/[escript]/trunk/cusplibrary/CHANGELOG
ViewVC logotype

Annotation of /trunk/cusplibrary/CHANGELOG

Parent Directory Parent Directory | Revision Log Revision Log


Revision 4955 - (hide annotations)
Tue May 20 04:33:15 2014 UTC (5 years, 2 months ago) by caltinay
Original Path: branches/diaplayground/cusplibrary/CHANGELOG
File size: 5946 byte(s)
added pristine copy of cusplibrary (apache license) to be used by ripley.

1 caltinay 4955 #######################################
2     # Cusp v0.4.0 #
3     #######################################
4    
5     Summary
6     Cusp v0.4.0 provides support for CUDA 5.5 and Thrust v1.7. Version
7     0.4.0 includes several new graph algorithms based on backend integration
8     with back40computing, a refactored smoothed aggregation solver allowing
9     user-defined functions for core components and experimental work concerning
10     sparse matrix visualization.
11    
12     New Features
13     array2d formats allow access to row and column elements regardless of orientation
14     gallery method to generate grids
15     OpenGL implementation of spy to view sparse matrices
16     Generalized multilevel class
17     Added Sparse * Dense multiplication routine
18    
19     Algorithms
20     cusp::graph::bfs
21     cusp::graph::connected_components
22     cusp::graph::hilbert_curve
23     cusp::graph::maximum_flow
24     cusp::graph::max_flow_to_min_cut
25     cusp::graph::pseudo_peripheral_vertex
26     cusp::graph::symmetric_rcm
27     cusp::krylov::cr
28    
29     Bug Fixes
30     Resolved issues related to Thrust 1.6 & Thrust 1.7
31     Resolved issues related to CUDA 5.5
32     Silenced warnings concerning uninitialized memory in generalized SpMV
33    
34     #######################################
35     # Cusp v0.3.1 #
36     #######################################
37    
38     Summary
39     Minor compatibility fixes for Thrust 1.6.x
40    
41     #######################################
42     # Cusp v0.3.0 #
43     #######################################
44    
45     Summary
46     Cusp v0.3.0 provides support for CUDA 4.1 and Thrust v1.5. Version
47     0.3.0 includes incremental improvements to MatrixMarket I/O support
48     and array format conversions.
49    
50     New Features
51     MatrixMarket I/O routines support complex matrices
52     array1d is convertible to and from all other formats (e.g. array1d<->array2d)
53    
54     Bug Fixes
55     Resize final output of CSR * CSR to account for excluded zeros
56    
57     Acknowledgements
58     Justin Luitjens and Simon Layton for diagnosing CSR SpMM bug
59     Kashif Rasul for Thrust v1.6 support patch
60     Filipe Maia and Steven Dalton for general support and maintenance
61    
62     #######################################
63     # Cusp v0.2.0 #
64     #######################################
65    
66     Summary
67     Cusp v0.2.0 provides support for CUDA 4.0 in addition to many new
68     features and performance enhancements. Support for complex numbers
69     is now available through the cusp::complex class. Lightweight views
70     to dense arrays and sparse matrix formats are now provided, allowing
71     user-allocated data to be interfaced efficiently with Cusp algorithms.
72     The performance of sparse matrix format conversions has been improved
73     dramatically on the device. Preconditioners based on algebraic multigrid
74     (AMG) and approximate inverse (AINV) methods have been added to the library.
75    
76     New Features
77     Complex numbers
78     cusp::complex
79     cusp/cmath.h
80     cusp/complex.h
81    
82     Sparse Matrices
83     cusp::coo_matrix_view
84     cusp::csr_matrix_view
85     cusp::dia_matrix_view
86     cusp::ell_matrix_view
87     cusp::hyb_matrix_view
88    
89     Krylov methods
90     cusp::krylov::bicg
91     cusp::krylov::gmres
92     cusp::krylov::cg_m
93     cusp::krylov::bicgstab_m
94    
95     Input/Output
96     cusp::io::read_matrix_market_stream
97    
98     Algorithms
99     cusp::add
100     cusp::subtract
101     cusp::graph::maximal_independent_set
102     cusp::copy
103     cusp::convert
104     cusp::blas::nrm1
105     cusp::blas::nrmmax
106     cusp::relaxation::polynomial
107    
108     Matrix Gallery
109     cusp::gallery::diffusion
110    
111     Preconditioners
112     cusp::precond::scaled_bridson_ainv
113     cusp::precond::bridson_ainv
114     cusp::precond::nonsym_bridson_ainv
115     cusp::precond::smoothed_aggregation
116    
117     Bug Fixes
118     #12 MatrixMarket file reader fails on OSX
119     #18 ensure that IndexType=long is supported
120     #27 bicgstab can't solve Ix = b
121     #31 device::spmm_coo requires huge amount of memory
122     #37 cusp::graph::maximal_independent_set should return the size of the MIS
123     #45 silence compiler warnings
124     #66 thrust hg version no longer compatible with hg version of cusp
125     #71 Several cusp::blas calls cause unexpected results by converting the output to ScalarType
126    
127     Deprecations
128     cusp::print_matrix has been replaced by cusp::print
129     cusp::blas functions only accept arrays and array views
130    
131     Known Issues
132     Cusp is known to cause the nvcc compiler to issue warnings of the form:
133     warning: calling a host function ... from a __device__/__global__ function ... is not allowed
134    
135     Acknowledgements
136     Filipe Maia for general maintenance, testing, and implementations of cusp::complex and BiCG
137     Greg van Anders for CG-M and BiCGstab-M implementations
138     Joe Gordon for GMRES implementation
139     Jonathan Cohen for AINV preconditioners
140     Steven Dalton for improvements to conversion routines and smoothed aggregation preconditioner
141    
142     #######################################
143     # Cusp v0.1.1 #
144     #######################################
145    
146     Summary
147     Small fixes for compatibility with CUDA 3.1
148    
149     #######################################
150     # Cusp v0.1.0 #
151     #######################################
152    
153     Summary
154     Initial release
155    
156     Features
157     Sparse Matrix Support
158     cusp::coo_matrix
159     cusp::csr_matrix
160     cusp::dia_matrix
161     cusp::ell_matrix
162     cusp::hyb_matrix
163     conversion between all formats
164     host<->device transfers
165     Dense Matrix Support
166     cusp::array1d
167     cusp::array2d
168     Algorithms
169     cusp::multiply(A,x,y) : y = A*x
170     cusp::transpose(A,At) : A^T
171     cusp::blas::*
172     Gallery
173     cusp::gallery::stencil
174     cusp::gallery::poisson5pt
175     Solvers
176     cusp::krylov::cg
177     cusp::krylov::bicgstab
178     Monitors
179     cusp::default_monitor
180     cusp::verbose_monitor
181     Preconditioners
182     cusp::precond::diagonal
183     User-Defined Linear Operators
184     cusp::linear_operator
185     File I/O
186     cusp::io::read_matrix_market_file
187     cusp::io::write_matrix_market_file
188     Utilities
189     cusp::print_matrix
190     cusp::is_valid_matrix etc.
191    
192     Known Issues
193     Many routines assume that several arguments have the exact same type
194    

  ViewVC Help
Powered by ViewVC 1.1.26