forked from GlobalArrays/ga
-
Notifications
You must be signed in to change notification settings - Fork 0
/
NEWS
170 lines (152 loc) · 6.29 KB
/
NEWS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
5.0 November 2010
* Now built using GNU autotools (autoconf,automake,libtool)
* Restricted arrays (see user manual)
* ARMCI runtime enhancements
+ On-demand connection management
+ Improved scalability for fence
* New Python interface
* Task Scheduling Library (tascel)
5.0b July 2010
* Now built using GNU autotools (autoconf,automake,libtool)
4.3 May 2010
* Optimized portals port to scale upto 200K procs
* Optimized OpenIB port
* BlueGene/P
* Support for Sparse Data Operations
+ (See GA user manual - Chapter 11 for more details)
4.2 July 2009
* Support for several new platforms
* Optimized portals port for Cray XT5
* BlueGene/P
* Optimized OpenIB port
* Support for Sparse Data Operations
+ (See GA user manual - Chapter 11 for more details)
4.1 May 2008
* Support for several new platforms
+ Cray XT4
+ BlueGene/L, BlueGene/P
+ OpenIB network
* Optimized one-sided non-blocking operations
* New networks. i.e. ARMCI_NETWORK
+ OPENIB
+ PORTALS
+ MPI-SPAWN (one-sided communication thru' MPI2 Dynamic
Process management and Send/Recv)
4.0 April 2006
* Support for multilevel parallelism: processor group awareness
* GA_Dgemm matrix multiplication based on SRUMMA
* Support for large arrays (a terabyte Global Array now possible)
* Optimized one-sided non-blocking operations
* Supports various platforms (Crays, IBM SPs, SGI Altix, ...) and
interconnects (Myrinet, Quadrics, Infiniband, ...)
3.4 2004 (beta)
* Initial support for multilevel parallelism
* processor group awareness
* nonblocking one-sided interfaces
* optimized port to Cray X1, SGI Altix
3.3 July, 2003
* nonblocking one-sided interfaces
* optimized matrix multiplication (ga_dgemm)
* mirrored arrays
* unoptimized port to Cray X1
* support for infiniband network (Mellanox VAPI)
* GA CCA component
3.2 October, 2002
* support for ghost cells
* port to Linux64/Itanium, Hitachi, Cray SV1, MAC X, and Cygwin
* support two underscores in fortran names on Linux/Cygwin
(enabled by setting F2C_TWO_UNDERSCORES)
* C++ bindings to Global Arrays
* element-wise and matrix operations
* synchronization control in collective operations
* additional datatypes: float & long in C, real in Fortran
* cluster information calls
* n-dim DRA
3.1 November 6th, 2000
* ports to 64-bit IBM, HP, Sun, Fujitsu, Linux/Alpha platforms
* new operations:
+ periodic version of put,get,accumulate
+ nga_scatter_acc
+ nga_select_elem
+ ga_nblock
+ nga_select_elem
* restructured makefiles to simplify changes and enable
export of make definitions to application makefiles
* new platforms supported by ARMCI (and thus GA)
+ Myrinet on Linux (Intel, Sparc) and Solaris/Sparc
+ Giganet on Linux/Intel
+ Compaq/Quadrics clusters
+ clusters of Windows NT under TCP/IP
3.0 November 15th, 1999
* This is the most significant update of the GA package to date:
all source code for the core functionality of the library
has been rewritten, ~90% of the library code is new/changed;
* added support for n-dimensional arrays (n<=7);
* added several n-dimensional test programs;
* updated utility tools: ga_print, ga_summarize, which
support n-dimensional operations;
* improved interoperability with MPI on clusters of
workstations
* ports to Cray J90 and Windows NT (single machine)
* ga_create uses default distribution for block size <1
rather than <2 (the old scheme can be restored see
OLD_DEFAULT_BLK in src/GNUmakefile)
* added a new, easier-to-use C interface (see ntestc.c)
* added User's Manual
2.4
* port to IBM SP running LAPI active messages
(significant performance improvement).
Version >=2.3 of the IBM PSSP parallel environment and
threaded version of MPI are required to run with LAPI.
* port to Fujitsu VX-4 and VPP-700.
* minor platform specific updates.
2.3
* added lock operations: ga_lock, ga_unlock,
ga_create_mutex, ga_destroy_mutex
* ga_acc works for integer data type
* port to the CRAY T3E (with streams disabled)
* environment variable TARGET_CPU can be use to guide
compilation for particular CPU (R10000 and R8000 are
currently recognized on the SGI systems)
* a new TARGET, SGI_N32 used for the 64-bit SGI systems
2.2
* added support for double complex datatype;
* port to HP and Convex SPP;
* latency optimizations for IBM SP running AIX 4.X;
* relaxed restriction on data types in ga_dscal,
ga_dscal_patch, ga_dadd, ga_dadd_patch,
ga_ifill_patch, ga_dfill_patch - these functions has
been renamed to: ga_scale, ga_scale_patch, ga_add,
ga_add_patch, ga_fill_patch;
* new functions: ga_fence and ga_init_fence, ga_zdot,
and ga_zdot_patch, ga_print_stats;
* new interface functions to ScaLAPACK: ga_cholesky,
ga_llt_solve, ga_spd_invert, ga_solve
* several configuration parameters are exposed to the
user in global/src/config.h to allow customization of
the package.
2.1
* selection of message-passing (TCGMSG or MPI) library
* new operations: ga_mpi_communicator and ga_proc_topology;
* performance optimizations;
* alternative C interface and C test programs;
* port to Sun Solaris and PC running Linux.
2.0
* support for networks of multiprocessors (SMP);
* (tightly) integrated shared and distributed memory versions;
* ability to restrict memory usage in global arrays;
* ga_scatter/gather supports integers;
* a program to test performance of GA primitives;
* different numbering of global array handles (< 0);
* and other internal improvements, like randomized order
of remote operations, prototypes of C-callable ops, etc.
1.3.1
* port to CRAY-T3D;
* new operation ga_summarize;
* improved ga_dgemm.
1.3
* introduced patch versions of ga_ddot, ga_dscal, ga_dadd;
* also ga_matmul_patch, ga_ifill_patch, ga_dfill_patch, and ga_duplicate.
* ga_copy_patch now has a transpose argument.
1.2
* (loosely) integrated shared and distributed memory versions