| 显卡型号 | 架构/计算能力 | FP32算力(TFLOPS) | FP16算力(TFLOPS) | INT8算力(TFLOPS) | FP64算力(TFLOPS) | 带宽(GB/s) | 显存容量 |
|---|---|---|---|---|---|---|---|
| GTX 1650 | Turing/7.5 | 2.984 TFLOPS(FP32) | 5.967 TFLOPS(FP16)(2:1) | - | 93.24 GFLOPS(FP64)(1:32) | 128.1 GB/s | 4 GB |
| Orin NX 16G | Ampere/8.6 | 1.880 TFLOPS (FP32) | 3.760 TFLOPS(FP16)(2:1) | - | 940.0 GFLOPS(FP64)(1:2) | 102.4 GB/s | 16G |
| Jetson Tx2 | Pascal/6.2 | 665.6 GFLOPS(FP32) | 1,331.2 GFLOPS(FP16)(2:1) | - | 0.80 GFLOPS(FP64)(1:32) | 59.71 GB/s | 8 GB |
| GTX 1050T | Pascal/6.1 | 2.488 TFLOPS(FP32) | 38.88 GFLOPS(FP16)(1:64) | - | 77.76 GFLOPS(FP64)(1:32) | 112.1 GB/s | 4 GB |
| Goldwasser-UL | ~/6.1 | - | 8TFLOPS (FP16) | 32TOPS INT8 | - | 38.4 GB/s | 8 GB |
-
Notifications
You must be signed in to change notification settings - Fork 0
InJuly7/High-performance-operators
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
About
调优一些常见的算子