GPU Insights

Comprehensive GPU Performance Benchmarks

About This Benchmark

This performance test evaluates GPU hardware performance in a training scenario using ResNet50. The output score reflects the time to complete a given training task - higher scores indicate better computational performance.

Test Configuration

  • Model: ResNet50 (23.5M parameters)
  • Default Command: python main.py -m -s 512 -e 2 -mt resnet50 -bs 256 -dt FP32/FP16 -gpu 0
  • Score Impact: Affected by video memory bandwidth and PCIe bus bandwidth
0
Total Devices Tested
0
Top FP32 Score
0
Top FP16 Score
0
GPU Vendors
Device Platform FP32 Score FP32 Batch Size FP16 Score FP16 Batch Size Notes Test Date