NVIDIA CMP 170HX vs NVIDIA CMP 90HX
Graphics Processor
| GPU model | GA100 | GA102 |
|---|---|---|
| GPU variant | GA100-105F-A1 | GA102-100-A1 |
| Architecture | Ampere | Ampere |
| Foundry | TSMC | Samsung |
| Manufacturing process | 7 nm | 8 nm |
| Number of transistors | 54,200 million | 28,300 million |
| Die size | 826 mm² | 628 mm² |
Graphics Card
| Release date | Sep 1st, 2021 | Jul 28th, 2021 |
|---|---|---|
| Generation | Mining GPUs | Mining GPUs |
| Production | Active | Active |
| Interface | PCIe 4.0 x4 | PCIe 4.0 x16 |
| Launch price | 4,299 USD | — |
Clocks
| Base clock | 1140 MHz | 1500 MHz |
|---|---|---|
| Boost clock | 1410 MHz | 1710 MHz |
| Memory clock | 1458 MHz 2.9 Gbps effective | 1188 MHz 19 Gbps effective |
Memory Configuration
| Memory size | 16 GB | 10 GB |
|---|---|---|
| Memory type | HBM2e | GDDR6X |
| Memory bus width | 4096 bit | 320 bit |
| Bandwidth | 1,493 GB/s | 760.3 GB/s |
Render Configuration
| Shading units | 4480 | 6400 |
|---|---|---|
| TMUs | 280 | 200 |
| ROPs | 128 | 80 |
| SM count | 70 | 50 |
| Tensor cores | 280 | 200 |
| Cache L1 | 192 KB (per SM) | 128 KB (per SM) |
| Cache L2 | 8 MB | 5 MB |
| RT cores | — | 50 |
Performance
| Pixel rate | 180.5 GPixel/s | 136.8 GPixel/s |
|---|---|---|
| Texture rate | 394.8 GTexel/s | 342.0 GTexel/s |
| FP16 (half) performance | 50.53 TFLOPS (4:1) | 21.89 TFLOPS (1:1) |
| FP32 (float) performance | 12.63 TFLOPS | 21.89 TFLOPS |
| FP64 (double) performance | 6.317 TFLOPS (1:2) | 342.0 GFLOPS (1:64) |
Dimensions & Outputs
| Slot width | IGP | Dual-slot |
|---|---|---|
| TDP | 250 W | 320 W |
| Suggested PSU | 600 W | 700 W |
| Outputs | No outputs | No outputs |
| Power connectors | 2x 8-pin | 2x 8-pin |
| Length | — | 285 mm 11.2 inches |
| Width | — | 112 mm 4.4 inches |
| Board number | — | PG132 SKU 100 |
API Support & Features
| DirectX | — | 12 Ultimate (12_2) |
|---|---|---|
| OpenGL | — | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | — | 1.2 |
| CUDA | 8.0 | 8.6 |
| Shader model | — | 6.6 |
