1 Reply Latest reply on Apr 23, 2015 5:22 PM by Horshack

    LR6 vs LR5 performance measurements

    Horshack Level 4

      Here are some quick measurements I've done so far. System is running Windows 7 64-bit, i7-3770K quad-core processor overclocked to 4.5ghz, 32GB DDR, LR catalog on SSD, raw files on 7200RPM HD. Video card is NVidia GeForce GTX 650 Ti - GPU acceleration confirmed functional under LR6.

       

      Importing 74 D800 raw files with 1:1 previews

      LR5.7: 4 minutes 43 seconds

        LR6: 4 minutes 58 seconds [same with GPU acceleration on/off]

       

      Exporting 74 D800 raw files to JPEG, full-size, 80% quality, all default develop options

      LR5.7: 4 minutes 3 seconds

        LR6: 2 minutes 53 seconds


      The Export module is making better use of multiprocessing. Here is a comparison of CPU utilization between LR5.7 and LR6 for the D800 export:

      i-Tvk2sWQ.png

      i-vK9DTSF.png

        • 1. Re: LR6 vs LR5 performance measurements
          Horshack Level 4

          Regarding utilization of GPU acceleration, I analyzed LR6's operation under Microsoft's Windows Performance Analyzer and found the following:

           

          Import+Preview Generation: Doesn't use GPU

          Export: Doesn't use GPU

          Develop Module 100% initial zoom/rendering: Modest amount of GPU utilization

          Develop Module Adjustment sliders: Significant GPU utilization. All sliders found to use GPU

          Develop Module Adjustment brush: Significant GPU utilization

           

          Here is a sample of the GPU utilization for an adjustment brush operation, with only the exposure setting used:

          i-NjPVQh9.jpg

          i-3bxShJv.png