A bit more information on the upcoming PGI CUDA x86 C compiler is over at GPU Science, including some important details like:
- It’s a C compiler, to complement their existing CUDA Fortran offerings.
- Works with AMD and Intel CPUs
- Utilizes SIMD streaming capabilities
And does it all at run-time, meaning a single universal binary can run with or without a CUDA-compliant GPU. See Update Below.
“CUDA C for x86 is a perfect complement to CUDA Fortran and PGI’s optimizing parallel Fortran and C compilers for multi-core x86,” said Douglas Miles, director, The Portland Group. “It’s another important element in our on-going strategy of providing HPC programmers with development tools that give PGI users a full range of options for optimizing compute-intensive applications, while allowing them to leverage the latest technical innovations from AMD, Intel and NVIDIA.”
Seems they’ll have it at the SC10 exhibit, but there are no details yet on availability or pricing.
Update 9/23 1pm: I just spoke with NVidia’s Sumit Gupta and found out I was mistaken on this. The PGI compiler will be a CPU-only compiler, turning CUDA instructions into CPU-executing code. The benefit being that people new to HPC and parallel algorithms can use the breadth of information on CUDA, which is far simpler than most other models, to write their code and run it right on CPU’s. For those people who already have GPU-based systems, they’ll be able to recompile for CPU-only without any modification of their code, and enable use of possibly underutilized CPU-only clusters.
via PGI to Demonstrate New PGI CUDA C Compiler at SC10 Supercomputing | GPU Science.
