NVIDIA Jetson TX1 GPU vs NVIDIA Riva 128 PCI
Graphics Processor
| GPU model | GM20B | NV3 |
|---|---|---|
| GPU variant | TM670D-A1 | — |
| Architecture | Maxwell 2.0 | — |
| Foundry | TSMC | ST |
| Manufacturing process | 20 nm | 350 nm |
| Number of transistors | 2,000 million | 4 million |
| Die size | 118 mm² | 90 mm² |
Integrated Graphics
| Release date | Jan 4th, 2015 | — |
|---|---|---|
| Generation | Tegra | — |
| Production | End-of-life | — |
| Interface | IGP | — |
Clocks
| GPU clock | 1000 MHz | 100 MHz |
|---|---|---|
| Memory clock | System Shared | 100 MHz |
Memory Configuration
| Memory size | System Shared | 4 MB |
|---|---|---|
| Memory type | System Shared | SDR |
| Memory bus width | System Shared | 128 bit |
| Bandwidth | System Dependent | 1.600 GB/s |
Render Configuration
| Shading units | 256 | — |
|---|---|---|
| TMUs | 16 | 1 |
| ROPs | 16 | 1 |
| SMM count | 2 | — |
| Cache L1 | 48 KB (per SMM) | — |
| Cache L2 | 256 KB | — |
| Pixel shaders | — | 1 |
| Vertex shaders | — | 0 |
Performance
| Pixel rate | 16.00 GPixel/s | 100.0 MPixel/s |
|---|---|---|
| Texture rate | 16.00 GTexel/s | 100.0 MTexel/s |
| FP16 (half) performance | 1,024 GFLOPS (2:1) | — |
| FP32 (float) performance | 512.0 GFLOPS | — |
| FP64 (double) performance | 16.00 GFLOPS (1:32) | — |
Dimensions & Outputs
| Slot width | IGP | Single-slot |
|---|---|---|
| TDP | 15 W | 4 W |
| Outputs | No outputs | 1x VGA1x DB13W3 |
| Suggested PSU | — | 200 W |
| Power connectors | — | None |
API Support & Features
| DirectX | 12 (12_1) | 5.0 |
|---|---|---|
| OpenGL | 4.6 | 1.0 |
| OpenCL | 1.2 | — |
| Vulkan | 1.1 | — |
| CUDA | 5.3 | — |
| Shader model | 6.4 | — |
| Pixel shader | — | — |
| Vertex shader | — | — |
Other Features
Graphics Card
| Release date | — | Apr 1st, 1997 |
|---|---|---|
| Generation | — | Riva |
| Production | — | End-of-life |
| Interface | — | PCI |
| Predecessor | — | NV1 |
| Successor | — | TNT2 |
