feat: Add metax build for cuda script#40

Open

baominghelly wants to merge 1 commit into

master from

feat/metax_hardware

Open

feat: Add metax build for cuda script #40
baominghelly wants to merge 1 commit into
master from
feat/metax_hardware

Conversation

@baominghelly

@baominghelly baominghelly commented May 25, 2026

Copy link

Copy Markdown

Collaborator

Description

feat: Add metax build for cuda script

Test

Test on nvidia and metax platform
MetaX

root@test:/home/libaoming/workplace/infinibench/InfiniBench# CUDA_VISIBLE_DEVICES=1 python main.py format_input_metax.json
2026年05月25日 04:32:17,562 - infinimetrics.utils.input_loader - INFO - Loaded 1 input(s) from format_input_metax.json
2026年05月25日 04:32:17,562 - infinimetrics.dispatcher - INFO - Processing 1 valid inputs (skipped 0 invalid)
2026年05月25日 04:32:17,563 - infinimetrics.dispatcher - INFO - Validation complete: 1 valid, 0 skipped
2026年05月25日 04:32:17,563 - infinimetrics.dispatcher - INFO - [1/1] Executing hardware.cudaUnified.Comprehensive
2026年05月25日 04:32:17,563 - infinimetrics.executor - INFO - Executor: Running hardware.cudaUnified.Comprehensive
2026年05月25日 04:32:17,564 - infinimetrics.hardware.hardware_adapter - INFO - HardwareTestAdapter: Processing hardware.cudaUnified.Comprehensive
2026年05月25日 04:32:17,564 - infinimetrics.hardware.hardware_adapter - INFO - GPU mode (device=metax): Executing CUDA tests
2026年05月25日 04:32:17,564 - infinimetrics.hardware.hardware_adapter - INFO - Executing: /home/libaoming/workplace/infinibench/InfiniBench/infinimetrics/hardware/cuda-memory-benchmark/build/cuda_perf_suite --all --array-size 67108864
2026年05月25日 04:35:29,641 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_h2d_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,642 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_d2h_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,642 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_d2d_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,643 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_bidirectional_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,644 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/cache_l1_bandwidth_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,644 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/cache_l2_bandwidth_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_043529.csv
2026年05月25日 04:35:29,645 - infinimetrics.executor - INFO - Executor: hardware.cudaUnified.Comprehensive completed in 192.08s
2026年05月25日 04:35:29,667 - infinimetrics.dispatcher - INFO - Summary saved to summary_output/dispatcher_summary_20260525_043529.json
2026年05月25日 04:35:29,667 - infinimetrics.dispatcher - INFO - Git info: commit=8aabd9ad, branch=metax_hardware
============================================================
Test Summary
============================================================
Total tests: 1
Successful: 1
Failed: 0
Success rate: 100.0%
============================================================

Nvidia

 python main.py format_input_hardware.json
2026年05月25日 07:59:55,583 - infinimetrics.utils.input_loader - INFO - Loaded 1 input(s) from format_input_hardware.json
2026年05月25日 07:59:55,583 - infinimetrics.dispatcher - INFO - Processing 1 valid inputs (skipped 0 invalid)
2026年05月25日 07:59:55,585 - infinimetrics.dispatcher - INFO - Validation complete: 1 valid, 0 skipped
2026年05月25日 07:59:55,586 - infinimetrics.dispatcher - INFO - [1/1] Executing hardware.cudaUnified.Comprehensive
2026年05月25日 07:59:55,586 - infinimetrics.executor - INFO - Executor: Running hardware.cudaUnified.Comprehensive
2026年05月25日 07:59:55,586 - infinimetrics.hardware.hardware_adapter - INFO - HardwareTestAdapter: Processing hardware.cudaUnified.Comprehensive
2026年05月25日 07:59:55,586 - infinimetrics.hardware.hardware_adapter - INFO - GPU mode (device=nvidia): Executing CUDA tests
2026年05月25日 07:59:55,586 - infinimetrics.hardware.hardware_adapter - INFO - Executing: /data/shared/baoming/workplace/infinibench/InfiniBench/infinimetrics/hardware/cuda-memory-benchmark/build/cuda_perf_suite --all --array-size 67108864
2026年05月25日 08:03:01,913 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_h2d_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,914 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_d2h_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,916 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_d2d_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,917 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/mem_sweep_bidirectional_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,919 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/cache_l1_bandwidth_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,921 - infinimetrics.common.csv_utils - INFO - CSV saved: output/hardware/cache_l2_bandwidth_hardware.cudaUnified.Comprehensive.fadfsdf_20260525_080301.csv
2026年05月25日 08:03:01,922 - infinimetrics.executor - INFO - Executor: hardware.cudaUnified.Comprehensive completed in 186.34s
2026年05月25日 08:03:01,948 - infinimetrics.dispatcher - INFO - Summary saved to summary_output/dispatcher_summary_20260525_080301.json
2026年05月25日 08:03:01,948 - infinimetrics.dispatcher - INFO - Git info: commit=2d73e5b2, branch=feat/metax_hardware
============================================================
Test Summary
============================================================
Total tests: 1
Successful: 1
Failed: 0
Success rate: 100.0%
============================================================