Consumercarts is reader-supported. Learn more
NVIDIA L40 vs NVIDIA GeForce RTX 5090 D V2: Comparison
NVIDIA L40 and NVIDIA GeForce RTX 5090 D V2 separate most on tdp: NVIDIA L40 at 300 W, NVIDIA GeForce RTX 5090 D V2 at 575 W, 92% in favour of NVIDIA L40. Slot Width splits them by another 75%, again favouring NVIDIA L40. They match on 4 other specifications, including gpu brand, tier and ray tracing support.
Specs
| Specification |  |  |
|---|---|---|
| GPU Brand | NVIDIA | NVIDIA |
| Series | RTX Ada Generation | GeForce RTX 50 |
| Board Partner | NVIDIA Reference (data center/OEM only, passive cooling) | NVIDIA Reference (China-market AIB partners: ASUS, MSI, Gigabyte, Colorful, Galax) |
| VRAM | 48 GB | 32 GB-33% |
| Tier | Flagship | Flagship |
| Memory Type | GDDR6 | GDDR7 |
| Boost Clock | 2490 MHz | 2010 MHz-19% |
| Ray Tracing Support | true | true |
| TDP | 300 W | 575 W-92% |
| Architecture | Ada Lovelace | Blackwell |
| Process Node | TSMC 4N | TSMC 4N |
| Card Length | 267 mm | 304 mm-14% |
| Slot Width | 2 | 3.5-75% |
| Memory Bus Width | 384 bit-25% | 512 bit |
| Memory Bandwidth | 864 GB/s-45% | 1568 GB/s |
| Recommended PSU | 550 W | 825 W-50% |
| Power Connector | 1x 16-pin (12VHPWR) | 1x 12V-2x6 |
| MSRP | 8000 | 2299 |
| Launch Date | 2022-10 | 2025-08 |
| Base / Game Clock | 735 MHz | 1590 MHz |
| PCIe Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |
| Display Outputs | 4x DisplayPort 1.4a (rarely used in server deployment) | 3x DisplayPort 2.1b, 1x HDMI 2.1b |
| Media Engine | 3x NVENC (8th gen), 3x NVDEC, AV1 encode/decode | 2x NVENC (9th gen), 2x NVDEC, AV1 + H.264/HEVC encode |
Verdict
NVIDIA L40
8.6 / 10 Expert Rating
A versatile data center GPU that genuinely splits the difference between graphics and AI workloads - since replaced by the L40S for AI-heavy deployments but still a strong all-rounder.
NVIDIA GeForce RTX 5090 D V2
8.8 / 10 Expert Rating
A minor running change rather than a new generation - functionally identical to the original 5090 D with a slightly more conservative power target.