From 1b37cb0c03665294c7955b619a4625fae35dcef2 Mon Sep 17 00:00:00 2001 From: WJJ1995 Date: Fri, 17 Mar 2023 17:30:47 +0800 Subject: [PATCH] [Benchmark]Add End2End and meminfo configs (#1645) * add GPL lisence * add GPL-3.0 lisence * add GPL-3.0 lisence * add GPL-3.0 lisence * support yolov8 * add pybind for yolov8 * add yolov8 readme * add cpp benchmark * add cpu and gpu mem * public part split * add runtime mode * fixed bugs * add cpu_thread_nums * deal with comments * deal with comments * deal with comments * rm useless code * add FASTDEPLOY_DECL * add FASTDEPLOY_DECL * fixed for windows * mv rss to pss * mv rss to pss * Update utils.cc * use thread to collect mem * Add ResourceUsageMonitor * rm useless code * fixed bug * fixed typo * update ResourceUsageMonitor * fixed bug * fixed bug * add note for ResourceUsageMonitor * deal with comments * add macros * deal with comments * deal with comments * deal with comments * re-lint * rm pmap and use mem api * rm pmap and use mem api * add mem api * Add PrintBenchmarkInfo func * Add PrintBenchmarkInfo func * Add PrintBenchmarkInfo func * deal with comments * fixed enable_paddle_to_trt * add log for paddle_trt * support ppcls benchmark * use new trt option api * update benchmark info * simplify benchmark.cc * simplify benchmark.cc * deal with comments * Add ppseg && ppocr benchmark * add OCR rec img * add ocr benchmark * fixed trt shape * add trt shape * resolve conflict * add ENABLE_BENCHMARK define * Add ClassifyDiff * Add Resize for ClassifyResult * deal with comments * add convert info script * resolve conflict * Add SaveBenchmarkResult func * fixed bug * fixed bug * fixed bug * add config.txt for option * fixed bug * fixed bug * fixed bug * add benchmark.sh * mv thread_nums from 8 to 1 * deal with comments * deal with comments * fixed readme * deal with comments * add all platform shell * Update config.arm.txt * Update config.gpu.txt * Update config.x86.txt * fixed printinfo bug * rm proxy * add more model support * all backend config.txt * deal with comments * Add MattingDiff compare * fixed predict bug * adjust warmup/repeat times * add e2e/mem configs * fixed typo --------- Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com> --- .../cpp/config/config.arm.lite.fp16.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.arm.lite.fp16.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.arm.lite.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.arm.lite.fp32.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.ort.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.ort.fp32.e2e.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.ort.fp32.txt | 2 +- .../cpp/config/config.gpu.paddle.fp32.e2e.mem.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.paddle.fp32.e2e.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.paddle.fp32.txt | 2 +- .../config/config.gpu.paddle_trt.fp16.e2e.mem.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.paddle_trt.fp16.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.paddle_trt.fp16.txt | 2 +- .../config/config.gpu.paddle_trt.fp32.e2e.mem.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.paddle_trt.fp32.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.gpu.paddle_trt.fp32.txt | 2 +- .../cpp/config/config.gpu.trt.fp16.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.trt.fp16.e2e.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.trt.fp16.txt | 2 +- .../cpp/config/config.gpu.trt.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.trt.fp32.e2e.txt | 14 ++++++++++++++ benchmark/cpp/config/config.gpu.trt.fp32.txt | 2 +- .../cpp/config/config.x86.ort.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.x86.ort.fp32.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.x86.ov.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.x86.ov.fp32.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.x86.paddle.fp32.e2e.mem.txt | 14 ++++++++++++++ .../cpp/config/config.x86.paddle.fp32.e2e.txt | 14 ++++++++++++++ .../cpp/config/config.xpu.lite.fp32.e2e.mem.txt | 14 ++++++++++++++ benchmark/cpp/config/config.xpu.lite.fp32.e2e.txt | 14 ++++++++++++++ benchmark/cpp/macros.h | 2 +- cmake/paddle_inference.cmake | 1 - 32 files changed, 343 insertions(+), 8 deletions(-) create mode 100755 benchmark/cpp/config/config.arm.lite.fp16.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.arm.lite.fp16.e2e.txt create mode 100755 benchmark/cpp/config/config.arm.lite.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.arm.lite.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.ort.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.ort.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.trt.fp16.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.trt.fp16.e2e.txt create mode 100755 benchmark/cpp/config/config.gpu.trt.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.gpu.trt.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.x86.ort.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.x86.ort.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.x86.ov.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.x86.ov.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.x86.paddle.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.x86.paddle.fp32.e2e.txt create mode 100755 benchmark/cpp/config/config.xpu.lite.fp32.e2e.mem.txt create mode 100755 benchmark/cpp/config/config.xpu.lite.fp32.e2e.txt diff --git a/benchmark/cpp/config/config.arm.lite.fp16.e2e.mem.txt b/benchmark/cpp/config/config.arm.lite.fp16.e2e.mem.txt new file mode 100755 index 0000000000..8fb3794882 --- /dev/null +++ b/benchmark/cpp/config/config.arm.lite.fp16.e2e.mem.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: true +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_arm_lite_fp16_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.arm.lite.fp16.e2e.txt b/benchmark/cpp/config/config.arm.lite.fp16.e2e.txt new file mode 100755 index 0000000000..87b1a7f798 --- /dev/null +++ b/benchmark/cpp/config/config.arm.lite.fp16.e2e.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_arm_lite_fp16_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.arm.lite.fp32.e2e.mem.txt b/benchmark/cpp/config/config.arm.lite.fp32.e2e.mem.txt new file mode 100755 index 0000000000..9fb93fb27f --- /dev/null +++ b/benchmark/cpp/config/config.arm.lite.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: true +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_arm_lite_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.arm.lite.fp32.e2e.txt b/benchmark/cpp/config/config.arm.lite.fp32.e2e.txt new file mode 100755 index 0000000000..2fad23322b --- /dev/null +++ b/benchmark/cpp/config/config.arm.lite.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_arm_lite_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.ort.fp32.e2e.mem.txt b/benchmark/cpp/config/config.gpu.ort.fp32.e2e.mem.txt new file mode 100755 index 0000000000..b95718160a --- /dev/null +++ b/benchmark/cpp/config/config.gpu.ort.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: ort +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_ort_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.ort.fp32.e2e.txt b/benchmark/cpp/config/config.gpu.ort.fp32.e2e.txt new file mode 100755 index 0000000000..a1bf97be97 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.ort.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: ort +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_ort_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.ort.fp32.txt b/benchmark/cpp/config/config.gpu.ort.fp32.txt index 3f48bde01a..cb1d2b5b44 100755 --- a/benchmark/cpp/config/config.gpu.ort.fp32.txt +++ b/benchmark/cpp/config/config.gpu.ort.fp32.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.mem.txt b/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.mem.txt new file mode 100755 index 0000000000..d06642b1e4 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: paddle +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.txt b/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.txt new file mode 100755 index 0000000000..1e3dabc59c --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: paddle +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle.fp32.txt b/benchmark/cpp/config/config.gpu.paddle.fp32.txt index ba06c5c5d1..cbe7bc2cb4 100755 --- a/benchmark/cpp/config/config.gpu.paddle.fp32.txt +++ b/benchmark/cpp/config/config.gpu.paddle.fp32.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.mem.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.mem.txt new file mode 100755 index 0000000000..8063c45f40 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: paddle_trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_trt_fp16_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.txt new file mode 100755 index 0000000000..525d8c3e99 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: paddle_trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_trt_fp16_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp16.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.txt index f731ce7fc5..6cdd4f9254 100755 --- a/benchmark/cpp/config/config.gpu.paddle_trt.fp16.txt +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp16.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.mem.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.mem.txt new file mode 100755 index 0000000000..4f4fb67acb --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: paddle_trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_trt_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.txt new file mode 100755 index 0000000000..58fe7d8a33 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: paddle_trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_paddle_trt_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.paddle_trt.fp32.txt b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.txt index a11394eb78..7a83160cd1 100755 --- a/benchmark/cpp/config/config.gpu.paddle_trt.fp32.txt +++ b/benchmark/cpp/config/config.gpu.paddle_trt.fp32.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.gpu.trt.fp16.e2e.mem.txt b/benchmark/cpp/config/config.gpu.trt.fp16.e2e.mem.txt new file mode 100755 index 0000000000..9819ba74e0 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.trt.fp16.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_trt_fp16_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.trt.fp16.e2e.txt b/benchmark/cpp/config/config.gpu.trt.fp16.e2e.txt new file mode 100755 index 0000000000..742129f652 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.trt.fp16.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: true +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_trt_fp16_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.trt.fp16.txt b/benchmark/cpp/config/config.gpu.trt.fp16.txt index e27837e227..4a842a69c7 100755 --- a/benchmark/cpp/config/config.gpu.trt.fp16.txt +++ b/benchmark/cpp/config/config.gpu.trt.fp16.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.gpu.trt.fp32.e2e.mem.txt b/benchmark/cpp/config/config.gpu.trt.fp32.e2e.mem.txt new file mode 100755 index 0000000000..993a4309e2 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.trt.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_trt_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.trt.fp32.e2e.txt b/benchmark/cpp/config/config.gpu.trt.fp32.e2e.txt new file mode 100755 index 0000000000..1130f34821 --- /dev/null +++ b/benchmark/cpp/config/config.gpu.trt.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: gpu +device_id: 3 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: trt +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_gpu_trt_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.gpu.trt.fp32.txt b/benchmark/cpp/config/config.gpu.trt.fp32.txt index 49c2695c15..c38dfcfd55 100755 --- a/benchmark/cpp/config/config.gpu.trt.fp32.txt +++ b/benchmark/cpp/config/config.gpu.trt.fp32.txt @@ -1,5 +1,5 @@ device: gpu -device_id: 0 +device_id: 3 cpu_thread_nums: 1 warmup: 20 repeat: 100 diff --git a/benchmark/cpp/config/config.x86.ort.fp32.e2e.mem.txt b/benchmark/cpp/config/config.x86.ort.fp32.e2e.mem.txt new file mode 100755 index 0000000000..4970345640 --- /dev/null +++ b/benchmark/cpp/config/config.x86.ort.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: ort +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_ort_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.x86.ort.fp32.e2e.txt b/benchmark/cpp/config/config.x86.ort.fp32.e2e.txt new file mode 100755 index 0000000000..17803cbf71 --- /dev/null +++ b/benchmark/cpp/config/config.x86.ort.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: ort +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_ort_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.x86.ov.fp32.e2e.mem.txt b/benchmark/cpp/config/config.x86.ov.fp32.e2e.mem.txt new file mode 100755 index 0000000000..0717bcb203 --- /dev/null +++ b/benchmark/cpp/config/config.x86.ov.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: ov +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_ov_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.x86.ov.fp32.e2e.txt b/benchmark/cpp/config/config.x86.ov.fp32.e2e.txt new file mode 100755 index 0000000000..cdcb8d0ad0 --- /dev/null +++ b/benchmark/cpp/config/config.x86.ov.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: ov +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_ov_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.x86.paddle.fp32.e2e.mem.txt b/benchmark/cpp/config/config.x86.paddle.fp32.e2e.mem.txt new file mode 100755 index 0000000000..d36413d9a1 --- /dev/null +++ b/benchmark/cpp/config/config.x86.paddle.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: paddle +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_paddle_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.x86.paddle.fp32.e2e.txt b/benchmark/cpp/config/config.x86.paddle.fp32.e2e.txt new file mode 100755 index 0000000000..b255d7378c --- /dev/null +++ b/benchmark/cpp/config/config.x86.paddle.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: cpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: paddle +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_x86_paddle_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.xpu.lite.fp32.e2e.mem.txt b/benchmark/cpp/config/config.xpu.lite.fp32.e2e.mem.txt new file mode 100755 index 0000000000..6a68958b24 --- /dev/null +++ b/benchmark/cpp/config/config.xpu.lite.fp32.e2e.mem.txt @@ -0,0 +1,14 @@ +device: xpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 10 +repeat: 20 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_xpu_lite_fp32_e2e_mem.txt \ No newline at end of file diff --git a/benchmark/cpp/config/config.xpu.lite.fp32.e2e.txt b/benchmark/cpp/config/config.xpu.lite.fp32.e2e.txt new file mode 100755 index 0000000000..6ae4cf2f4a --- /dev/null +++ b/benchmark/cpp/config/config.xpu.lite.fp32.e2e.txt @@ -0,0 +1,14 @@ +device: xpu +device_id: 0 +cpu_thread_nums: 1 +warmup: 20 +repeat: 100 +backend: lite +profile_mode: end2end +include_h2d_d2h: false +use_fp16: false +collect_memory_info: false +sampling_interval: 1 +precision_compare: false +xpu_l3_cache: 0 +result_path: benchmark_xpu_lite_fp32_e2e.txt \ No newline at end of file diff --git a/benchmark/cpp/macros.h b/benchmark/cpp/macros.h index 021e7db2e5..018a7c8810 100755 --- a/benchmark/cpp/macros.h +++ b/benchmark/cpp/macros.h @@ -79,7 +79,7 @@ std::cout << "gpu_rss_mb: " << __gpu_mem__ << "MB." << std::endl; \ __ss__ << "gpu_rss_mb: " << __gpu_mem__ << "MB." << std::endl; \ std::cout << "gpu_util: " << __gpu_util__ << std::endl; \ - __ss__ << "gpu_rss_mb: " << __gpu_mem__ << "MB." << std::endl; \ + __ss__ << "gpu_util: " << __gpu_util__ << "MB." << std::endl; \ __resource_moniter__.Stop(); \ } \ fastdeploy::benchmark::ResultManager::SaveBenchmarkResult(__ss__.str(), \ diff --git a/cmake/paddle_inference.cmake b/cmake/paddle_inference.cmake index 5c125fcc53..fc9cbd8e12 100755 --- a/cmake/paddle_inference.cmake +++ b/cmake/paddle_inference.cmake @@ -134,7 +134,6 @@ else() endif(PADDLEINFERENCE_DIRECTORY) if(UNIX AND (NOT APPLE) AND (NOT ANDROID)) - message("?????????????? ${PATCHELF_EXE}") add_custom_target(patchelf_paddle_inference ALL COMMAND bash -c "PATCHELF_EXE=${PATCHELF_EXE} python ${PROJECT_SOURCE_DIR}/scripts/patch_paddle_inference.py ${PADDLEINFERENCE_INSTALL_DIR}/paddle/lib/libpaddle_inference.so" DEPENDS ${LIBRARY_NAME}) endif()