How well does NVIDIA GeForce RTX 3090 Ti scale across multiple GPUs?
Sagot
NVIDIA GeForce RTX 3090 Ti performance headline: 40 FP16 TFLOPS, 20 FP32 TFLOPS, 1,008 GB/s bandwidth, 24 GB VRAM.
Converted into practical benchmarks: model training a 7B-parameter LLM in FP16 with reasonable batch sizes typically saturates compute before bandwidth; real-time serving on the same model is usually bandwidth-bound and tracks the 1,008 GB/s figure. Diffusion image generation benchmarks sit between the two — compute-heavy steps utilise tensor cores well, while attention blocks still touch bandwidth.
Review full specs and related comparisons on the NVIDIA GeForce RTX 3090 Ti page.