-
Notifications
You must be signed in to change notification settings - Fork 3.7k
Closed
Description
I have build tvm with CUDA and LLVM support(LLVM_CONFIG = /mypath/llvm-config). when I use nnvm.compiler.build(..., target="llvm", ...), there is no problem. But if I use target="cuda" then report error:attribute after last parameter! i32 (%0*, i8*, i32*, i8*, i8*) @fuse_sqrt_elemwise_add_compute_ LLVM ERROR: Broken function found, compilation aborted!
When I build tvm only with CUDA, there is no error.
Anyone can help me? Thank you!
Code: I use the example of Get Started with NNVM.
Metadata
Metadata
Assignees
Labels
No labels