NVIDIA A10 PCIe vs NVIDIA A10G
What is the difference between NVIDIA A10 PCIe and NVIDIA A10G. Find out which graphics card has better performance.
Graphics Processor (GPU)
| GA102 | GPU Name | GA102 |
| GA102-890-A1 | Model number | GA102-890-A1 |
| Ampere | Architecture | Ampere |
| Samsung | Foundry | Samsung |
| 8 nm | Process Size | 8 nm |
| 28,300 million | Transistors | 28,300 million |
| 628 mm² | Die Size | 628 mm² |
Graphics Card
| Apr 12th, 2021 | Release Date | Apr 12th, 2021 |
| Tesla (Axx) | Family | Tesla (Axx) |
| Active | Production | Active |
| 2021 | Availability | 2021 |
| PCIe 4.0 x16 | Bus Interface | PCIe 4.0 x16 |
Memory
| 24 GB | Memory Size | 12 GB |
| GDDR6 | Memory Type | GDDR6 |
| 384 bit | Memory Bus | 384 bit |
| 600.2 GB/s | Bandwidth | 600.2 GB/s |
Performance
| 162.7 GPixel/s | Pixel fillrate | 164.2 GPixel/s |
| 488.2 GTexel/s | Texture fillrate | 492.5 GTexel/s |
| 31.24 TFLOPS (1:1) | FP16 (half) performance | 31.52 TFLOPS (1:1) |
| 31.24 TFLOPS | FP32 (float) performance | 31.52 TFLOPS |
| 976.3 GFLOPS (1:32) | FP64 (double) performance | 985.0 GFLOPS (1:32) |
Clock Speeds
| 885 MHz | Base Clock | 1320 MHz |
| 1695 MHz | Boost Clock | 1710 MHz |
| 1563 MHz 12.5 Gbps effective | Memory Clock | 1563 MHz 12.5 Gbps effective |
Render Config
| 9216 | Shading Units | 9216 |
| 288 | TMUs | 288 |
| 96 | ROPs | 96 |
| 72 | RT Cores | 72 |
| 128 KB (per SM) | L1 Cache | 128 KB (per SM) |
| 6 MB | L2 Cache | 6 MB |
| 72 | SM Count | 72 |
| 288 | Tensor Cores | 288 |
Board Design
| Single-slot | Slot Width | Single-slot |
| 267 mm 10.5 inches | Length | 267 mm 10.5 inches |
| 112 mm 4.4 inches | Width | 112 mm 4.4 inches |
| 150 W | Thermal design power (TDP) | 150 W |
| 450 W | Suggested PSU | 450 W |
| No outputs | Display Connectors | No outputs |
| 8-pin EPS | Power Connectors | 8-pin EPS |
| PG133 SKU 215 | Board Number | PG133 SKU 215 |
API support
| 12 Ultimate (12_2) | DirectX | 12 Ultimate (12_2) |
| 4.6 | OpenGL | 4.6 |
| 3.0 | OpenCL | 3.0 |
| 1.2 | Vulkan | 1.2 |
| 6.6 | Shader Model | 6.6 |
| 8.6 | CUDA | 8.6 |