| GPU Name | AD103 |
| GPU Codename | AD103 |
| Architecture | Ada Lovelace |
| GPCs | 7 |
| TPCs | 38 |
| SMs | 76 |
| CUDA Cores / SM | 128 |
| CUDA Cores / GPU | 9728 |
| Tensor Cores / SM | 4 (4th Gen) |
| Tensor Cores / GPU | 304 (4th Gen) |
| OFA TOPS | 305 |
| RT Cores | 76 (3rd Gen) |
| GPU Boost Clock (MHz) | 2505 |
| Peak FP32 TFLOPS (non-Tensor) | 48.7 |
| Peak FP16 TFLOPS (non-Tensor) | 48.7 |
| Peak BF16 TFLOPS (non-Tensor) | 48.7 |
| Peak INT32 TOPS (non-Tensor) | 24.4 |
| RT TFLOPS | 112.7 |
| Peak FP8 Tensor TFLOPS with FP16 Accumulate | 389.9/779.82 |
| Peak FP8 Tensor TFLOPS with FP32 Accumulate | 389.9/779.82 |
| Peak FP16 Tensor TFLOPS with FP16 Accumulate | 194.9/389.82 |
| Peak FP16 Tensor TFLOPS with FP32 Accumulate | 97.5/1952 |
| Peak BF16 Tensor TFLOPS with FP32 Accumulate | 97.5/1952 |
| Peak TF32 Tensor TFLOPS | 48.7/97.42 |
| Peak INT8 Tensor TOPS | 389.9/779.822 |
| Peak INT4 Tensor TOPS | 779.8/1559.62 |
| Frame Buffer Memory Size and Type | 16 GB GDDR6X |
| Memory Interface | 256-bit |
| Memory Clock (Data Rate) | 22.4 Gbps |
| Memory Bandwidth | 716.8 GB/sec |
| ROPs | 112 |
| Pixel Fill-rate (Gigapixels/sec) | 280.6 |
| Texture Units | 304 |
| Texel Fill-rate (Gigatexels/sec) | 761.5 |
| L1 Data Cache/Shared Memory | 9728 KB |
| L2 Cache Size | 65536 KB |
| Register File Size | 19456 KB |
| Video Engines | 2 x NVENC (8th Gen), 1 x NVDEC (5th Gen) |
| TGP (Total Graphics Power) | 320 W |
| Transistor Count | 45.9 Billion |
| Die Size | 378.6 mm2 |
| Manufacturing Process | TSMC 4N NVIDIA Custom Process |
| PCI Express Interface | Gen 4 |
