You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Due to mutiple iterations, the the JitKernel class became messy. PR #896
There are at least two levels of caching for the same unspecialized kernel, i.e. self._cache and self._kernel_bundle_cache
For the specialized kernel, the caching is done in three individual levels. i.e. self._cache, self._kernel_bundle_cache and self._specialization_cache
These multiple levels of LRU caching is causing unnecessary overhead. Everything in that class needs to be made streamlined and clean. e.g. self._cache and self._kernel_bundle_cache can be consolidated.
Also due to these reasons, the cache-hit/miss counts (self._cache_hits) became ambiguous.
Either _check_range() and _check_ndrange() functions need to be moved inside Range, NdRange class or moved into the numba_dpex.core.kernel_interface.utils module.
The text was updated successfully, but these errors were encountered:
Due to mutiple iterations, the the
JitKernel
class became messy. PR #896self._cache
andself._kernel_bundle_cache
self._cache
,self._kernel_bundle_cache
andself._specialization_cache
self._cache
andself._kernel_bundle_cache
can be consolidated.self._cache_hits
) became ambiguous._check_range()
and_check_ndrange()
functions need to be moved insideRange
,NdRange
class or moved into thenumba_dpex.core.kernel_interface.utils
module.The text was updated successfully, but these errors were encountered: