RTX 4080 vs V100

The FP16 dense figure reaches 97.5 TFLOPS on the RTX 4080 against 125 TFLOPS on the V100 while the FP32 figure reaches 48.7 TFLOPS on the RTX 4080 against 15.7 TFLOPS on the V100. This FP16 to FP32 ratio equals roughly 2 times on the RTX 4080 and 8 times on the V100 indicating the V100 delivers greater relative acceleration for FP16 dense workloads during training and inference. The RTX 4080 also lists FP16 with sparsity at 195 TFLOPS yet the V100 lists no sparsity figure. Memory bandwidth of 900 GB/s on the V100 versus 717 GB/s on the RTX 4080 allows larger batch sizes on the V100 in memory bound scenarios. Direct dense FP16 comparisons therefore favor the V100 while sparsity options remain available only on the RTX 4080.

The RTX 4080 suits workloads that benefit from its 48.7 TFLOPS FP32 performance which exceeds the V100 by more than 3 times. Its 16 GB GDDR6X configuration and PCIe 4.0 interconnect support inference tasks where sparsity can double effective FP16 throughput to 195 TFLOPS.

The V100 suits workloads that benefit from its 125 TFLOPS FP16 dense performance which exceeds the RTX 4080 by 28 percent. Its 900 GB/s bandwidth and NVLink interconnect support training tasks that require higher memory throughput and multi GPU scaling.