GTX 1080 vs Quadro RTX 5000

The Quadro RTX 5000 lists dense FP16 performance at 89.2 TFLOPS while the GTX 1080 has no published FP16 figure. The ratio of dense FP16 to FP32 on the Quadro RTX 5000 equals 89.2 divided by 11.2. This gap indicates that dense FP16 operations complete faster than FP32 operations on the Quadro RTX 5000 for training and inference tasks that support the format. Memory bandwidth of 448 GB/s on the Quadro RTX 5000 exceeds the 320 GB/s of the GTX 1080 by a factor of 1.4. The higher bandwidth supports larger batch sizes when data movement limits throughput. The GTX 1080 remains limited to its published 8.9 TFLOPS FP32 figure without a comparable dense FP16 rating.

The GTX 1080 suits workloads that operate entirely within its 8.9 TFLOPS FP32 rating and 320 GB/s bandwidth. Its 180W TDP fits systems with tighter power constraints than the 230W requirement of the Quadro RTX 5000. Users select the GTX 1080 when FP32 tasks match the 8 to 11 GB VRAM capacity and no dense FP16 acceleration is required.

The Quadro RTX 5000 fits workloads that utilize its dense FP16 rating of 89.2 TFLOPS and 448 GB/s bandwidth. Its 16 GB VRAM accommodates larger data sets than the 8 to 11 GB range of the GTX 1080. The device is chosen when dense FP16 or dense INT8 operations at 178.4 TOPS align with the target application.