NVIDIA GRID K2 vs NVIDIA Tesla P40
Graphics Processor
| GPU model | GK104 | GP102 |
|---|---|---|
| GPU variant | GK104-895-A2 | — |
| Architecture | Kepler | Pascal |
| Foundry | TSMC | TSMC |
| Manufacturing process | 28 nm | 16 nm |
| Number of transistors | 3,540 million | 11,800 million |
| Die size | 294 mm² | 471 mm² |
Graphics Card
| Release date | May 11th, 2013 | Sep 13th, 2016 |
|---|---|---|
| Generation | GRID | Tesla |
| Production | End-of-life | End-of-life |
| Launch price | 5,199 USD | 5,699 USD |
| Interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Clocks
| GPU clock | 745 MHz | — |
|---|---|---|
| Memory clock | 1250 MHz 5 Gbps effective | 1808 MHz 14.5 Gbps effective |
| Base clock | — | 1303 MHz |
| Boost clock | — | 1531 MHz |
Memory Configuration
| Memory size | 4 GB | 24 GB |
|---|---|---|
| Memory type | GDDR5 | GDDR5X |
| Memory bus width | 256 bit | 384 bit |
| Bandwidth | 160.0 GB/s | 694.3 GB/s |
Render Configuration
| Shading units | 1536 | 3840 |
|---|---|---|
| TMUs | 128 | 240 |
| ROPs | 32 | 96 |
| SMX count | 8 | — |
| Cache L1 | 16 KB (per SMX) | 48 KB (per SM) |
| Cache L2 | 512 KB | 3 MB |
| SM count | — | 30 |
Performance
| Pixel rate | 23.84 GPixel/s | 147.0 GPixel/s |
|---|---|---|
| Texture rate | 95.36 GTexel/s | 367.4 GTexel/s |
| FP32 (float) performance | 2.289 TFLOPS | 11.76 TFLOPS |
| FP64 (double) performance | 95.36 GFLOPS (1:24) | 367.4 GFLOPS (1:32) |
| FP16 (half) performance | — | 183.7 GFLOPS (1:64) |
Dimensions & Outputs
| Slot width | Dual-slot | Dual-slot |
|---|---|---|
| Length | 267 mm 10.5 inches | 267 mm 10.5 inches |
| TDP | 225 W | 250 W |
| Suggested PSU | 550 W | 600 W |
| Outputs | No outputs | No outputs |
| Power connectors | 1x 8-pin | 1x 6-pin + 1x 8-pin |
| Board number | P2055 | PG610 SKU 200 |
| Width | — | 111 mm 4.4 inches |
API Support & Features
| DirectX | 12 (11_0) | 12 (12_1) |
|---|---|---|
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.1 | 1.2 |
| CUDA | 3.0 | 6.1 |
| Shader model | 5.1 | 6.4 |
