Specifications Compared
| Spec | L40S | P100 |
|---|---|---|
| TDP | 350W | 250W |
| VRAM | 48 GB | 16 GB |
| CUDA Cores | 18,176 | 3,584 |
| FP8 (dense) | 733 TFLOPS | Not published |
| Memory Type | GDDR6 | HBM2 |
| Architecture | Ada Lovelace | Pascal |
| FP16 (dense) | 362.05 TFLOPS | 18.7 TFLOPS |
| Form Factors | PCIe | SXM2, PCIe |
| INT8 (dense) | 733 TOPS | Not published |
| Interconnect | PCIe 4.0 | NVLink, PCIe 3.0 |
| Tensor Cores | 568 | Not published |
| FP32 Performance | 91.6 TFLOPS | 9.3 TFLOPS |
| FP64 Performance | 1.4 TFLOPS | 4.7 TFLOPS |
| Memory Bandwidth | 864 GB/s | 732 GB/s |
| FP8 (with sparsity) | 1,466 TFLOPS | Not published |
| FP16 (with sparsity) | 733 TFLOPS | Not published |
| INT8 (with sparsity) | 1,466 TOPS | Not published |
Performance Analysis
Dense FP16 performance measures 362.05 TFLOPS on the L40S against 18.7 TFLOPS on the P100. This ratio indicates the L40S processes matrix operations over nineteen times faster in dense FP16 mode for training and inference tasks. FP32 figures follow a similar pattern at 91.6 TFLOPS dense on the L40S versus 9.3 TFLOPS on the P100. Memory bandwidth of 864 GB/s on the L40S versus 732 GB/s on the P100 supports larger batch sizes during model execution. The L40S also lists FP16 with sparsity at 733 TFLOPS while the P100 provides only dense FP16 data. Such differences translate to reduced iteration times on the L40S for precision formats that match exactly between the two GPUs.
Current On-Demand Offers
Cheapest secure on-demand offer per provider that is reported in stock right now, per GPU per hour; multi-GPU instances show the whole-instance price alongside. Deploy opens the DeployGPU console for providers on the platform, otherwise the provider's own site. Stock is rechecked every minute.
L40S
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| Massed Compute | us-central-3 | 2 | $0.97 | $1.94 | Deploy |
| RunPod | global | 1 | $1.09 | — | Deploy |
| QuantaCloud | us-midwest-1 | 1 | $1.09 | — | Deploy |
| Ori | lille-4 | 1 | $1.55 | — | Deploy |
| Lyceum | Europe | 1 | $1.99 | — | Deploy |
5 providers in stock, 16 offers (cheapest per provider shown). All L40S offers, price history and alerts
P100
1 provider in stock, 3 offers (cheapest per provider shown). All P100 offers, price history and alerts
Notify me when L40S drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $0.97/GPU-hr.
QuantaCloud
Comparing providers? We broker across all of them.
Stop tab-switching between pricing pages. Tell us what you need, 16+ GPUs reserved or cluster capacity, and we return one quote at partner rates within 24 hours.
When to Choose the L40S
The L40S suits LLM training scenarios that demand 48 GB VRAM and 362.05 TFLOPS dense FP16 capacity. Its 864 GB/s bandwidth further enables handling of larger datasets without frequent memory swaps compared to the P100 specifications.
When to Choose the P100
The P100 fits scientific computing jobs that operate within 16 GB HBM2 limits and require 250W TDP operation. Its NVLink interconnect option provides an alternative path for multi GPU setups that align with Pascal era software stacks.
Use Cases
The L40S supplies 362.05 TFLOPS dense FP16 and 48 GB VRAM that exceed P100 values of 18.7 TFLOPS dense FP16 and 16 GB VRAM.
Dense FP16 throughput of 362.05 TFLOPS on the L40S supports faster token processing than the 18.7 TFLOPS dense FP16 on the P100.
The L40S 48 GB GDDR6 capacity accommodates larger fine tuning batches compared to the P100 16 GB HBM2 limit.
FP32 performance at 91.6 TFLOPS on the L40S surpasses the 9.3 TFLOPS on the P100 for image generation pipelines.
Both GPUs deliver FP32 results though the L40S reaches 91.6 TFLOPS while the P100 reaches 9.3 TFLOPS.
Frequently Asked Questions
What is the FP16 dense performance difference between L40S and P100?▾
The L40S achieves 362.05 TFLOPS dense FP16 while the P100 achieves 18.7 TFLOPS dense FP16. This gap affects training duration for models that rely on dense FP16 operations.
Which GPU offers higher memory bandwidth?▾
The L40S provides 864 GB/s memory bandwidth. The P100 provides 732 GB/s memory bandwidth. Data movement rates favor the L40S during large tensor transfers.
What FP32 performance do the L40S and P100 deliver?▾
The L40S reaches 91.6 TFLOPS FP32. The P100 reaches 9.3 TFLOPS FP32. Scientific simulations that use FP32 execute more rapidly on the L40S.
How do TDP values differ between these GPUs?▾
The L40S lists a TDP of 350W. The P100 lists a TDP of 250W. Power budgets must account for the higher draw on the L40S in dense deployments.
Which is cheaper to rent, the L40S or the P100?▾
Cloud rental prices for both the L40S and P100 vary by provider, configuration, and availability. This page shows live pricing from 25+ providers updated every 60 seconds. Scroll to the Live Cloud Pricing section to compare current rates.
How much VRAM does the L40S have compared to the P100?▾
The L40S has 48 GB of GDDR6 memory. The P100 has 16 GB of HBM2 memory.
Can I find L40S and P100 GPUs available to rent right now?▾
Yes. This page shows real-time availability across 25+ cloud GPU providers. The Live Cloud Pricing section displays only in-stock offers with current pricing.
What is the main difference between the L40S and the P100?▾
The L40S uses the Ada Lovelace architecture (2023) while the P100 uses Pascal (2016). The L40S delivers 19.4x the dense FP16 throughput (362.05 vs 18.7 TFLOPS, both without sparsity) and 1.2x the memory bandwidth of the P100.
Rent these GPUs
Each GPU page lists every current on-demand offer by provider, per GPU-hour, updated every minute.
Related comparisons
How this page is made
- Specifications come from the NVIDIA, AMD and Intel datasheets for the L40S and the P100. Dense and sparse throughput are listed separately.
- Prices and availability are live: every offer shown is an in-stock, on-demand listing from a provider we track, rechecked every minute.
- The written comparison (overview, performance notes, when to choose each card, verdict, use cases and FAQ) was drafted with an AI model from the specification table above and passed an automated check that rejects any figure not in that table. It was last generated on . It contains no prices; those are always read live.
- Read how we collect the data, or report an error on this page.
Next steps
- Rent L40SEvery current offer by provider, daily price history and a price alert.
- Rent P100Every current offer by provider, daily price history and a price alert.
- GPU price indexHow on-demand prices for the major GPUs have moved, updated daily.
- All GPUsEvery GPU we track, with current lows and provider counts.