Regardless CUDA or OpenCL... Will there be anything on the Mac side...? I'm sick all these developments meant for Windows and some for Linux, but none for Mac... My Pro's dying for more.
Since the developer previews of Snow Leopard are the only useful OpenCL development environment right now I would expect at least the first OpenCL Einstein@hone App to be issued for MacOS ;-)
If I understand right, devices "smaller" than GTX260/280 and Tesla S1060/C1060 cannot operate with double precision numbers and "truncate" operands to float precision type.
Nope. For these the double precision operations will be emulated by the software using multiple single precision operations, which is way slower than on GPUs supporting double precision on hardware, but will still give correct double precision results.
The core functions of the "HierarchicalSearch" are all single precision, there are a few variables that collect many small numeric values which will add up to a large error if simply switched to single precision. But in the code we use since S5R2 the use of such double precision variables has been purposefully reduced to a minimum, I don't think they will be the limiting factor.
Einstein support OpenCL ?
)
There are people working on both. AFAIK CUDA is somewhat further, maybe we get something working in March there.
BM
BM
RE: Regardless CUDA or
)
Since the developer previews of Snow Leopard are the only useful OpenCL development environment right now I would expect at least the first OpenCL Einstein@hone App to be issued for MacOS ;-)
BM
BM
RE: If I understand right,
)
Nope. For these the double precision operations will be emulated by the software using multiple single precision operations, which is way slower than on GPUs supporting double precision on hardware, but will still give correct double precision results.
The core functions of the "HierarchicalSearch" are all single precision, there are a few variables that collect many small numeric values which will add up to a large error if simply switched to single precision. But in the code we use since S5R2 the use of such double precision variables has been purposefully reduced to a minimum, I don't think they will be the limiting factor.
BM
BM