NVIDIA A40 PCIe vs NVIDIA Tesla P100 SXM2
Graphics Processor
| GPU model | GA102 | GP100 |
|---|---|---|
| Architecture | Ampere | Pascal |
| Foundry | Samsung | TSMC |
| Manufacturing process | 8 nm | 16 nm |
| Number of transistors | 28,300 million | 15,300 million |
| Die size | 628 mm² | 610 mm² |
| GPU variant | — | GP100-890-A1 |
Graphics Card
| Release date | Oct 5th, 2020 | Apr 5th, 2016 |
|---|---|---|
| gpu.details.availability | 2021 | — |
| Generation | Tesla | Tesla |
| Production | Active | End-of-life |
| Interface | PCIe 4.0 x16 | PCIe 3.0 x16 |
Clocks
| Base clock | 1305 MHz | 1328 MHz |
|---|---|---|
| Boost clock | 1740 MHz | 1480 MHz |
| Memory clock | 1812 MHz 14.5 Gbps effective | 715 MHz 1430 Mbps effective |
Memory Configuration
| Memory size | 48 GB | 16 GB |
|---|---|---|
| Memory type | GDDR6 | HBM2 |
| Memory bus width | 384 bit | 4096 bit |
| Bandwidth | 695.8 GB/s | 732.2 GB/s |
Render Configuration
| Shading units | 10752 | 3584 |
|---|---|---|
| TMUs | 336 | 224 |
| ROPs | 112 | 96 |
| SM count | 84 | 56 |
| Tensor cores | 336 | — |
| RT cores | 84 | — |
| Cache L1 | 128 KB (per SM) | 24 KB (per SM) |
| Cache L2 | 6 MB | 4 MB |
Performance
| Pixel rate | 194.9 GPixel/s | 142.1 GPixel/s |
|---|---|---|
| Texture rate | 584.6 GTexel/s | 331.5 GTexel/s |
| FP16 (half) performance | 37.42 TFLOPS (1:1) | 21.22 TFLOPS (2:1) |
| FP32 (float) performance | 37.42 TFLOPS | 10.61 TFLOPS |
| FP64 (double) performance | 1,169 GFLOPS (1:32) | 5.304 TFLOPS (1:2) |
Dimensions & Outputs
| Slot width | Dual-slot | — |
|---|---|---|
| Length | 267 mm 10.5 inches | — |
| Width | 112 mm 4.4 inches | — |
| TDP | 300 W | 300 W |
| Suggested PSU | 700 W | 700 W |
| Outputs | 3x DisplayPort | No outputs |
| Power connectors | 8-pin EPS | None |
API Support & Features
| DirectX | 12 Ultimate (12_2) | 12 (12_1) |
|---|---|---|
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.2 | 1.2 |
| CUDA | 8.6 | 6.0 |
| Shader model | 6.6 | 6.4 |
