[Info-vax] OT: Intel dusts off old supercomputing technology

[Michael Kraemer]

Michael Moroney schrieb:

> It wasn't a question about numbers (I'm fully aware about the number of PC
> GPUs vs. VMS systems), it was a question on how easy it must be to
> implement a client on VMS vs. GPUs, although the motivation to tap that
> admittedly large supply of CPU horsepower would be rather high.  A PC
> graphics card doesn't exactly come with a run time library suitable to run
> number crunchers, and I'm sure loading an actual program onto, and getting
> results from, a graphics card requires some Windoze hackery.  Meanwhile,
> an OS that was originally designed as a number cruncher in part doesn't
> have a client.  Seti at home was dropped since the newer BOINC code wasn't
> easily ported.

The days VMS was used for raw number crunching are long gone
(did it ever make it in the upper half of the Top 500 ?)
and somehow I doubt that it was better suited for this
purpose than any other OS.
OTOH the number crunching crowd has looked on GPUs
for quite some time now. A couple of months ago I had
a conversation with a guy using the Roadrunner installation.
He told me that GPUs are at least partially attractive,
but will have a power consumption/dissipation problem when
installed in masses. Hence for dedicated HPC installations,
IBM Cell Blades or Bluegenes would be the better solution.
This does not hold for GPUs in home PCs distributed all over the planet.

All that said, it is well possible that you will find
more people familiar with GPU hacks than with intimate
VMS knowledge these days. I'm almost sure the necessary
GPU "libraries" already exist for quite some time now.
And again, the sheer number of GPUs vs the number of
home VMS systems (which hardly exceeds a few thousands)
gives a strong incentive.