diff --git a/CHANGELOG.md b/CHANGELOG.md --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -10,8 +10,14 @@ because NVIDIA are A-OK introducing breaking changes in minor updates. -## [next] +## [0.10.2.0] - 2020-08-26 ### Added + * Add support for CUDA-10.2 + * Add support for Cabal-3 + * Add device properties for SM7.x, SM8 + +## [0.10.1.0] - 2019-04-29 +### Added * Add support for CUDA-10.1 ### Changed @@ -154,7 +160,8 @@ * Add functions from CUDA-6.5 -[next]: https://github.com/tmcdonell/cuda/compare/v0.10.0.0...HEAD +[0.10.2.0]: https://github.com/tmcdonell/cuda/compare/v0.10.1.0...v0.10.2.0 +[0.10.1.0]: https://github.com/tmcdonell/cuda/compare/v0.10.0.0...v0.10.1.0 [0.10.0.0]: https://github.com/tmcdonell/cuda/compare/0.9.0.3...v0.10.0.0 [0.9.0.3]: https://github.com/tmcdonell/cuda/compare/0.9.0.2...0.9.0.3 [0.9.0.2]: https://github.com/tmcdonell/cuda/compare/0.9.0.1...0.9.0.2 diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -20,5 +20,43 @@ For important information on installing on Windows, see: - + + + +## Missing functionality + +An incomplete list of missing bindings. Pull requests welcome! + +### CUDA-9 + +- cuLaunchCooperativeKernelMultiDevice + +### CUDA-10.0 + +- cuDeviceGetLuid (windows only?) +- cuLaunchHostFunc +- cuGraphHostNode[Get/Set]Params +- cuGraphKernelNode[Get/Set]Params +- cuGraphMemcpyNode[Get/Set]Params +- cuGraphMemsetNode[Get/Set]Params + +### CUDA-10.2 + +- cuDeviceGetNvSciSyncAttributes +- cuMemAddressFree +- cuMemAddressReserve +- cuMemCreate +- cuMemExportToShareableHandle +- cuMemGetAccess +- cuMemGetAllocationGranularity +- cuMemGetAllocationPrepertiesFromHandle +- cuMemImportFromShareableHandle +- cuMemMap +- cuMemRelease +- cuMemSetAccess +- cuMemUnmap +- cuGraphExecHostNodeSetParams +- cuGraphExecMemcpyNodeSetParams +- cuGraphExecMemsetNodeSetParams +- cuGraphExecUpdate diff --git a/Setup.hs b/Setup.hs --- a/Setup.hs +++ b/Setup.hs @@ -64,7 +64,6 @@ main = defaultMainWithHooks customHooks where readHook get_verbosity a flags = do - noExtraFlags a getHookedBuildInfo (fromFlag (get_verbosity flags)) preprocessors = hookedPreProcessors simpleUserHooks @@ -105,7 +104,6 @@ currentPlatform = hostPlatform lbi compilerId_ = compilerId (compiler lbi) -- - noExtraFlags args generateAndStoreBuildInfo verbosity profile @@ -196,7 +194,11 @@ , extraLibDirs = extraLibDirs' , frameworks = frameworks' , extraFrameworkDirs = frameworkDirs' +#if MIN_VERSION_Cabal(3,0,0) + , options = PerCompilerFlavor (if os /= Windows then ghcOptions else []) [] +#else , options = [(GHC, ghcOptions) | os /= Windows] +#endif , customFieldsBI = [c2hsExtraOptions] } diff --git a/cuda.cabal b/cuda.cabal --- a/cuda.cabal +++ b/cuda.cabal @@ -1,5 +1,5 @@ Name: cuda -Version: 0.10.1.0 +Version: 0.10.2.0 Synopsis: FFI binding to the CUDA interface for programming NVIDIA GPUs Description: The CUDA library provides a direct, general purpose C-like SPMD programming @@ -28,7 +28,7 @@ . * "Foreign.CUDA.Runtime" . - Tested with library versions up to CUDA-9.2. See also the + Tested with library versions up to CUDA-10.2. See also the build matrix for version compatibility. . @@ -65,7 +65,7 @@ Bug-reports: https://github.com/tmcdonell/cuda/issues Category: Foreign Cabal-version: >= 1.24 -Tested-with: GHC >= 7.8 +Tested-with: GHC >= 8.0 Build-type: Custom Extra-tmp-files: @@ -86,7 +86,7 @@ Library hs-source-dirs: src - Exposed-Modules: + exposed-modules: Foreign.CUDA Foreign.CUDA.Path Foreign.CUDA.Ptr @@ -138,16 +138,16 @@ -- Extras Foreign.C.Extra - Other-modules: + other-modules: Foreign.CUDA.Internal.C2HS Text.Show.Describe - Include-dirs: . - C-sources: cbits/stubs.c + include-dirs: . + c-sources: cbits/stubs.c cbits/init.c - Build-tools: c2hs >= 0.21 - Build-depends: + build-tools: c2hs >= 0.21 + build-depends: base >= 4.7 && < 5 , bytestring >= 0.10.4 , filepath >= 1.0 @@ -155,7 +155,7 @@ , uuid-types >= 1.0 default-language: Haskell98 - Extensions: + ghc-options: -Wall -O2 @@ -172,10 +172,10 @@ Executable nvidia-device-query - Main-is: DeviceQuery.hs + main-is: DeviceQuery.hs hs-source-dirs: examples/src/deviceQueryDrv - Build-depends: + build-depends: base >= 4 && < 5 , cuda , pretty @@ -190,6 +190,6 @@ source-repository this type: git location: https://github.com/tmcdonell/cuda - tag: v0.10.1.0 + tag: v0.10.2.0 -- vim: nospell diff --git a/src/Foreign/CUDA/Analysis/Device.chs b/src/Foreign/CUDA/Analysis/Device.chs --- a/src/Foreign/CUDA/Analysis/Device.chs +++ b/src/Foreign/CUDA/Analysis/Device.chs @@ -150,6 +150,8 @@ -- wikipedia entry: -- data Allocation = Warp | Block + deriving Show + data DeviceResources = DeviceResources { threadsPerWarp :: !Int -- ^ Warp size , coresPerMP :: !Int -- ^ Number of SIMD arithmetic units per multiprocessor @@ -168,6 +170,7 @@ , warpRegAllocUnit :: !Int -- ^ Warp register allocation granularity , maxGridsPerDevice :: !Int -- ^ Maximum number of resident grids per device (concurrent kernels) } + deriving Show -- | @@ -322,8 +325,7 @@ , maxGridsPerDevice = 16 } - Compute 7 0 -> resources (Compute 7 2) -- Volta GV100 - Compute 7 2 -> DeviceResources + Compute 7 0 -> DeviceResources -- Volta GV100 { threadsPerWarp = 32 , coresPerMP = 64 , warpsPerMP = 64 @@ -342,21 +344,35 @@ , maxGridsPerDevice = 128 } - Compute 7 5 -> DeviceResources -- Turing TU1xx - { threadsPerWarp = 32 - , coresPerMP = 64 - , warpsPerMP = 32 - , threadsPerMP = 1024 + Compute 7 2 -> (resources (Compute 7 0)) -- Volta GV10B + { maxGridsPerDevice = 16 + , maxSharedMemPerBlock = 49152 + } + + Compute 7 5 -> (resources (Compute 7 0)) -- Turing TU1xx + { warpsPerMP = 32 , threadBlocksPerMP = 16 + , threadsPerMP = 1024 + , maxGridsPerDevice = 128 , sharedMemPerMP = 65536 -- of 96KB , maxSharedMemPerBlock = 65536 + } + + Compute 8 0 -> DeviceResources -- Ampere GA100 + { threadsPerWarp = 32 + , coresPerMP = 64 + , warpsPerMP = 64 + , threadsPerMP = 2048 + , threadBlocksPerMP = 32 + , sharedMemPerMP = 167936 -- of 192KB + , maxSharedMemPerBlock = 163840 , regFileSizePerMP = 65536 , maxRegPerBlock = 65536 , regAllocUnit = 256 , regAllocationStyle = Warp , maxRegPerThread = 255 - , sharedMemAllocUnit = 256 - , warpAllocUnit = 2 + , sharedMemAllocUnit = 128 + , warpAllocUnit = 4 , warpRegAllocUnit = 256 , maxGridsPerDevice = 128 } diff --git a/src/Foreign/CUDA/Analysis/Occupancy.hs b/src/Foreign/CUDA/Analysis/Occupancy.hs --- a/src/Foreign/CUDA/Analysis/Occupancy.hs +++ b/src/Foreign/CUDA/Analysis/Occupancy.hs @@ -57,7 +57,7 @@ import Data.Ord import Data.List -import Foreign.CUDA.Analysis.Device +import Foreign.CUDA.Analysis.Device hiding ( maxSharedMemPerBlock ) -- GPU Occupancy per multiprocessor @@ -97,7 +97,8 @@ -- Physical resources -- - gpu = deviceResources dev + gpu = deviceResources dev + maxSharedMemPerBlock = fromIntegral (sharedMemPerBlock dev) -- 2080Ti reports 48KB, but should be 64KB ! -- Allocated resource limits -- @@ -116,16 +117,16 @@ -- Maximum thread blocks per multiprocessor -- - limitDueToWarps = threadBlocksPerMP gpu `min` (warpsPerMP gpu `div` limitWarpsPerBlock) + limitDueToWarps = threadBlocksPerMP gpu `min` (warpsPerMP gpu `div` limitWarpsPerBlock) limitDueToRegs - | regs > maxRegPerThread gpu = 0 - | regs > 0 = (limitRegsPerSM `div` limitRegsPerBlock) * (regFileSizePerMP gpu `div` maxRegPerBlock gpu) - | otherwise = threadBlocksPerMP gpu + | regs > maxRegPerThread gpu = 0 + | regs > 0 = (limitRegsPerSM `div` limitRegsPerBlock) * (regFileSizePerMP gpu `div` maxRegPerBlock gpu) + | otherwise = threadBlocksPerMP gpu limitDueToSMem - | smem > maxSharedMemPerBlock gpu = 0 - | smem > 0 = sharedMemPerMP gpu `div` limitSMemPerBlock - | otherwise = threadBlocksPerMP gpu + | smem > maxSharedMemPerBlock = 0 + | smem > 0 = sharedMemPerMP gpu `div` limitSMemPerBlock + | otherwise = threadBlocksPerMP gpu -- | diff --git a/src/Foreign/CUDA/Driver/Device.chs b/src/Foreign/CUDA/Driver/Device.chs --- a/src/Foreign/CUDA/Driver/Device.chs +++ b/src/Foreign/CUDA/Driver/Device.chs @@ -290,8 +290,6 @@ #endif --- TODO: cuDeviceGetLuid, introduced CUDA-10.0 (windows only) - -- | -- Return the properties of the selected device -- diff --git a/src/Foreign/CUDA/Driver/Exec.chs b/src/Foreign/CUDA/Driver/Exec.chs --- a/src/Foreign/CUDA/Driver/Exec.chs +++ b/src/Foreign/CUDA/Driver/Exec.chs @@ -352,8 +352,6 @@ } -> `Status' cToEnum #} #endif --- TODO: cuLaunchCooperativeKernelMultiDevice introduced CUDA-9.0 --- TODO: cuLaunchHostFunc introduced CUDA-10.0 -------------------------------------------------------------------------------- -- Deprecated diff --git a/src/Foreign/CUDA/Driver/Graph/Build.chs b/src/Foreign/CUDA/Driver/Graph/Build.chs --- a/src/Foreign/CUDA/Driver/Graph/Build.chs +++ b/src/Foreign/CUDA/Driver/Graph/Build.chs @@ -659,12 +659,6 @@ #endif --- TODO: since CUDA-10.0 --- * cuGraphHostNode[Get/Set]Params --- * cuGraphKernelNode[Get/Set]Params --- * cuGraphMemcpyNode[Get/Set]Params --- * cuGraphMemsetNode[Get/Set]Params - -------------------------------------------------------------------------------- -- Internal --------------------------------------------------------------------------------