# Conflicts:
#	Compiler/src/exastencils/baseExt/ir/IR_ContainerDatatypeSTL.scala
#	Compiler/src/exastencils/communication/ir/IR_LocalRecv.scala
#	Compiler/src/exastencils/communication/ir/IR_LocalSend.scala
#	Compiler/src/exastencils/communication/ir/IR_RemoteCommunicationFinish.scala
#	Compiler/src/exastencils/communication/ir/IR_RemoteCommunicationStart.scala
#	Compiler/src/exastencils/communication/ir/IR_RemoteSend.scala
#	Compiler/src/exastencils/timing/ir/IR_AutomaticFunctionTiming.scala

Improve existing timing infrastructure

See merge request exastencils/exastencils!148

Encapsulate cuda device count and property getters and add corresponding IVs

See merge request exastencils/exastencils!151

Richard Angersbach authored 4 months ago and Richard Angersbach committed 4 months ago

(cherry picked from commit a1ded09b)

 Conflicts:
	Compiler/src/exastencils/baseExt/ir/IR_LoopOverDimensions.scala
	Compiler/src/exastencils/parallelization/api/cuda/CUDA_Flags.scala
	Compiler/src/exastencils/parallelization/api/cuda/CUDA_HandleFragmentLoops.scala
	Compiler/src/exastencils/parallelization/api/cuda/CUDA_MemoryTransfer.scala
	Compiler/src/exastencils/parallelization/api/cuda/CUDA_PrepareFragmentLoops.scala

CUDA: Fix reduction of fragment copies in LoopOverDimensions (host)

See merge request exastencils/exastencils!142

CUDA: Wait until packing completion before MPI statement is called

See merge request exastencils/exastencils!143

Consider min OMP work items for fragment loops

See merge request exastencils/exastencils!150

CUDA: Fix dirty flag handling

See merge request exastencils/exastencils!147

Adjusted python scripts accordingly

Add transferred buffers/fields that are not modified by kernel to intermediate flag to avoid re-transfer of unused data

Add transferred buffers/fields that are not modified by kernel to intermediate flag to avoid re-transfer of unused data