Includes changes equivalent to https://github.com/inducer/pycuda/pull/423



Co-authored-by: Antoine Martin <antoine@xpra.org>

* feat: support cl_amd_copy_buffer_p2p

* clean: shorter line

* feat: use optional for byte_count && handle failure

* feat: add test

* feat: add docs

* fix: flake8

* feat: copy function signature over

* feat: remove suffix?

* feat: change version added

Andreas Klöckner authored 1 year ago and Andreas Klöckner committed 1 year ago

The chance of immediate reuse is low, but the chance of memory clutter is high

* runtime_program.rst: small fix for program.kernel_name

* Explain why `prg.sum(queue,...)` is discouraged for enqueue

* Further doc improvements on recommended approach for kernel enqueue

---------

Co-authored-by: Andreas Klöckner <inform@tiker.net>

* fix and move print-binary.py

* remove the file

Andreas Klöckner authored 2 years ago and Andreas Klöckner committed 2 years ago

https://github.com/inducer/compyte/pull/43

* Enable extended alignment for complex types with a define

* Do not use reserved C identifiers

See https://wiki.sei.cmu.edu/confluence/display/c/DCL37-C.+Do+not+declare+or+define+a+reserved+identifier

* add a comment that this is undocumented