| Product | AMD Instinct™ MI100 Accelerator - 32GB HBM2 - PCIe 4.0 x16 - Passive Cooling | AMD Instinct™ MI210 Accelerator - 64GB HBM2e - PCIe 4.0 x16 - Passive Cooling | NVIDIA® A10 GPU Computing Accelerator - 24GB GDDR6 - PCIe 4.0 x16 - Passive Cooler (w/o CEC) | NVIDIA® A16 GPU Computing Accelerator - 64GB (4x 16GB) GDDR6 - PCIe 4.0 x16 - Passive Cooler | NVIDIA® A30 GPU Computing Accelerator - 24GB HBM2 - PCIe 4.0 x16 - Passive Cooler | NVIDIA® A40 GPU Computing Accelerator - 48GB GDDR6 - PCIe 4.0 x16 - Passive Cooling | NVIDIA® L4 ADA GPU Computing Accelerator - 24GB GDDR6X - PCIe 4.0 x16 - Passive Cooling | NVIDIA® L40S ADA GPU Computing Accelerator - 48GB GDDR6 - PCIe 4.0 x16 - Passive Cooling | NVIDIA® RTX A6000 - 48GB GDDR6 - PCIe 4.0 x16 - Active Cooling (4xDP) | NVIDIA® RTX 6000 Ada Generation - 48GB GDDR6 ECC - PCIe 4.0 x16 - Active Cooling (4xDP) |
| Action | Select | Select | Select | Select | Select | Select | Select | Select | Select | Select |
| Main Specifications | ||||||||||
| Product Series | AMD Instinct | AMD Instinct | Nvidia A10 | Nvidia A16 | Nvidia A30 | Nvidia A40 | Nvidia L4 | Nvidia L40S | ||
| Core Type | NVIDIA TENSOR | NVIDIA TENSOR | NVIDIA TENSOR | NVIDIA TENSOR | NVIDIA TENSOR | NVIDIA TENSOR | ||||
| Core Clock Speed | 1502 MHz | 1700 MHz | 885 MHz (1695 MHz Boost Clock) | 795 MHz Base | 2040 MHz Boost | ||||||
| Host Interface | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 64GB/s | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 | PCI Express 4.0 x16 |
| GPU Architecture | CDNA | CDNA2 | Ampere | Ampere | Ampere | Ampere | Ada Lovelace | Ada Lovelace | ||
| Product Type | Workstation | Workstation | ||||||||
| Product Line | NVIDIA Professional Graphics | NVIDIA Professional Graphics | ||||||||
| Memory Technology | GDDR6 | GDDR6 | ||||||||
| Memory Capacity | 48 GB | 48 GB with ECC | ||||||||
| Max Displays | 4 Displays | |||||||||
| Detailed Specifications | ||||||||||
| Streaming Processor Cores | 7,680 | 6656 | 10752 CUDA Cores | 18,176 | 10752 Shading Units | 18,176 | ||||
| Compute Units | 120 | 104 | ||||||||
| NVIDIA Tensor Cores | 336 Tensor Cores | 568 | Gen 4 | 336 | 568 | ||||||
| NVIDIA RT Cores | 72 RT Cores | 84 RT Cores | 142 | Gen 3 | 84 | 142 | |||||
| Memory Clock Speed | 1.2 GHz | 1.6 GHz | 1563 MHz | 6251 MHz | 2000 MHz 16 Gbps effective | |||||
| Memory Interface | 4096-bit | 4096-bit | 384-bit | 192-bit | 384-bit | 384-bit | ||||
| Memory Speeds (GT/s) | 14.5Gbps GDDR6 | |||||||||
| Max Memory Size | 32 GB HBM2 | 64 GB HBM2e | 24 GB GDDR6 | 4x 16GB GDDR6 with error-correcting code (ECC) | 24 GB HBM2 | 48 GB GDDR6 with error-correcting code (ECC) | 24 GB | 48GB GDDR6 with ECC | ||
| Max Memory Bandwidth | Up to 1228.8 GB/s | Up to 1638.4 GB/s | 600 GB/s | 4x 232GB/s | 933 GB/s | 696 GB/s | 300 GB/s | 864 GB/s | ||
| Infinity Fabric™ Links | 3 | 3 | ||||||||
| Peak Infinity Fabric™ Link Bandwidth | 92 GB/s | 100 GB/s | ||||||||
| Peak FP64 | 5.2 teraFLOPS | |||||||||
| Peak FP64 Tensor Core | 10.3 teraFLOPS | |||||||||
| Peak FP32 | 22.6 TFLOPs | 31.2 teraFLOPS | 10.3 teraFLOPS | 30.3 TFLOPS | 91.6 teraFLOPS | |||||
| Peak TF32 Tensor Core | 62.5 teraFLOPS | 125 teraFLOPS | 82 teraFLOPS | 165 teraFLOPS | 120 TFLOPS | Sparsity | 183 teraFLOPS | ||||||
| Peak BFLOAT16 Tensor Core | 125 teraFLOPS | 250 teraFLOPS | 165 teraFLOPS | 330 teraFLOPS | 242 TFLOPS | Sparsity | 362.05 teraFLOPS | ||||||
| Peak FP16 Tensor Core | 125 teraFLOPS | 250 teraFLOPS | 165 teraFLOPS | 330 teraFLOPS | 242 TFLOPS | Sparsity | 362.05 teraFLOPS | ||||||
| Peak FP8 Tensor Core | 485 TFLOPS | Sparsity | 733 teraFLOPS | ||||||||
| Peak INT8 Tensor Core | 250 TOPS | 500 TOPS | 330 TOPS | 661 TOPS | 485 TOPS | Sparsity | 733 teraFLOPS | ||||||
| Peak INT4 Tensor Core | 500 TOPS | 1,000 TOPS | 661 TOPS | 1321 TOPS | 733 teraFLOPS | |||||||
| NVIDIA NVLink™ Interconnect Bandwidth | Third-gen NVLINK: 200GB/s | NVIDIA NVLink 112.5 GB/s (bidirectional) PCIe Gen4 16 GB/s | Not supported | |||||||
| Multi-Instance GPUs | No | |||||||||
| Tensor Performance | 1457.0 TFLOPS | |||||||||
| NVIDIA CUDA™ Technology | Yes | |||||||||
| NVENC | NVDEC | 2 | 4 | 4 | JPEG Decoders | AV1 Encode and Decode | 3x l 3x (includes AV1 encode and decode) | ||||||||
| Secure Boot with Root of Trust | Yes | Yes | ||||||||
| NEBS Ready | Yes | Level 3 | Level 3 | ||||||||
| Peak Half Precision (FP16) Performance | 184.6 TFLOPs | |||||||||
| Peak Single Precision Matrix (FP32) Performance | 46.1 TFLOPs | 45.3 TFLOPs | ||||||||
| Peak Double Precision Matrix (FP64) Performance | 45.3 TFLOPs | |||||||||
| Peak Single Precision (FP32) Performance | 23.1 TFLOPs | |||||||||
| Peak Double Precision (FP64) Performance | 11.5 TFLOPs | 22.6 TFLOPs | ||||||||
| Peak INT4 Performance | 184.6 TOPs | 181 TOPs | ||||||||
| Peak INT8 Performance | 184.6 TOPs | |||||||||
| Peak bfloat16 | 92.3 TFLOPs | 181 TFLOPs | ||||||||
| ECC Protection | Yes (Full-Chip) | On by Default | ||||||||
| Transistor Count | 28.3 Billion | 76.3 billion | ||||||||
| DisplayPort Connectors | 3x DisplayPort 1.4 A40 is configured for virtualization by default with physical display connectors disabled. The display outputs can be enabled via management software tools. | None | vGPU Only | 4x DisplayPort 1.4a | |||||||
| OS Support | Linux x86_64 | Linux x86_64 | ||||||||
| Cooling | Passive | Passive | Passive | Passive | Passive | Passive | Passive | |||
| Dual Slot | yes | yes | Single-slot | Dual-slot | Dual-slot | 2-slot Low-profile | No | |||
| Dimensions | 10.5" (267 mm) Board Length | 10.5" (267 mm) Board Length | FHFL | 4.4" (H) x 10.5" (L) | 4.4" (H) x 10.5" (L) | 4.4" (H) x 10.5" (L) | 4.4" H x 10.5" L | |||
| Form Factor | Full Height | 6.61” L x 2.71” H (Low-profile) | ||||||||
| Lithography | TSMC 7nm FinFET | 8 nm | Samsung 8nm | Samsung 8nm | 4 nm NVIDIA Custom Process | |||||
| Supplementary Power Connectors | 2x PCIe 8-pin connectors | 1x8 pin 12V EPS | None | 8-pin CPU | 1x 8-pin CPU (EPS12V) | 1x 8-pin CPU (EPS12V) | 1x 16-pin | 1x 8-pin EPS | 1x PCIe CEM5 16-pin | |
| Max Graphics Card Power (W) | 300W | 300W Peak | 150W | 250W | 165W | 300W | 72W | 350W | 300W | 300W |
| Processor | Ampere (GA102) | NVIDIA Ada Lovelace | ||||||||
| Memory Bandwidth | 768 GB/s | 960 GB/s | ||||||||
| Core Clock Speed | 1455 MHz Base Clock 1860 MHz Boost Clock | |||||||||
| L2 Cache Size | 6 MB | |||||||||
| API Support | CUDA 8.5, OpenCL 2.0 Shader Model 6.5, OpenGL 4.6, DirectX 12 Ultimate (12_2), Vulkan 1.2 | |||||||||
| Texture Fill Rate | 625 GTexel/s | |||||||||
| Graphics Resolution | 7680 x 4320 x36 bpp at 60 Hz | |||||||||
| Peak Double Precision FP64 Performance | 1,250 GFLOPS (1:32) | |||||||||
| Peak Single Precision FP32 Performance | 38.7 TFLOPS | 91.1 TFLOPS | ||||||||
| Peak Half Precision FP16 Performance | 40.00 TFLOPS (1:1) | |||||||||
| Multi-GPU Scalability | NVLINK 2-way low profile (2-slot and 3-slot bridges) connects 2x NVIDIA RTX A6000 | |||||||||
| NVLink Interconnect | 112.5 GB/s (bidirectional) | |||||||||
| RT Core Performance | 210.6 TFLOPS | |||||||||
| VR Ready | Yes | |||||||||
| Vulkan API | 1.2 | |||||||||
| DisplayPort Output | 4x DisplayPort 1.4a | 4x DP 1.4a | ||||||||
| Minimum Recommended Power, Single Card (W) | 700W | 600 | ||||||||
| Minimum Recommended Power, 2-Way (W) | 850 | 750 | ||||||||
| Minimum Recommended Power, 3-Way (W) | 1000 | 850 | ||||||||
| Minimum Recommended Power, 4-Way (W) | 1200 | 1000 | ||||||||
| Thermal Solution | Active Heatsink | Blower Active Fan | ||||||||
| Slot Height | 2-Slot | 2-Slot | ||||||||
| Action | Select | Select | Select | Select | Select | Select | Select | Select | Select | Select |