vinx13 commented on a change in pull request #4546: [CODEGEN] Support cuda
tensorcore subbyte int data type in auto tensorcore
URL: https://github.com/apache/incubator-tvm/pull/4546#discussion_r380751172
##########
File path: include/tvm/runtime/data_type.h
##########
@@ -231,7 +231,15 @@ inline int GetVectorBytes(DataType dtype) {
int data_bits = dtype.bits() * dtype.lanes();
// allow bool to exist
if (dtype == DataType::Bool()) return 1;
- return data_bits;
+ // allow int4/uint4/int1 to exist
+ if (dtype == DataType::Int(4) ||
+ dtype == DataType::UInt(4) ||
+ dtype == DataType::Int(1)) {
+ return data_bits;
Review comment:
why date_bits here, instead of bytes? is there any trick here?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
