RTX 4080 Super

Ada Lovelace2024320W TDPBF16FP8
NVIDIA Product Page

Specifications

CUDA Cores10,240
Tensor Cores320
AI TOPS836
FP32 TFLOPS52.0
Memory16 GB GDDR6X
Memory Bandwidth736 GB/s
Memory Interface256-bit
TDP320W

Pricing

Proshop (cheapest)-
Vast.ai$0.30/hr
Google Cloud-
Modal-

Efficiency

Power Efficiency
CUDA/Watt
32.00
Tensor/Watt
1.00
Bandwidth/Watt
2.30GB/s/W
Rental Value (CUDA/$/hr)
Vast.ai
34,133.33