gemini-code-assist[bot] commented on code in PR #59:
URL: https://github.com/apache/tvm-ffi/pull/59#discussion_r2382338643
##########
python/tvm_ffi/cython/tvm_ffi_python_helpers.h:
##########
@@ -424,9 +435,10 @@ class TVMFFIPyCallManager {
private:
TVMFFIPyCallManager() {
static constexpr size_t kDefaultDispatchCapacity = 32;
- static constexpr size_t kDefaultStackSize = 32;
+ // keep it 4K as default stack size so it is page aligned
+ static constexpr size_t kDefaultStackSize = 4096;
dispatch_map_.reserve(kDefaultDispatchCapacity);
- temp_stack_.resize(kDefaultStackSize * 2);
+ temp_stack_.resize(kDefaultStackSize / sizeof(TVMFFIAny));
Review Comment:
This change to `kDefaultStackSize` and the `resize` logic is a good
improvement, increasing the on-stack argument capacity and clarifying the code.
However, it appears to be a separate logical change from the main goal of this
PR, which is to add `always_inline` hints. For better maintainability and
clearer git history, it's recommended to split unrelated changes into separate
pull requests. This makes each PR easier to review and understand.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
