GPU Comparison
NVIDIA RTX 4090 vs NVIDIA V100.
Side-by-side comparison of specs, performance, and cloud rental pricing. NVIDIA RTX 4090 is the cheaper option at $0.36/hr.
TL;DR
Choose NVIDIA V100 if you need 32GB+ on a single GPU (8GB more than NVIDIA RTX 4090).
Choose NVIDIA RTX 4090 if you need raw FP16 throughput (425% faster than NVIDIA V100).
Switch comparison
vs
Ada Lovelace
NVIDIA RTX 4090
$0.36/hrCheaper
$0.16-$0.37/hr range · 9 regions
VRAM24GB GDDR6X
FP16165 TFLOPS
Bandwidth1.0 TB/s
Volta
NVIDIA V100
$0.55/hr
$0.45-$0.65/hr range · 4 regions
VRAM32GB HBM2
FP1631.4 TFLOPS
Bandwidth900 GB/s
At a glance
Spec by spec.
Visual comparison across the four specs that matter most for AI workloads. Green bar marks the winner.
VRAM
NVIDIA RTX 4090
24 GB
NVIDIA V100
32 GB
Memory Bandwidth
NVIDIA RTX 4090
1.0 TB/s
NVIDIA V100
900 GB/s
FP16 Performance
NVIDIA RTX 4090
165 TFLOPS
NVIDIA V100
31.4 TFLOPS
Hourly Price
NVIDIA RTX 4090
$0.36/hr
NVIDIA V100
$0.55/hr
Full specs
Complete spec sheet.
SpecNVIDIA RTX 4090NVIDIA V100
ArchitectureAda LovelaceVolta
VRAM24 GB32 GB●
Memory TypeGDDR6XHBM2
Memory Bandwidth1.0 TB/s●900 GB/s
FP32 Performance82.6 TFLOPS●15.7 TFLOPS
FP16 Performance165 TFLOPS●31.4 TFLOPS
INT8 Performance661 TOPS●125 TOPS
TDP450 W300 W
Form FactorConsumerPCIe
NVLinkNoYes
Max GPU/Node28
Hourly Price$0.36/hr●$0.55/hr
FAQ
Frequently asked.
Pick a GPU and deploy.
Both GPUs available on Runcrate · per-second billing · no commitments
Other comparisons