oneapi-agent-toolkit

GPU test: softmax_hecbench

step status summary
reference_build ok exit 0: g++ -O2 -std=c++17 -I/tmp/tmp.H4U7IjIsE5/task /tmp/tmp.H4U7IjIsE5/task/softmax_ref.cpp -o build/reference
native_build ok exit 0: /opt/rocm/bin/hipcc -O2 -std=c++17 –offload-arch=gfx950 -I/tmp/tmp.H4U7IjIsE5/task /tmp/tmp.H4U7IjIsE5/out/work/softmax.hip.cpp -o build/native
native_verify_vs_reference ok PASS: 1/1 case(s) match (worst max abs err 5.25e-07, atol 1e-06)
sycl_compile ok compiled 1 source(s) -> /tmp/tmp.H4U7IjIsE5/out/work/build/sycl (0 warning(s))
sycl_verify_vs_reference ok PASS: 1/1 case(s) match (worst max abs err 9.69e-08, atol 1e-06)
sycl_verify_vs_native ok PASS: 1/1 case(s) match (worst max abs err 4.69e-07, atol 1e-06)
sycl_benchmark ok 10 run(s) each; hip:0: median 183.52 ms, p95 184.10 ms
native_benchmark ok 10 run(s) each; none: median 155.96 ms, p95 160.09 ms

Limitations: