 Unterschiede hervorheben |  Nvidia L4 mit 24 GB GDDR6-RAM |
Nvidia A10 mit 24 GB GDDR6-RAM |
Nvidia A16 mit 4x 16 GB GDDR6 ECC-RAM |
Nvidia A30 mit 24 GB HBM2 ECC-RAM |
Nvidia A40 mit 48 GB GDDR6 ECC-RAM |
Nvidia L40 mit 48 GB GDDR6 ECC-RAM |
Nvidia L40S mit 48 GB GDDR6 ECC-RAM |
| Chipsatz | L4 | A10 | A16 | A30 | A40 | L40 | L40S |
| GPU Architektur | Ada | Ampere | Ampere | Ampere | Ampere | Ada | Ada |
| Speicher | 24 GB | 24 GB | 4x 16 GB | 24 GB | 48 GB | 48 GB | 48 GB |
| Speicherart | GDDR6 | GDDR6 | GDDR6 ECC | HBM2 ECC | GDDR6 ECC | GDDR6 ECC | GDDR6 ECC |
| Speicheranbindung | 192 Bit | 384 Bit | 128 Bit | 3072 Bit | 384 Bit | 384 Bit | 384 Bit |
| Speicherbandbreite | 300 GB/s | 600 GB/s | 4x 232 GB/s | 933 GB/s | 696 GB/s | 864 GB/s | 864 GB/s |
| GPU-Taktung | 795 MHz | 885 MHz | 885 MHz | 930 MHz | 1305 MHz | 735 MHz | 1110 MHz |
| CUDA / Streamprozessoren | 7680 | 9216 | 4x 1280 | 3804 | 10752 | 18176 | 18176 |
| Gleitkomma-Rechenleistung (FP32) | 30,31 TFLOPs | 31,2 TFLOPS | 4x 4,5 TFLOPs | 10,30 TFLOPS | 37,42 TFLOPs | 90,52 TFLOPS | 91,61 TFLOPs |
| Steckplatz | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 4.0 x16 |
| Slotbreite | 1 Slot | 1 Slot | 2 Slot | 2 Slot | 2 Slot | 2 Slot | 2 Slot |
| Anschlüsse | 3x DisplayPort 1.4, Off by Default | 4x DP 1.4a | 4x DP 1.4a | ||||
| Kühlung | passive | passive | passive | passive | passive | passive | passive |
| SLI / CrossFire | NVLink | ||||||
| Verbrauch (Watt) | 72W | 150W | 250W | 165W | 300W | 300W | 350W |
| DirectX | 12_2 Ultimate | 12_2 Ultimate | 12_2 Ultimate | 12_2 Ultimate | 12_2 Ultimate | 12_2 Ultimate | |
| Open-GL | 4.6 | 4.68 | 4.68 | 4.68 | 4.6 | 4.6 | |
| Open-CL | 3.0 | 3.0 | 3.0 | 3.0 | 3.0 | 3.0 | 3.0 |
| Vulkan | 1.3 | 1.3 | 1.3 | 1.3 | 1.3 | 1.3 | |
| Shader Model | 6.7 | 6.7 | 6.7 | 6.7 | 6.7 | 6.7 | |
| Max. Auflösung (Pixel) | 0 | 7680x4320 | 7680x4320 | 7680x4320 | |||
| Max. Anzahl Monitore | 4 | 3 | 4 | 4 | |||
| Zusatzfunktionen | 60 RT Cores, 240 Tensor Cores, NVIDIA vPC/vApps/vWS, NVIDIA vGPU, NVIDIA vCS, NVIDIA AI Enterprise, DLSS 3, 2 NVENC, 4 NVDEC, 4 JPEG decoders, AV1 | NVIDIA Virtual PC (vPC), NVIDIA RTX Virtual Workstation (vWS), NVIDIA Virtual Applications (vApps), NVIDIA Virtual Compute Server (vCS) | NVIDIA Virtual PC (vPC), NVIDIA RTX Virtual Workstation (vWS), NVIDIA Virtual Applications (vApps), NVIDIA Virtual Compute Server (vCS) | Multi-Instance GPU (MIG), NVIDIA Virtual PC (vPC), NVIDIA RTX Virtual Workstation (vWS), NVIDIA Virtual Applications (vApps), NVIDIA Virtual Compute Server (vCS) | NVIDIA Virtual PC, NVIDIA RTX Virtual Workstation (vWS), NVIDIA Virtual Compute Server (vCS, no MIG) | 209 RT Cores, 568 Tensor Cores, NVIDIA vPC/vApps/vWS, NVIDIA vGPU, NVIDIA Omniserve Enterprise, DirectCompute, Nvidia Mosaic, 3 NVENC, 3 NVDEC, AV1 | 209 RT Cores, 568 Tensor Cores, NVIDIA vPC/vApps/vWS, NVIDIA vGPU, NVIDIA Omniserve Enterprise, DirectCompute, NVIDIA RTX IO, Nvidia Mosaic, 3 NVENC, 3 NVDEC, AV1 |
