oneAPI Deep Neural Network Library (oneDNN)  1.4.0
Performance library for Deep Learning
Inspecting JIT Code

oneDNN uses just-in-time compilation (JIT) to generate optimal code for some functions based on input parameters and instruction set supported by the system. The library provides a mechanism to save the generated code into a file for inspection.

This behavior can be enabled with DNNL_JIT_DUMP environment variable or dnnl_set_jit_dump function.

Value Behav
0 JIT dump is disabled (default)
any other value JIT dump is enabled

The function setting takes precedence over the environment variable.

Example

$ DNNL_JIT_DUMP=1 ./simple-net-cpp

This will produce the following output files if running on a CPU supporting Intel(R) Advanced Vector Extensions 2 (Intel(R) AVX2):

dnnl_dump_jit_avx2_conv_fwd_kernel_f32.1.bin
...
dnnl_dump_jit_avx_gemm_f32_xbyak_gemm.40.bin

Use any disassembler to view the code. For example:

XED is a decoder tool available as part as Intel Software Development Emulator (Intel SDE).