We are currently investigating what workarounds could be automatically applied. Until then, you have to modify your code manually to circumvent the above limitations.
Translation of data directives: !$acc enter data, !$acc exit data, !$acc data.Synchronization directives: !$acc wait, !$acc update self/host/device.Kernel and loop constructs !$acc kernels plus !$acc loop in subsequent line, !$acc kernels loop, !$acc parallel plus !$acc loop.In subsequent line, !$acc parallel loop, !$acc loop Support for !$acc routine seq functions with scalar arguments.Majority of CUDA libary functionality via HIPFORT.Kernel and loop constructs: !$cuf kernel do.Overloaded intrinsics: allocate, allocated, deallocate, deallocated, =.Support for CUDA Fortran attributes(global) (array and scalar arguments),Īnd attributes(host,device), attributes(device) procedures (only scalar arguments supported for the latter).) Planned features (or: "more limitations") Initial support for !$acc declare (detected but not considered in codegen yet).Improve support for !$acc parallel (loop).Add support for !$acc parallel without !$acc loop in next line).Add support for !$acc kernels without !$acc loop in next line).Auto detection of offloadable code parts.use other compiler for syntax checks.AppGraphics comes with every copy of Simply Fortran.Īdd option for prescribing syntax checker (e.g.Support of derived types with allocatable, pointer members.Rewrite of GPUFORT Fortran runtime in (HIP) C++.
SIMPLY FORTRAN CUDA SUPPORT INSTALL
You don't need to install anything else, and you don't need an SF Package Manager subscription.
0 kommentar(er)
