Version 0.6.4 : Minor changes and bug fixes

[eyalroz]

Changes since v0.6.3:

Features

• mapped memory allocation currently insists on CU_MEMHOSTALLOC_DEVICEMAP #528 Can now register memory areas with the CUDA driver as read-only (in addition to other parameters)
• Support use of external memory resources #519 Support use of "external memory resources"

Bug fixes

• Add support for the read-only attribute for cuda::memory::host::register() #527 copy_parameters_t::set_single_context no longer mistakenly taking an endpoint_t parameter
• mapped memory allocation currently insists on CU_MEMHOSTALLOC_DEVICEMAP #528 When mapping a region pair, not insisting on the same address on both the host and the device (which had made it impossible for this to succeed with older GPUs).
• Reduce boilerplate + avoid warning when overriding context for the current scope #521 Avoid a compiler compiler warning when overriding the context for the current scope
• Don't use a context-wrapper scoped-context-override when unnecessary #520 Removed unnecessary uses of a context-wrapper scoped-context-override
• typed_set mistakenly accepts values of size 8 #517 cuda::memory::typed_set() no longer mistakenly accepts values of size 8 (which have no special CUDA API call).
• Use explicit cast and correct types for write_single_value #516 Corrected types and casting in stream_t::enqueue_t::write_single_value()
• Missing includes #515 Resolved a case of missing includes when including only certain headers
• Definitions of cuda::memory::managed::allocate(device, num_bytes) #514 Now providing a definition of cuda::memory::managed::allocate(device, num_bytes) (which was declared but not defined)

Other changes

• Rename synch_ to sync_ on everything #522 Renamed synch to sync in multiple identifiers
• program_t::add_registered_globals should be able to take a container of any string-like type #529 program_t::add_registered_globals() can now take any container of any string-like type.
• Pass device_t's by const-ref more #523 Now passing device_t's by const-ref in more cases, avoiding copying and enabling re-use of a reference to the primary context.
• Reduce boilerplate + avoid warning when overriding context for the current scope #521 Reduce boilerplate + avoid warning when overriding context for the current scope resolved-on-development task

Build issues

• Build fails with cooperative_groups on GitHub Actions Windows runners with CUDA >= 11.7 #504 Fixed build failure with cooperative_groups on GitHub Actions Windows runners and CUDA >= 11.7

---

This discussion was created from the release Version 0.6.4 : Minor changes and bug fixes.