cusparse 0.2.0.0 → 0.3.0.0

raw patch · 5 files changed

+50/−3 lines, 5 filessetup-changedPVP ok

version bump matches the API change (PVP)

API changes (from Hackage documentation)

Files

• CHANGELOG.md +6/−0
• Foreign/CUDA/BLAS/Sparse/Convert.chs +18/−0
• Foreign/CUDA/BLAS/Sparse/Internal/Types.chs +19/−0
• Setup.hs +4/−0
• cusparse.cabal +3/−3

@@ -4,6 +4,11 @@  The format is based on [Keep a Changelog](http://keepachangelog.com/). +## [0.3.0.0] - 2020-08-26+### Added+  * Operations from CUDA-10.1+  * Support for Cabal-3+ ## [0.2.0.0] - 2018-10-02 ### Added   * Level 3 operations from CUDA-9.2@@ -19,6 +24,7 @@    * First version; split off from [bmsherman/cublas](https://github.com/bmsherman/cublas). Released on an unsuspecting world. +[0.3.0.0]:    https://github.com/tmcdonell/cusparse/compare/v0.2.0.0...v0.3.0.0 [0.2.0.0]:    https://github.com/tmcdonell/cusparse/compare/0.1.0.1...v0.2.0.0 [0.1.0.1]:    https://github.com/tmcdonell/cusparse/compare/0.1.0.0...0.1.0.1 

@@ -26,6 +26,7 @@   Action(..),   Hybrid,   HybridPartition(..),+  Algorithm_csr2csc(..),   Info_csru2csr,   Info_prune,   sbsr2csr,@@ -182,6 +183,8 @@   dnnz_compress,   cnnz_compress,   znnz_compress,+  csr2cscEx2,+  csr2cscEx2_bufferSize,  ) where @@ -875,4 +878,19 @@  znnz_compress :: Handle -> Int -> MatrixDescriptor -> DevicePtr (Complex Double) -> DevicePtr Int32 -> DevicePtr Int32 -> Ptr Int32 -> (Complex Double) -> IO () znnz_compress _ _ _ _ _ _ _ _ = cusparseError "'znnz_compress' requires at least cuda-9.0"+#endif+#if CUDA_VERSION >= 10010++{-# INLINEABLE csr2cscEx2 #-}+{# fun unsafe cusparseCsr2cscEx2 as csr2cscEx2 { useHandle `Handle', `Int', `Int', `Int', useDevP `DevicePtr ()', useDevP `DevicePtr Int32', useDevP `DevicePtr Int32', useDevP `DevicePtr ()', useDevP `DevicePtr Int32', useDevP `DevicePtr Int32', cFromEnum `Type', cFromEnum `Action', cFromEnum `IndexBase', cFromEnum `Algorithm_csr2csc', useDevP `DevicePtr ()' } -> `()' checkStatus*- #}++{-# INLINEABLE csr2cscEx2_bufferSize #-}+{# fun unsafe cusparseCsr2cscEx2_bufferSize as csr2cscEx2_bufferSize { useHandle `Handle', `Int', `Int', `Int', useDevP `DevicePtr ()', useDevP `DevicePtr Int32', useDevP `DevicePtr Int32', useDevP `DevicePtr ()', useDevP `DevicePtr Int32', useDevP `DevicePtr Int32', cFromEnum `Type', cFromEnum `Action', cFromEnum `IndexBase', cFromEnum `Algorithm_csr2csc', alloca- `Int' peekIntConv* } -> `()' checkStatus*- #}+#else++csr2cscEx2 :: Handle -> Int -> Int -> Int -> DevicePtr () -> DevicePtr Int32 -> DevicePtr Int32 -> DevicePtr () -> DevicePtr Int32 -> DevicePtr Int32 -> Type -> Action -> IndexBase -> Algorithm_csr2csc -> DevicePtr () -> IO ()+csr2cscEx2 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = cusparseError "'csr2cscEx2' requires at least cuda-10.0"++csr2cscEx2_bufferSize :: Handle -> Int -> Int -> Int -> DevicePtr () -> DevicePtr Int32 -> DevicePtr Int32 -> DevicePtr () -> DevicePtr Int32 -> DevicePtr Int32 -> Type -> Action -> IndexBase -> Algorithm_csr2csc -> Int -> IO ()+csr2cscEx2_bufferSize _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = cusparseError "'csr2cscEx2_bufferSize' requires at least cuda-10.0" #endif

@@ -85,3 +85,22 @@   { underscoreToCase }   with prefix="CUSPARSE_SOLVE_POLICY" deriving (Eq, Show) #} +-- | Indices the algorithm to use for CSR to CSC matrix conversion+--+-- Algorithm 1 requires extra storage proportional to the number of nonzero+-- values @nnz@. It is in general faster than algorithm 2 and the result is+-- deterministic.+--+-- Algorithm 2 requires extra storage proportional to the number of rows @m@.+-- It is non-deterministic, and does not ensure always the same ordering of+-- CSC column indices and values. It is faster than algorithm 1 for regular+-- matrices.+--+#if CUDA_VERSION < 10010+data Algorithm_csr2csc+#else+{# enum cusparseCsr2CscAlg_t as Algorithm_csr2csc+  { underscoreToCase }+  with prefix="CUSPARSE" deriving (Eq, Show) #}+#endif+

@@ -151,7 +151,11 @@     , extraLibs      = extraLibs'     , extraGHCiLibs  = extraGHCiLibs'     , extraLibDirs   = extraLibDirs'+#if MIN_VERSION_Cabal(3,0,0)+    , options        = PerCompilerFlavor (if os /= Windows then ghcOptions else []) []+#else     , options        = [(GHC, ghcOptions) | os /= Windows]+#endif     , customFieldsBI = [c2hsExtraOptions]     } 

@@ -1,5 +1,5 @@ name:                   cusparse-version:                0.2.0.0+version:                0.3.0.0 synopsis:               FFI bindings to the CUDA Sparse BLAS library description:     The cuSPARSE library contains a set of basic linear algebra subroutines for@@ -18,7 +18,7 @@ license-file:           LICENSE author:                 Trevor L. McDonell maintainer:             Trevor L. McDonell <trevor.mcdonell@gmail.com>-copyright:              Copyright (c) [2017]. Trevor L. McDonell <trevor.mcdonell@gmail.com>+copyright:              Copyright (c) [2017..2018]. Trevor L. McDonell <trevor.mcdonell@gmail.com> category:               Foreign cabal-version:          >=1.24 @@ -86,6 +86,6 @@ source-repository this     type:               git     location:           https://github.com/tmcdonell/cusparse-    tag:                v0.2.0.0+    tag:                v0.3.0.0  -- vim: nospell