[BugFix][CUDA] Increase FloatImm precision when printing 64 bit values in CUDA codegen

apache / tvm

Open deep learning compiler stack for cpu, gpu and specialized accelerators

https://tvm.apache.org/

Apache License 2.0

11.81k stars 3.48k forks source link

[BugFix][CUDA] Increase FloatImm precision when printing 64 bit values in CUDA codegen #17530

Open SerodioJ opened 1 week ago

SerodioJ commented 1 week ago

This bugfix is related to issue #17276. The reduced precision when printing floating point constants in source code files, leads to undesired behavior, more noticeable while working with trigonometric functions.