| 1 |
|
| 2 |
Platform: AMD Accelerated Parallel Processing |
| 3 |
Device: Cypress |
| 4 |
Driver version : 1348.4 (Linux x64) |
| 5 |
Compute units : 20 |
| 6 |
|
| 7 |
Global memory bandwidth (GBPS) |
| 8 |
float : 127.53 |
| 9 |
float2 : 125.89 |
| 10 |
float4 : 94.18 |
| 11 |
float8 : 64.77 |
| 12 |
float16 : 34.23 |
| 13 |
|
| 14 |
Single-precision compute (GFLOPS) |
| 15 |
float : 542.26 |
| 16 |
float2 : 1077.85 |
| 17 |
float4 : 2139.41 |
| 18 |
float8 : 2130.85 |
| 19 |
float16 : 2133.98 |
| 20 |
|
| 21 |
Double-precision compute (GFLOPS) |
| 22 |
double : 540.76 |
| 23 |
double2 : 539.13 |
| 24 |
double4 : 535.31 |
| 25 |
double8 : 537.66 |
| 26 |
double16 : 534.95 |
| 27 |
|
| 28 |
Integer compute (GIOPS) |
| 29 |
int : 270.28 |
| 30 |
int2 : 540.54 |
| 31 |
int4 : 540.12 |
| 32 |
int8 : 540.88 |
| 33 |
int16 : 541.23 |
| 34 |
|
| 35 |
Transfer bandwidth (GBPS) |
| 36 |
enqueueWriteBuffer : 3.61 |
| 37 |
enqueueReadBuffer : 3.36 |
| 38 |
enqueueMapBuffer(for read) : 206.81 |
| 39 |
memcpy from mapped ptr : 3.41 |
| 40 |
enqueueUnmap(after write) : 584.32 |
| 41 |
memcpy to mapped ptr : 3.32 |
| 42 |
|
| 43 |
Kernel launch latency : 117.88 us |
| 44 |
|