Skip to content

cp async fallback #761

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
alihassanijr wants to merge 2 commits into from
Closed

cp async fallback #761

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5c03690
Adds missing semicolon
alihassanijr Jan 9, 2023
6e701f5
Partial specializations for cp_async_zfill
alihassanijr Jan 10, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
78 changes: 78 additions & 0 deletions include/cutlass/arch/memory_sm80.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,6 +104,40 @@ static const uint32_t OOB_NAN_F16x2 = ((OOB_NAN_F16 << 16) | OOB_NAN_F16);

////////////////////////////////////////////////////////////////////////////////////////////////////

/// Fallback specialization for 1 byte copies
template <
/// Cache operation - ignored
CacheOperation::Kind cache_op>
struct cp_async<1, cache_op> {

/// Copy
CUTLASS_DEVICE
cp_async(void *smem_ptr, void const *global_ptr, bool pred_guard = true) {
using AccessType = Array<uint8_t, 1>;

if (pred_guard) {
*static_cast<AccessType *>(smem_ptr) = *static_cast<AccessType const *>(global_ptr);
}
}
};

/// Fallback specialization for 2 byte copies
template <
/// Cache operation - ignored
CacheOperation::Kind cache_op>
struct cp_async<2, cache_op> {

/// Copy
CUTLASS_DEVICE
cp_async(void *smem_ptr, void const *global_ptr, bool pred_guard = true) {
using AccessType = Array<uint8_t, 2>;

if (pred_guard) {
*static_cast<AccessType *>(smem_ptr) = *static_cast<AccessType const *>(global_ptr);
}
}
};

/// Partial specialization
template <
/// Size of the access in bytes
Expand Down Expand Up @@ -143,6 +177,50 @@ struct cp_async<SizeInBytes, CacheOperation::Always> {
}
};

/// Fallback specialization for 1 byte copies
template <
/// Cache operation - ignored
CacheOperation::Kind cache_op>
struct cp_async_zfill<1, cache_op> {

/// Copy with zero fill
CUTLASS_DEVICE
cp_async_zfill(void *smem_ptr, void const *global_ptr, bool pred_guard) {
using AccessType = Array<uint8_t, 1>;

if (pred_guard) {
*static_cast<AccessType *>(smem_ptr) = *static_cast<AccessType const *>(global_ptr);
}
else {
AccessType zeros;
zeros.clear();
*static_cast<AccessType *>(smem_ptr) = zeros;
}
}
};

/// Fallback specialization for 2 byte copies
template <
/// Cache operation - ignored
CacheOperation::Kind cache_op>
struct cp_async_zfill<2, cache_op> {

/// Copy with zero fill
CUTLASS_DEVICE
cp_async_zfill(void *smem_ptr, void const *global_ptr, bool pred_guard) {
using AccessType = Array<uint8_t, 2>;

if (pred_guard) {
*static_cast<AccessType *>(smem_ptr) = *static_cast<AccessType const *>(global_ptr);
}
else {
AccessType zeros;
zeros.clear();
*static_cast<AccessType *>(smem_ptr) = zeros;
}
}
};

/// Partial specialization
template <
/// Size of the access in bytes
Expand Down
2 changes: 1 addition & 1 deletion include/cutlass/arch/mma.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,7 +70,7 @@ struct OpMultiplyAddFastF16 {};

/// Tag indicating the input is converted to 2 (big and small) TF32 components
// Perform 3xTF32 or 4xTF32 for every F32 output element
struct OpMultiplyAddFastF32 {}
struct OpMultiplyAddFastF32 {};

/// Tag indicating the input is converted to 2 (big and small) TF32 components
// Perform 3xTF32 or 4xTF32 for every complex<F32> output element
Expand Down