…
L40S vs RTX 6000 Ada
Cloud rental price, compute per dollar and specs, side by side.
Open in the interactive comparer →
L40S$0.60lowest on-demand per GPU·h
$1.66 per PFLOP·h · 17 providersRTX 6000 Ada$0.60lowest on-demand per GPU·h
$1.65 per PFLOP·h · 6 providersWhat actually differs
- The L40S and RTX 6000 Ada have the same dense FP16/BF16 throughput (0.362 PF vs 0.364 PF).
- The RTX 6000 Ada has 1.1× more memory bandwidth than the L40S (864 vs 960 GB/s), which usually limits LLM token generation. Memory: 48 GB vs 48 GB.
- Both cost about the same per hour at the lowest on-demand prices ($0.60 vs $0.60).
- Per unit of compute, the RTX 6000 Ada is the better deal: $1.65 vs $1.66 per PFLOP-hour of dense FP16/BF16.
| Spec / price | L40S | RTX 6000 Ada |
|---|---|---|
| Memory | 48 GB GDDR6 | 48 GB GDDR6 |
| Memory bandwidth | 864 GB/s | 960 GB/sbetter |
| FP16 / BF16 (dense) | 0.362 PF | 0.364 PFbetter |
| FP8 (dense) | 0.733 PFbetter | 0.729 PF |
| FP4 (dense) | — | — |
| INT8 (dense) | 0.733 PFbetter | 0.729 PF |
| TF32 (dense) | 0.183 PFbetter | 0.182 PF |
| FP32 (dense) | 0.0916 PFbetter | 0.0911 PF |
| FP64 (dense) | 0.0014 PF | 0.0014 PF |
| Power | 350 W | 300 W |
| Architecture | Ada Lovelace (2023) | Ada Lovelace (2022) |
| Lowest on-demand | $0.60 | $0.60better |
| Median on-demand | $1.50 | $0.86better |
| Lowest spot | $0.38better | $0.58 |
| FP16/BF16 $ per PFLOP·h | $1.66 | $1.65better |
| Providers | 17 | 6 |
L40S vs RTX 6000 Ada: questions
Is the L40S faster than the RTX 6000 Ada?
The L40S and RTX 6000 Ada have the same dense FP16/BF16 compute. The RTX 6000 Ada has 1.1× more memory bandwidth than the L40S. Real-world speedups depend on the workload: training and prefill scale with compute, while LLM decoding scales mostly with memory bandwidth.
Which is cheaper per FLOP, the L40S or the RTX 6000 Ada?
Right now the RTX 6000 Ada is cheaper per unit of compute: $1.65 vs $1.66 per PFLOP-hour at the lowest on-demand prices.
How much does it cost to rent an L40S vs an RTX 6000 Ada for a month?
Running one GPU around the clock at the lowest on-demand price costs about $440 per month for the L40S and $438 for the RTX 6000 Ada.
Which has more memory, the L40S or the RTX 6000 Ada?
Both have 48 GB of VRAM; the RTX 6000 Ada has more bandwidth.