| 1 |
|
| 2 |
Platform: NVIDIA CUDA |
| 3 |
Device: GeForce RTX 2080 Ti |
| 4 |
Driver version : 415.27 (Linux x64) |
| 5 |
Compute units : 68 |
| 6 |
Clock frequency : 1650 MHz |
| 7 |
|
| 8 |
Global memory bandwidth (GBPS) |
| 9 |
float : 506.69 |
| 10 |
float2 : 532.16 |
| 11 |
float4 : 548.03 |
| 12 |
float8 : 556.57 |
| 13 |
float16 : 492.17 |
| 14 |
|
| 15 |
Single-precision compute (GFLOPS) |
| 16 |
float : 16909.53 |
| 17 |
float2 : 16894.22 |
| 18 |
float4 : 16866.23 |
| 19 |
float8 : 16798.47 |
| 20 |
float16 : 16672.67 |
| 21 |
|
| 22 |
No half precision support! Skipped |
| 23 |
|
| 24 |
Double-precision compute (GFLOPS) |
| 25 |
double : 529.92 |
| 26 |
double2 : 529.30 |
| 27 |
double4 : 527.99 |
| 28 |
double8 : 525.44 |
| 29 |
double16 : 519.98 |
| 30 |
|
| 31 |
Integer compute (GIOPS) |
| 32 |
int : 15480.90 |
| 33 |
int2 : 15398.06 |
| 34 |
int4 : 15411.76 |
| 35 |
int8 : 15226.44 |
| 36 |
int16 : 15304.72 |
| 37 |
|
| 38 |
Transfer bandwidth (GBPS) |
| 39 |
enqueueWriteBuffer : 9.61 |
| 40 |
enqueueReadBuffer : 8.49 |
| 41 |
enqueueMapBuffer(for read) : 10.79 |
| 42 |
memcpy from mapped ptr : 11.37 |
| 43 |
enqueueUnmap(after write) : 12.27 |
| 44 |
memcpy to mapped ptr : 11.84 |
| 45 |
|
| 46 |
Kernel launch latency : 3.81 us |
| 47 |
|