NVIDIA L20 48GB vs NVIDIA L40S 48GB

The L40S delivers 3.1x the FP8 throughput of the L20 (733 vs 239 TFLOPS dense).

L20: Ada Lovelace, 2024 L40S: Ada Lovelace, 2023
FP8 dense lead
3.1x
L40S: 733 vs 239 TFLOPS
VRAM
48 vs 48 GB
Same capacity
Bandwidth
864 GB/s vs 864 GB/s
Even
TDP
275 W vs 350 W
L20 draws 75 W less

Specifications

L20 L40S
Architecture
Ada Lovelace Ada Lovelace
Launch Year
2024 2023
Form Factor
PCIe PCIe
VRAM
48 GB 48 GB
Memory Bandwidth
864 GB/s 864 GB/s
TDP
275 W 350 W
Process Node
4nm

Performance (TFLOPS)

L20 L40S
FP64
1.4 TFlops
FP32
59.8 TFlops 91.6 TFlops BEST
FP16
119.5 TFlops
sparse not published
362.1 TFlops BEST
733.0 TFLOPS sparse
BF16
119.5 TFlops
sparse not published
362.1 TFlops BEST
733.0 TFLOPS sparse
FP8
239.0 TFlops
sparse not published
733.0 TFlops BEST
1466.0 TFLOPS sparse
INT8
239.0 TFlops
sparse not published
733.0 TFlops BEST
1466.0 TOPS sparse

FLOPS by Precision

What actually differs

The L20 is the newer part: Ada Lovelace, launched in 2024, against the L40S's Ada Lovelace from 2023. Newer architectures typically add lower-precision formats and better throughput per watt, so check the precision rows your workload actually uses.

Dense throughput for the L20 against the L40S: FP32 60 vs 92 TFLOPS, FP16 120 vs 362 TFLOPS, FP8 239 vs 733 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 48 GB against 48 GB, fed at 864 GB/s versus 864 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 275 W for the L20 and 350 W for the L40S. At FP32 that works out to 0.22 against 0.26 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 L20 faster than the L40S?

At FP8 precision the L40S reaches 733 TFLOPS dense against 239 TFLOPS for the L20. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the L20 or the L40S?

Both GPUs carry 48 GB of VRAM. Memory bandwidth is 864 GB/s for the L20 and 864 GB/s for the L40S.

How much power do the L20 and the L40S draw?

The L20 is rated at 275 W TDP and the L40S at 350 W. On FP32 throughput per watt, the L40S is the more efficient part.

Can I rent the L20 or the L40S in the cloud?

Yes. Live cloud listings tracked by Flopper start at $0.69 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

NVIDIA L40S 48GB

ProviderInstancePrice/GPU-hrChecked
DataCrunch
1L40S.20V (spot) $0.69 5h ago View →
Nebius
$0.74 5h ago View →
RunPod
L40S (Community Cloud) $0.79 5h ago View →
Cudo Compute
epyc-genoa-l40s-graphics $0.87 5h ago View →
CoreWeave
nvidia-l40s (spot) $0.98 5h ago View →
RunPod
L40S (Secure Cloud) $0.99 5h ago View →
DataCrunch
1L40S.20V (on-demand) $1.37 5h ago View →
Crusoe
$1.50 5h ago View →
Nebius
$1.55 5h ago View →
DigitalOcean
NVIDIA L40S $1.57 5h ago View →
Vultr
vcg-l40s-16c-180g-48vram $1.67 5h ago View →
OVHcloud
l40s-360 $1.80 5h ago View →
CoreWeave
nvidia-l40s (on-demand) $2.25 5h ago View →
All L40S listings and price history →

Get Comparison Updates

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