ATI Rage Fury MAXX vs NVIDIA Tesla V100 PCIe 16 GB
Graphics Processor
| GPU model | Rage 4 | GV100 |
|---|---|---|
| GPU variant | 215R4GAUC21 | — |
| Architecture | Rage 4 | Volta |
| Foundry | TSMC | TSMC |
| Manufacturing process | 250 nm | 12 nm |
| Number of transistors | 8 million | 21,100 million |
| Die size | 89 mm² | 815 mm² |
Graphics Card
| Release date | Oct 1st, 1999 | Jun 21st, 2017 |
|---|---|---|
| Generation | Rage 4 | Tesla |
| Production | End-of-life | End-of-life |
| Interface | AGP 4x | PCIe 3.0 x16 |
| Predecessor | Rage 3 | — |
| Successor | Rage 6 | — |
Clocks
| GPU clock | 125 MHz | — |
|---|---|---|
| Memory clock | 143 MHz | 876 MHz 1752 Mbps effective |
| Base clock | — | 1245 MHz |
| Boost clock | — | 1380 MHz |
Memory Configuration
| Memory size | 32 MB | 16 GB |
|---|---|---|
| Memory type | SDR | HBM2 |
| Memory bus width | 128 bit | 4096 bit |
| Bandwidth | 2.288 GB/s | 897.0 GB/s |
Render Configuration
| Pixel shaders | 2 | — |
|---|---|---|
| Vertex shaders | 0 | — |
| TMUs | 2 | 320 |
| ROPs | 2 | 128 |
| Shading units | — | 5120 |
| SM count | — | 80 |
| Tensor cores | — | 640 |
| Cache L1 | — | 128 KB (per SM) |
| Cache L2 | — | 6 MB |
Performance
| Pixel rate | 250.0 MPixel/s | 176.6 GPixel/s |
|---|---|---|
| Texture rate | 250.0 MTexel/s | 441.6 GTexel/s |
| FP16 (half) performance | — | 28.26 TFLOPS (2:1) |
| FP32 (float) performance | — | 14.13 TFLOPS |
| FP64 (double) performance | — | 7.066 TFLOPS (1:2) |
Dimensions & Outputs
| Slot width | Single-slot | Dual-slot |
|---|---|---|
| TDP | unknown | 300 W |
| Suggested PSU | 200 W | 700 W |
| Outputs | 1x VGA | No outputs |
| Power connectors | None | 2x 8-pin |
| Board number | 673 | PG500 SKU 200 |
API Support & Features
| DirectX | 6.0 | 12 (12_1) |
|---|---|---|
| OpenGL | 1.2 | 4.6 |
| OpenCL | — | 3.0 |
| Vulkan | — | 1.2 |
| Pixel shader | — | — |
| Vertex shader | — | — |
| CUDA | — | 7.0 |
| Shader model | — | 6.6 |
