| 1 |
|
| 2 |
Platform: AMD Accelerated Parallel Processing |
| 3 |
Device: Hawaii (R9 390) |
| 4 |
Driver version : 1800.5 (VM) (Linux x64) |
| 5 |
Compute units : 40 |
| 6 |
Clock frequency : 1000 MHz |
| 7 |
|
| 8 |
Global memory bandwidth (GBPS) |
| 9 |
float : 325.28 |
| 10 |
float2 : 335.50 |
| 11 |
float4 : 326.06 |
| 12 |
float8 : 185.19 |
| 13 |
float16 : 92.65 |
| 14 |
|
| 15 |
Single-precision compute (GFLOPS) |
| 16 |
float : 5079.23 |
| 17 |
float2 : 5074.38 |
| 18 |
float4 : 5060.63 |
| 19 |
float8 : 5033.64 |
| 20 |
float16 : 4978.18 |
| 21 |
|
| 22 |
Double-precision compute (GFLOPS) |
| 23 |
double : 638.55 |
| 24 |
double2 : 637.93 |
| 25 |
double4 : 636.60 |
| 26 |
double8 : 633.93 |
| 27 |
double16 : 628.44 |
| 28 |
|
| 29 |
Integer compute (GIOPS) |
| 30 |
int : 1701.12 |
| 31 |
int2 : 1700.71 |
| 32 |
int4 : 1699.93 |
| 33 |
int8 : 1699.02 |
| 34 |
int16 : 1697.04 |
| 35 |
|
| 36 |
Transfer bandwidth (GBPS) |
| 37 |
enqueueWriteBuffer : 8.66 |
| 38 |
enqueueReadBuffer : 5.75 |
| 39 |
enqueueMapBuffer(for read) : 36584.05 |
| 40 |
memcpy from mapped ptr : 3.93 |
| 41 |
enqueueUnmap(after write) : 429496.75 |
| 42 |
memcpy to mapped ptr : 3.71 |
| 43 |
|
| 44 |
Kernel launch latency : 34.63 us |
| 45 |
|