] Purpose:
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
|
|
|
|
|
|
|
|
#define H5F_ACC_RDONLY (H5CHECK 0x0000u) /*absence of rdwr => rd-only */
|
|
|
|
|
#define H5F_ACC_RDWR (H5CHECK 0x0001u) /*open for read and write */
|
|
|
|
|
#define H5F_ACC_TRUNC (H5CHECK 0x0002u) /*overwrite existing files */
|
|
|
|
|
#define H5F_ACC_EXCL (H5CHECK 0x0004u) /*fail if file already exists*/
|
|
|
|
|
#define H5F_ACC_DEBUG (H5CHECK 0x0008u) /*print debug info */
|
|
|
|
|
#define H5F_ACC_CREAT (H5CHECK 0x0010u) /*create non-existing files */
|
1998-02-20 02:19:48 +08:00
|
|
|
|
2001-12-12 03:53:44 +08:00
|
|
|
#define H5F_OBJ_FILE (0x0001u)
|
|
|
|
|
#define H5F_OBJ_DATASET (0x0002u)
|
|
|
|
|
#define H5F_OBJ_GROUP (0x0004u)
|
|
|
|
|
#define H5F_OBJ_DATATYPE (0x0008u)
|
2003-06-19 04:50:22 +08:00
|
|
|
#define H5F_OBJ_ATTR (0x0010u)
|
|
|
|
|
#define H5F_OBJ_ALL (H5F_OBJ_FILE|H5F_OBJ_DATASET|H5F_OBJ_GROUP|H5F_OBJ_DATATYPE|H5F_OBJ_ATTR)
|
1998-02-20 02:19:48 +08:00
|
|
|
|
1999-11-23 02:08:32 +08:00
|
|
|
#ifdef H5_HAVE_PARALLEL
|
1998-02-20 02:19:48 +08:00
|
|
|
/*
|
1999-08-11 04:21:32 +08:00
|
|
|
* Use this constant string as the MPI_Info key to set H5Fmpio debug flags.
|
|
|
|
|
* To turn on H5Fmpio debug flags, set the MPI_Info value with this key to
|
|
|
|
|
* have the value of a string consisting of the characters that turn on the
|
|
|
|
|
* desired flags.
|
1998-02-20 02:19:48 +08:00
|
|
|
*/
|
1999-08-11 04:21:32 +08:00
|
|
|
#define H5F_MPIO_DEBUG_KEY "H5F_mpio_debug_key"
|
2002-06-20 01:07:52 +08:00
|
|
|
#endif /* H5_HAVE_PARALLEL */
|
1998-02-20 02:19:48 +08:00
|
|
|
|
1999-08-11 04:21:32 +08:00
|
|
|
/* The difference between a single file and a set of mounted files */
|
1998-10-16 23:40:37 +08:00
|
|
|
typedef enum H5F_scope_t {
|
|
|
|
|
H5F_SCOPE_LOCAL = 0, /*specified file handle only */
|
|
|
|
|
H5F_SCOPE_GLOBAL = 1, /*entire virtual file */
|
|
|
|
|
H5F_SCOPE_DOWN = 2 /*for internal use only */
|
|
|
|
|
} H5F_scope_t;
|
|
|
|
|
|
1998-03-05 00:20:23 +08:00
|
|
|
/* Unlimited file size for H5Pset_external() */
|
1998-04-09 05:43:02 +08:00
|
|
|
#define H5F_UNLIMITED ((hsize_t)(-1L))
|
1998-02-20 02:19:48 +08:00
|
|
|
|
2001-12-12 03:53:44 +08:00
|
|
|
/* How does file close behave?
|
|
|
|
|
* H5F_CLOSE_DEFAULT - Use the degree pre-defined by underlining VFL
|
|
|
|
|
* H5F_CLOSE_WEAK - file closes only after all opened objects are closed
|
|
|
|
|
* H5F_CLOSE_SEMI - if no opened objects, file is close; otherwise, file
|
|
|
|
|
close fails
|
|
|
|
|
* H5F_CLOSE_STRONG - if there are opened objects, close them first, then
|
|
|
|
|
close file
|
|
|
|
|
*/
|
|
|
|
|
typedef enum H5F_close_degree_t {
|
|
|
|
|
H5F_CLOSE_DEFAULT = 0,
|
|
|
|
|
H5F_CLOSE_WEAK = 1,
|
|
|
|
|
H5F_CLOSE_SEMI = 2,
|
|
|
|
|
H5F_CLOSE_STRONG = 3
|
|
|
|
|
} H5F_close_degree_t;
|
|
|
|
|
|
1997-08-16 00:51:34 +08:00
|
|
|
#ifdef __cplusplus
|
1998-01-23 00:41:32 +08:00
|
|
|
extern "C" {
|
1997-08-02 06:48:01 +08:00
|
|
|
#endif
|
|
|
|
|
|
1997-07-31 05:17:56 +08:00
|
|
|
/* Functions in H5F.c */
|
2002-09-21 04:36:09 +08:00
|
|
|
H5_DLL htri_t H5Fis_hdf5 (const char *filename);
|
|
|
|
|
H5_DLL hid_t H5Fcreate (const char *filename, unsigned flags,
|
2001-12-12 03:53:44 +08:00
|
|
|
hid_t create_plist, hid_t access_plist);
|
2002-09-21 04:36:09 +08:00
|
|
|
H5_DLL hid_t H5Fopen (const char *filename, unsigned flags,
|
2001-12-12 03:53:44 +08:00
|
|
|
hid_t access_plist);
|
2002-09-21 04:36:09 +08:00
|
|
|
H5_DLL hid_t H5Freopen(hid_t file_id);
|
|
|
|
|
H5_DLL herr_t H5Fflush(hid_t object_id, H5F_scope_t scope);
|
|
|
|
|
H5_DLL herr_t H5Fclose (hid_t file_id);
|
|
|
|
|
H5_DLL hid_t H5Fget_create_plist (hid_t file_id);
|
|
|
|
|
H5_DLL hid_t H5Fget_access_plist (hid_t file_id);
|
2003-06-19 04:50:22 +08:00
|
|
|
H5_DLL int H5Fget_obj_count(hid_t file_id, unsigned types);
|
|
|
|
|
H5_DLL int H5Fget_obj_ids(hid_t file_id, unsigned types, int max_objs, hid_t *obj_id_list);
|
2002-10-01 00:31:55 +08:00
|
|
|
H5_DLL herr_t H5Fget_vfd_handle(hid_t file_id, hid_t fapl, void** file_handle);
|
2002-09-21 04:36:09 +08:00
|
|
|
H5_DLL herr_t H5Fmount(hid_t loc, const char *name, hid_t child, hid_t plist);
|
|
|
|
|
H5_DLL herr_t H5Funmount(hid_t loc, const char *name);
|
2003-10-07 21:32:32 +08:00
|
|
|
H5_DLL hssize_t H5Fget_freespace(hid_t file_id);
|
2004-07-08 01:03:02 +08:00
|
|
|
H5_DLL herr_t H5Fget_filesize(hid_t file_id, hsize_t *size);
|
2005-04-29 00:40:07 +08:00
|
|
|
H5_DLL herr_t H5Fget_mdc_config(hid_t file_id,
|
|
|
|
|
H5AC_cache_config_t * config_ptr);
|
|
|
|
|
H5_DLL herr_t H5Fset_mdc_config(hid_t file_id,
|
|
|
|
|
H5AC_cache_config_t * config_ptr);
|
|
|
|
|
H5_DLL herr_t H5Fget_mdc_hit_rate(hid_t file_id, double * hit_rate_ptr);
|
|
|
|
|
H5_DLL herr_t H5Fget_mdc_size(hid_t file_id,
|
|
|
|
|
size_t * max_size_ptr,
|
|
|
|
|
size_t * min_clean_size_ptr,
|
|
|
|
|
size_t * cur_size_ptr,
|
2005-05-03 05:36:43 +08:00
|
|
|
int * cur_num_entries_ptr);
|
2005-04-29 00:40:07 +08:00
|
|
|
H5_DLL herr_t H5Freset_mdc_hit_rate_stats(hid_t file_id);
|
2004-06-30 21:45:07 +08:00
|
|
|
H5_DLL ssize_t H5Fget_name(hid_t obj_id, char *name, size_t size);
|
1997-07-31 05:17:56 +08:00
|
|
|
|
1997-08-16 00:51:34 +08:00
|
|
|
#ifdef __cplusplus
|
1997-07-31 05:17:56 +08:00
|
|
|
}
|
1997-08-16 00:51:34 +08:00
|
|
|
#endif
|
1998-01-17 06:23:43 +08:00
|
|
|
#endif
|
