Not moving to BlueSky even if the #hpc community is trying to go there. But then I probably do social media all wrong.
I've seen things you people wouldn't believe. Jobs running in the GPU queue without requesting any GPUs. I watched tape drives storing 10,000 files of 100 bytes each. All those moments will be timed out, with no possibility for extensions. Time to sudo kill -9.
@jobs I've been visiting the Wellcome Center this week. Lovely campus, although very difficult to reach without a car. #HPC
via @hpcnotes :
The #1 supercomputer on the Top500 is Frontier at Oak Ridge National Lab, which packs 1.194 Exaflops on HPL and apparently cost $600M.
The aggregate HPL performance of the Top500 is 5.24 Exaflops.
So, extrapolating Frontier's cost/EF, that is a total of approximately $2.6B for all of the Top500 systems.
Interested in contributing to science by writing code? Take a look at the "Good First Issues" board here, or add your own repository if you maintain scientific software!
If you are involved with a High Performance Computing cluster that uses GPU's, could you quickly answer two questions on GPU monitoring on your machines?
Just compiling up a version of NCBI Blast for use on a large node. Anyone know why it looks for Boost/LaPack/GSL but doesn't require them? Will it be faster if I make them available?
Lawrence Livermore training materials
#introduction Hi! I am a research software engineer at Indiana University. I develop the CAFE application, an analyzer of gene family #evolution and support the software on IU's #HPC clusters. I like #bikes #cities and #opensource software.