Fortran Compiler

I’ve read online that there is a CUDA Fortran compiler coming out by the end of the year ( I saw people talking about it online, but I didn’t see an actual date. Does anyone know if/when a Fortran compiler is coming out? And will it allow to write pure CUDA programs without having to write C functions?

I have seen Fortran support on some slides presented at NVISION, but there was no timeline mentioned there, personally I would be surprised if it was already in 2.1 Do you already know about Flagon? Otherwise I would take a look at that first.

There isn’t a Fortran kernel language coming out this year; it’s not in 2.1.

xHelios, I think you read about the FORTRAN library FLAGON.
That’s not a native FORTRAN->CUDA compiler, it’s a FORTRAN library for using CUDA inside FORTRAN (which is in CUDA’s C-like language).

It’s at:

Here’s a blurb from

The PGI 8.0 release from The Portland Group includes a technology preview of the PGI accelerator programming strategy. PGI 8.0 compilers accept new directives that allow users to select compute intensive regions of Linux x64 Fortran and C99 programs and automatically offload them to an NVIDIA GPU. Until now HPC developers targeting GPU accelerators have had to rely on libraries or language extensions, and use of GPUs from Fortran has been extremely limited. Using the provisional support in PGI Release 8.0, programmers can accelerate Linux applications on x64+NVIDIA platforms by adding OpenMP-like compiler directives to existing high-level standard- compliant Fortran and C99 programs. At Supercomputing 2008 you can see the PGI x64+GPU compilers in action, and learn about PGI’s accelerator programming model and how you can use it to experiment with and embrace accelerated computing. You can also attend the PGI Vendor presentation by Michael Wolfe in room 19A/19B of the Austin convention center on Wednesday, November 19 from 10:30-11:00AM.

Thanks a lot for your help!