* build: add csignal header to rpc.h
- Include csignal for signal handling in RPC module
Signed-off-by: staryxchen <staryxchen@tencent.com>
* build(benchmark): set RPATH for tebench executable
- Configure tebench target to use relative RPATH for runtime library resolution
- Ensure libasio.so can be found at runtime from the ../lib director
Signed-off-by: staryxchen <staryxchen@tencent.com>
* feat(benchmark): support graceful interruption during execution
- Update runners to return error when stopped
- Add early exit logic in main loop
Signed-off-by: staryxchen <staryxchen@tencent.com>
* feat(benchmark): support specifying single GPU ID or all GPUs
- Update flags to accept -1 for all GPUs.
- Adjust buffer allocation and address calculation for specific GPU selection.
Signed-off-by: staryxchen <staryxchen@tencent.com>
* refactor(benchmark): unify buffer allocation logic for DRAM and VRAM
- Extract common logic for device prefix and buffer count
- Merge allocation and registration loops to reduce code duplication
- Improve variable const-correctness
Signed-off-by: staryxchen <staryxchen@tencent.com>
* feat(benchmark): add rpc server port configuration
- Add`rpc_server_port`field to`XferBenchConfig`
- Load port value from command line flags
Signed-off-by: staryxchen <staryxchen@tencent.com>
* Update mooncake-transfer-engine/benchmark/tent_backend.cpp
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
* refactor(benchmark): simplify gpu id validation and offset calculation
- Replace ternary operators with`std::max`for offset calculation
- Update assertions to check for non-negative GPU IDs
Signed-off-by: staryxchen <staryxchen@tencent.com>
* retrigger CI
Signed-off-by: staryxchen <staryxchen@tencent.com>
---------
Signed-off-by: staryxchen <staryxchen@tencent.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Stary