Specifications Compared
| Spec | L40 | MI300X |
|---|---|---|
| TDP | 300W | 750W |
| VRAM | 48 GB | 192 GB |
| CUDA Cores | 18,176 | Not published |
| FP8 (dense) | 362 TFLOPS | 2,614.9 TFLOPS |
| Memory Type | GDDR6 | HBM3 |
| Architecture | Ada Lovelace | CDNA 3 |
| FP16 (dense) | 181.05 TFLOPS | 1,307.4 TFLOPS |
| Form Factors | PCIe | OAM |
| INT8 (dense) | 362 TOPS | 2,614.9 TOPS |
| Interconnect | PCIe 4.0 | Infinity Fabric, PCIe 5.0 |
| Tensor Cores | 568 | Not published |
| FP32 Performance | 90.5 TFLOPS | 163.4 TFLOPS |
| FP64 Performance | Not published | 81.7 TFLOPS |
| Memory Bandwidth | 864 GB/s | 5,300 GB/s |
| FP8 (with sparsity) | 724 TFLOPS | 5,229.8 TFLOPS |
| FP16 (with sparsity) | 362.1 TFLOPS | 2,614.9 TFLOPS |
| INT8 (with sparsity) | 724 TOPS | 5,229.8 TOPS |
Performance Analysis
The MI300X FP16 dense performance of 1307.4 TFLOPS exceeds the L40 FP16 dense performance of 181.05 TFLOPS by a factor derived solely from those two numbers. This gap means training workloads that rely on dense FP16 operations complete more iterations per unit time on the MI300X. Inference tasks using the same dense FP16 metric benefit similarly from the higher throughput. The MI300X memory bandwidth of 5300 GB/s versus the L40 bandwidth of 864 GB/s permits larger batch sizes before memory capacity of 192 GB on the MI300X or 48 GB on the L40 becomes the limit. The MI300X FP16 with sparsity rating of 2614.9 TFLOPS compared with the L40 FP16 with sparsity rating of 362.1 TFLOPS follows the same proportional advantage when sparsity is applied consistently.
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.
L40
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| ThunderCompute | USA | 2 | $0.79 | $1.58 | Deploy |
| Massed Compute | us-central-2 | 1 | $0.86 | — | Deploy |
| QuantaCloud | us-midwest-2 | 4 | $0.94 | $3.75 | Deploy |
| Hyperstack | CANADA-1 | 4 | $1.00 | $4.00 | Deploy |
4 providers in stock, 17 offers (cheapest per provider shown). All L40 offers, price history and alerts
MI300X
1 provider in stock, 1 offer. All MI300X offers, price history and alerts
Notify me when L40 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $0.79/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 L40
The L40 suits environments where power draw must remain at 300W and PCIe interconnect suffices. Its 48 GB GDDR6 configuration supports inference workloads that fit within that memory limit while maintaining FP32 throughput of 90.5 TFLOPS. Organizations running multiple cards under constrained cooling or power budgets select the L40 over higher TDP alternatives.
When to Choose the MI300X
The MI300X fits workloads that require 192 GB HBM3 capacity and 5300 GB/s bandwidth for large model training. Its FP16 dense rating of 1307.4 TFLOPS accelerates dense matrix operations beyond the L40 capability. Systems equipped with Infinity Fabric interconnect benefit from the MI300X OAM form factor when scaling across multiple GPUs.
Use Cases
The MI300X supplies 192 GB memory and 1307.4 TFLOPS dense FP16 compared with the L40 48 GB and 181.05 TFLOPS dense FP16.
The MI300X 5300 GB/s bandwidth and 2614.9 TFLOPS dense FP8 exceed the L40 864 GB/s and 362 TFLOPS dense FP8 for larger batch processing.
The MI300X FP32 rating of 163.4 TFLOPS and 192 GB capacity support fine-tuning steps that exceed the L40 90.5 TFLOPS and 48 GB limits.
The L40 300W TDP and 48 GB GDDR6 fit diffusion workloads that do not require the MI300X 750W or 192 GB specifications.
Both GPUs deliver FP32 performance with the L40 at 90.5 TFLOPS and the MI300X at 163.4 TFLOPS depending on power and memory needs.
Frequently Asked Questions
What memory types do the L40 and MI300X use?▾
The L40 uses 48 GB GDDR6 while the MI300X uses 192 GB HBM3. These capacities determine maximum model sizes that fit without partitioning.
How do the FP16 dense ratings compare?▾
The MI300X FP16 dense rating reaches 1307.4 TFLOPS and the L40 FP16 dense rating reaches 181.05 TFLOPS. The ratio follows directly from these two published figures.
What interconnect options exist for each card?▾
The L40 supports PCIe 4.0 while the MI300X supports Infinity Fabric and PCIe 5.0. Multi GPU scaling therefore differs by fabric choice.
Which GPU has higher memory bandwidth?▾
The MI300X provides 5300 GB/s bandwidth compared with the L40 864 GB/s bandwidth. This difference affects achievable batch sizes in memory bound operations.
What are the TDP values for the L40 and MI300X?▾
The L40 TDP is 300W and the MI300X TDP is 750W. Power delivery and cooling requirements scale with these wattage numbers.
Do both GPUs list FP8 performance figures?▾
The L40 lists FP8 dense at 362 TFLOPS and the MI300X lists FP8 dense at 2614.9 TFLOPS. Direct comparison uses only the dense values from each specification block.
Which is cheaper to rent, the L40 or the MI300X?▾
Cloud rental prices for both the L40 and MI300X 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 L40 have compared to the MI300X?▾
The L40 has 48 GB of GDDR6 memory. The MI300X has 192 GB of HBM3 memory.
Can I find L40 and MI300X 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 L40 and the MI300X?▾
The L40 uses the Ada Lovelace architecture (2023) while the MI300X uses CDNA 3 (2023). The MI300X delivers 7.2x the dense FP16 throughput (1,307.4 vs 181.05 TFLOPS, both without sparsity) and 6.1x the memory bandwidth of the L40.
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 L40 and the MI300X. 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 L40Every current offer by provider, daily price history and a price alert.
- Rent MI300XEvery 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.