[TVM Discuss] [Questions] Break in C++ shared library in gdb (or lldb) from python

Hi, I'm currently working on TVM to add new operator and custom backend. I'm wondering if we could debug a C++ shared library in gdb or any other debugger from python since TVM provides most their interface throughout python language. I tried `gdb --args python-dbg {my-script}` but it cannot st

[TVM Discuss] [Questions] Need help reading Relay IR

I get relay IR printed by call build_module.lower() : // func_name is "fused_sqrt_2" produce T_sqrt { parallel (ax0, 0, 128) { T_sqrt[ax0] = sqrt(placeholder[ax0]) } } I want to know what really happens with "parallel (ax0, 0, 128)" , will there be 128 tasks to be tun at runtime?

[TVM Discuss] [Questions] XGBoost's cuda acceleration

It seems [XGBoost supports GPU acceleration](http://tracking.discuss.tvm.ai/tracking/click?d=16FMB7EwcxvJDCA2R-NliyJmm7vGiGsXVdu32-HbyXzgzHrax6cTTZF8vPk3tcPUdOYhHHQGI8McfylgvP47UvwPiIJFsNZq28iWJAHqZiWQAUNj2QyjvxwXmLYmOoAbUc_Qx_XPrgLlsOX54dR0pLn7p1ZfVXr664BMoqjouLIT0) via cuda (9?) with the `gpu