Move shared hook and threading-model types out of senseApiDesc.h so both
stimbuff and comparator libraries can include them without pulling in API descs.
Co-authored-by: Cursor <cursoragent@cursor.com>
Register comparator types via a dedicated manager and expose lookup/create
hooks through SmoCallbacks so stimbuff libs can resolve comparators at load time.
Co-authored-by: Cursor <cursoragent@cursor.com>
Introduce ExportedComparatorTypeDesc/ComparatorLibDesc with inline
sanity checks, and add a core comparator shlib exporting three stub types.
Co-authored-by: Cursor <cursoragent@cursor.com>
Move Comparator and ComparatorExpression out of cologex.h so comparator
types can be shared by loadable comparator libraries without pulling in
the full cologex surface.
Co-authored-by: Cursor <cursoragent@cursor.com>
SenseApiDesc's exported API now uses coro pointers instead of
CPS fn pointers.
* Do not build this version of SMO with the Livox drivers enabled,
because SMO has been changed at the smocore level to use coros
when calling into stimbuffAPI libs. But the Livox drivers
haven't yet been ported from CPS to coros.
xcbWindow has been ported to expose coros to SMO in its
senseApiDesc exported iface.
We haven't ported everything. Just the top-level methods. We'll
dig in to the leaf stuff later. Surprisingly, this all went without
any real difficulties.
Runs like a charm on first try.
We now specify intrins as separate DAPS lines. This syntax is much
nicer and well-grouped than the previous negtrin-*/postrin-* param
names.
Alas, we're about to replace it in the next few commits already though.
We use io_uring_register_buffers() for IoUringAssemblyEngine instead
of using mlock(). This __appears__ to have reduced CPU utilization on
the Dell laptop. Could also be that we recently upgraded total RAM
from 8GiB to 32GiB.
We directly use an instance of RangeDescriptor to avoid incurring
the memory cost of using a StagingBuffer here. It's unnecessary
since these stencils will always be 32bits large.
These two classes represent our first foray into stencil
construction. One of them standardizes PcloudAmbience stencils
across all stimbuffs, and the other specifies the internal
memory constraints and requirements for a LivoxGen1 device's
stencils.
Reduces code duplication, centralizes checking and enforces consistent
behaviour across producers.
Also reordered the writes to the sh_ptr<StimulusBuffer>s such that
the pointers are written last.
Slots whose stride size is larger than the slot alignment value
should have their size rounded up to the alignment size so that
the slots that follow them will also be aligned.
We added a new centralized OpenCL Compute manager. This can later
be extended to support CUDA, SyCL, etc. SMO can be configured at
build time to choose which API it will use for compute.
Moreover, the ComputeMgr allows us to register buffers which are
available to all cl_contexts.
We now allocate all the stimFrames for a StimBuffer using a
single StagingBuffer. This gives us all the benefits we're
looking for (pinning, alignment, etc).
hayodea