In order for interrupt-based wake-ups (as introduced by #1142) to work concurrently with time.Sleep, we need to make some per-board changes.

Previously, sleepTicks (the function used as an interface between the scheduler and the hardware timer) was defined to block until the timer completed, since there was nothing else to do. Now we need to change this so that it bails out when an interrup

PR #6447 adds a public API to get the maximum number of registers per thread (numba.cuda.Dispatcher.get_regs_per_thread()). There are other attributes that might be nice to provide - shared memory per block, local memory per thread, const memory usage, maximum block size.

These are all available in the FuncAttr named tuple: https://github.com/numba/numba/blob/master/numba/cuda/cudadrv/drive

Hey everyone!

mapd-core-cpu is already available on conda-forge (https://anaconda.org/conda-forge/omniscidb-cpu)

now we should add some instructions on the documentation.

at this moment it is available for linux and osx.

some additional information about the configuration:

1. for now, always install omniscidb-cpu inside a conda environment (also it is a good practice), eg:

While the test_suite presented in #589 does work, it is still pretty simple and can be improved and enhanced.

• Add tests for gnutils and coreutils.
• Add necessary utilities so more complex programs can be compiled from sources.
• Integrate CMake, so one could write something like make validate and the subset of test that is deemed necessary (for example everything with min t

Scenario:
I have a project, with build tools of cmake, when I try to run it with cling. It prompts some headers are not included.

So I was wondering if there is a way to make use of the cmake script to load the build environment into cling and run directly?

Appreciate the great work!

Currently, the architecture of the CLI is based on (sub)commands and options. Commands are expected to be provided as the first argument, and do effectively decide which feature is to be used. OTOH, options provide parameters to the commands. However, there is no syntactical difference, as both commands and options start with -- or -i. As a result, we rely on properly formating --help and on

enable cyclic-import check in pylint and fix the imports if there are problems

LLVM will soon support adding per-function attributes about available C library functions. We should add support for that, to solve issues like a memcpy implementation being replaced by a memcpy call: https://forum.dlang.org/thread/clptjumxigcozfcyhzzx@forum.dlang.org
Probably ldc.attributes.llvmAttrwill already work for this, but I think it'd be nice to have a more userfriendly attribute.
Som

Refactor host -> guest context switch assembly codes that have been duplicated when we introduced AVX support.