This is a benchmark of the Caffe deep learning framework and currently supports the AlexNet and Googlenet model and execution on both CPUs and NVIDIA GPUs.
To run this test with the Phoronix Test Suite, the basic command is: phoronix-test-suite benchmark caffe.
C/C++ Compiler Toolchain + CMake + Python + BLAS (Basic Linear Algebra Sub-Routine) + C++ Boost + Linear Algebra Pack + Snappy Compression + GFlags + OpenCV + HDF5
Accolades
150k+ Downloads
Supported Platforms
* Uploading of benchmark result data to OpenBenchmarking.org is always optional (opt-in) via the Phoronix Test Suite for users wishing to share their results publicly. ** Data based on those opting to upload their test results to OpenBenchmarking.org and users enabling the opt-in anonymous statistics reporting while running benchmarks from an Internet-connected platform. *** Test profile page view reporting began March 2021. Data updated weekly as of 18 November 2024.
Revision History
pts/caffe-1.5.0 [View Source] Sat, 26 Sep 2020 21:35:45 GMT Overhaul Caffe test profile with latest Git snapshot, switch to CMake build system, clean up test options, etc.
pts/caffe-1.4.0 [View Source] Sat, 29 Dec 2018 11:15:41 GMT Update Caffe to latest Git snapshot to hopefully workaround build problems on newer distros.
pts/caffe-1.3.2 [View Source] Wed, 04 Jan 2017 11:07:36 GMT Fix for OpenCV 3.2.
pts/caffe-1.3.1 [View Source] Wed, 28 Dec 2016 20:36:42 GMT Don't show title string of "Caffe AlexNet" but "Caffe" with recent test profile versions supporting more than just AlexNet.
pts/caffe-1.3.0 [View Source] Wed, 28 Dec 2016 20:34:27 GMT Update to latest Git snapshot to fix OpenCV compatibility.
pts/caffe-1.2.0 [View Source] Mon, 15 Aug 2016 16:11:16 GMT Add Googlenet support, decrease CPU only iteration count.
pts/caffe-1.1.1 [View Source] Sun, 12 Jun 2016 18:32:44 GMT Add OpenCV and OpenBLAS support.
pts/caffe-1.1.0 [View Source] Sat, 11 Jun 2016 19:32:55 GMT Update
pts/caffe-1.0.0 [View Source] Sat, 14 Nov 2015 15:29:45 GMT Initial commit of Caffe deep learning framework and with this benchmark using the AlexNet model for benchmarking.
Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 100
OpenBenchmarking.org metrics for this test profile configuration based on 62 public results since 26 September 2020 with the latest data as of 23 February 2024.
Additional benchmark metrics will come after OpenBenchmarking.org has collected a sufficient data-set.
Based on OpenBenchmarking.org data, the selected test / test configuration (Caffe 2020-02-13 - Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 100) has an average run-time of 2 minutes. By default this test profile is set to run at least 3 times but may increase if the standard deviation exceeds pre-defined defaults or other calculations deem additional runs necessary for greater statistical accuracy of the result.
Based on public OpenBenchmarking.org results, the selected test / test configuration has an average standard deviation of 0.2%.
Notable Instruction Set Usage
Notable instruction set extensions supported by this test, based on an automatic analysis by the Phoronix Test Suite / OpenBenchmarking.org analytics engine.
Requires passing a supported compiler/build flag (verified with targets: sandybridge, skylake, tigerlake, cascadelake, sapphirerapids, alderlake, znver2, znver3). Found on Intel processors since Sandy Bridge (2011). Found on AMD processors since Bulldozer (2011).
Requires passing a supported compiler/build flag (verified with targets: skylake, tigerlake, cascadelake, sapphirerapids, alderlake, znver2, znver3). Found on Intel processors since Haswell (2013). Found on AMD processors since Excavator (2016).
Requires passing a supported compiler/build flag (verified with targets: skylake, tigerlake, cascadelake, sapphirerapids, alderlake, znver2, znver3). Found on Intel processors since Haswell (2013). Found on AMD processors since Bulldozer (2011).
This benchmark has been successfully tested on the below mentioned architectures. The CPU architectures listed is where successful OpenBenchmarking.org result uploads occurred, namely for helping to determine if a given test is compatible with various alternative CPU architectures.