NVIDIA GeForce RTX 3090 24GB vs NVIDIA RTX A6000 48GB

The RTX A6000 delivers 2.2x the FP16 throughput of the GeForce RTX 3090 (155 vs 71 TFLOPS dense). The RTX A6000 also carries 24 GB more memory (48 GB vs 24 GB).

GeForce RTX 3090: Ampere, 2020 RTX A6000: Ampere, 2020
FP16 dense lead
2.2x
RTX A6000: 155 vs 71 TFLOPS
VRAM
24 vs 48 GB
24 GB more for the RTX A6000
Bandwidth
936 GB/s vs 768 GB/s
GeForce RTX 3090 moves data faster
TDP
350 W vs 300 W
RTX A6000 draws 50 W less

Specifications

GeForce RTX 3090
RTX A6000
Architecture
AmpereAmpere
Launch Year
20202020
Form Factor
PCIePCIe
VRAM
24 GB48 GB
Memory Bandwidth
936 GB/s768 GB/s
TDP
350 W300 W
Process Node
8nm8nm

Performance (TFLOPS)

GeForce RTX 3090
RTX A6000
FP64
0.6 TFlops 0.6 TFlops
FP32
35.6 TFlops 38.7 TFlops
FP16
71.0 TFlops
142.0 TFLOPS sparse
154.8 TFlops
309.6 TFLOPS sparse
BF16
71.0 TFlops
142.0 TFLOPS sparse
154.8 TFlops
309.6 TFLOPS sparse
INT8
284.0 TFlops
568.0 TOPS sparse
309.7 TFlops
619.4 TOPS sparse

FLOPS by Precision

What actually differs

Both GPUs launched in 2020: the GeForce RTX 3090 on Ampere and the RTX A6000 on Ampere.

Dense throughput for the GeForce RTX 3090 against the RTX A6000: FP64 0.6 vs 0.6 TFLOPS, FP32 36 vs 39 TFLOPS, FP16 71 vs 155 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 24 GB against 48 GB, fed at 936 GB/s versus 768 GB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

Power budgets are 350 W for the GeForce RTX 3090 and 300 W for the RTX A6000. At FP32 that works out to 0.10 against 0.13 TFLOPS per watt.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.

Frequently asked questions

Is the GeForce RTX 3090 faster than the RTX A6000?

At FP16 precision the RTX A6000 reaches 155 TFLOPS dense against 71 TFLOPS for the GeForce RTX 3090. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the GeForce RTX 3090 or the RTX A6000?

The RTX A6000 carries 48 GB of VRAM versus 24 GB for the GeForce RTX 3090. Memory bandwidth is 936 GB/s for the GeForce RTX 3090 and 768 GB/s for the RTX A6000.

How much power do the GeForce RTX 3090 and the RTX A6000 draw?

The GeForce RTX 3090 is rated at 350 W TDP and the RTX A6000 at 300 W. On FP32 throughput per watt, the RTX A6000 is the more efficient part.

Can I rent the GeForce RTX 3090 or the RTX A6000 in the cloud?

Yes. Live cloud listings tracked by Flopper start at $0.22 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

NVIDIA GeForce RTX 3090 24GB

ProviderConfigurationPrice/GPU-hrChecked
RunPod
1× GeForce RTX 3090 24GB Community $0.22 just now View →
RunPod
1× GeForce RTX 3090 24GB $0.50 just now View →
All GeForce RTX 3090 listings and price history →

NVIDIA RTX A6000 48GB

ProviderConfigurationPrice/GPU-hrChecked
DataCrunch
1× RTX A6000 48GB Spot $0.30 just now View →
RunPod
1× RTX A6000 48GB Community $0.33 just now View →
Hyperstack
1× RTX A6000 48GB Spot $0.40 just now View →
Hyperstack
1× RTX A6000 48GB $0.50 just now View →
RunPod
1× RTX A6000 48GB $0.53 just now View →
Massed Compute
8× RTX A6000 48GB $0.57 just now View →
DataCrunch
1× RTX A6000 48GB $0.61 just now View →
AceCloud
1× RTX A6000 48GB $0.93 just now View →
Paperspace
1× RTX A6000 48GB $1.93 just now View →
All RTX A6000 listings and price history →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.