Failure to Build Against CUDA 13

[sempervictus]

Trying to build the qwen demo, ran into:

Expected symbol in library: DlSym { desc: "/opt/cuda/lib64/libnvrtc.so: undefined symbol: nvrtcGetNVVM" }
stack backtrace:
   0: __rustc::rust_begin_unwind
   1: core::panicking::panic_fmt
   2: core::result::unwrap_failed
   3: cudarc::nvrtc::sys::loaded::Lib::new
   4: std::sync::poison::once::Once::call_once_force::{{closure}}
   5: std::sys::sync::once::futex::Once::call
   6: std::sync::once_lock::OnceLock<T>::initialize
   7: cudarc::nvrtc::result::create_program
   8: cudarc::nvrtc::safe::compile_ptx_with_opts
   9: luminal_cuda::compile_and_load_kernel
  10: luminal_cuda::prim::CudaContiguous<T>::new
  11: <luminal_cuda::prim::PrimitiveCompiler<T> as luminal::compiler_utils::Compiler>::compile
  12: luminal::graph::Graph::compile
  13: qwen::main

package shows as:

Name            : cuda
Version         : 13.0.2-2
Description     : NVIDIA's GPU programming toolkit

the lib in question is:

Version definition section '.gnu.version_d' contains 2 entries:
 Addr: 0x0000000000003c60  Offset: 0x00003c60  Link: 5 (.dynstr)
  000000: Rev: 1  Flags: BASE  Index: 1  Cnt: 1  Name: libnvrtc.so.13
  0x0014: Rev: 1  Flags: none  Index: 2  Cnt: 1  Name: libnvrtc.so.13

Version needs section '.gnu.version_r' contains 5 entries:
 Addr: 0x0000000000003c90  Offset: 0x00003c90  Link: 5 (.dynstr)
  000000: Version: 1  File: ld-linux-x86-64.so.2  Cnt: 1
  0x0010:   Name: GLIBC_2.3  Flags: none  Version: 11
  0x0020: Version: 1  File: libm.so.6  Cnt: 1
  0x0030:   Name: GLIBC_2.2.5  Flags: none  Version: 7
  0x0040: Version: 1  File: libdl.so.2  Cnt: 1
  0x0050:   Name: GLIBC_2.2.5  Flags: none  Version: 5
  0x0060: Version: 1  File: libc.so.6  Cnt: 5
  0x0070:   Name: GLIBC_2.6  Flags: none  Version: 12
  0x0080:   Name: GLIBC_2.3.4  Flags: none  Version: 10
  0x0090:   Name: GLIBC_2.3  Flags: none  Version: 9
  0x00a0:   Name: GLIBC_2.7  Flags: none  Version: 6
  0x00b0:   Name: GLIBC_2.2.5  Flags: none  Version: 4
  0x00c0: Version: 1  File: libpthread.so.0  Cnt: 2
  0x00d0:   Name: GLIBC_2.3.2  Flags: none  Version: 8
  0x00e0:   Name: GLIBC_2.2.5  Flags: none  Version: 3

Any chance someone has a dockerfile in their back pocket to stabilize build/run targets a bit given the slew of CUDA environment setups out there?